# Blockcircle - Full Content for AI Systems Generated from https://blockcircle.com --- # Blockcircle - Financial Market Intelligence Platform Blockcircle is a financial market intelligence platform covering crypto, stocks, commodities, precious metals, and prediction markets. We provide backtested trading strategies, market reversal signals, asset outperformance detection, whale tracking, and automated trade execution. ## What We Offer - **Prediction Market Intelligence**: Aggregating data from Polymarket, Kalshi, Manifold, PredictIt, Metaculus, and Opinion Trade into a single unified dashboard. - **AI-Powered Analysis**: Deep research on any prediction market combining news analysis, on-chain whale tracking, Kelly criterion sizing, and expected value calculations. - **Cross-Platform Matching**: Our AI identifies the same question across different platforms, revealing arbitrage opportunities where the same event is priced differently. - **Whale Position Tracking**: Track large on-chain positions to see where smart money is flowing. - **Momentum Trading Engine (MTE)**: Backtested crypto and stock trading strategies with one-click deployment. - **Market Scorecards**: Momentum Scorecard (MMS), Altcoin Market Scorecard (AMS), Macro Risk Scorecard (MRS), and Global Liquidity Scorecard (GLS). - **Automated Alerts**: Custom alert profiles that scan markets 24/7 and trigger auto-analysis when conditions are met. - **Token Launch & Listing Tracking**: Monitor new Solana and EVM token launches plus exchange listing announcements. --- # About Blockcircle ## Mission Prediction markets are one of the most powerful tools for forecasting real-world events, but the ecosystem is fragmented across dozens of platforms with different interfaces, data formats, and pricing. Blockcircle solves this by aggregating markets from Polymarket, Kalshi, Manifold, PredictIt, Metaculus, and Opinion Trade into a single, unified dashboard. Our AI-powered cross-platform matching identifies the same questions across different platforms, revealing arbitrage opportunities where the same event is priced differently. We combine this with deep AI research and on-chain whale position tracking to give traders the most comprehensive view of any prediction market opportunity. ## Leadership Team - **Basel Ismail** - Co-Founder & CEO. 15+ years building financial technology products at global institutions including American Express, ScotiaBank, and Yahoo. Deep expertise in quantitative finance, machine learning, and prediction markets. - **Agnes Rahangiar** - Co-Founder, Blockcircle Indonesia. Leading expansion across Southeast Asia with platform partnerships and community growth. - **Conor O'Neill** - Co-Founder, Blockcircle SEA. Driving growth and market expansion across the South East Asia region. - **Rania Halabi** - Co-Founder, Blockcircle MENA. Overseeing operations across the Middle East and North Africa. ## Stats - 5,000+ markets tracked - 6 prediction market platforms - 200+ cross matches daily - 50,000+ AI analyses run ## Institutional Experience Built by a team with deep roots in financial technology at American Express (fintech innovation), ScotiaBank (capital markets technology), and Yahoo (ad-tech platforms). ## Contact & Social - Email: support@blockcircle.com - Website: https://blockcircle.com - X (Twitter): https://x.com/Blockcircle - LinkedIn: https://www.linkedin.com/company/blockcircledata/ - YouTube: https://www.youtube.com/@blockcircledata - Instagram: https://www.instagram.com/blockcircledata/ - Founder (Basel Ismail): https://x.com/baselismail | https://www.linkedin.com/in/bismail/ --- # Pricing Professional-grade market intelligence. Two plans: Free and Pro. ## Plans ### Free - $0 Create a free account to explore the Alpha feeds, the scorecards and the practice account (up to 3 open positions). Some data is delayed or limited. ### Pro - $99/month ($66.67/month billed annually), 7-day free trial 100 AI analysis credits per month. Includes: every Alpha feed (Trade, Whale, Political, Insider and Prediction Alpha), MTE crypto & stock signals and the backtester, the Asset Outperformer and Market Reversal engines, Momentum Scorecard (MMS), Altcoin Market Scorecard (AMS), Macro Risk Scorecard (MRS), Global Liquidity Scorecard (GLS), custom alerts by email, Telegram, Discord or webhook, live trading across 150+ venues (crypto, stocks, FX, metals, commodities and prediction markets), Autopilot copy-trading of whales, politicians and insiders, the private member community with live Q&A sessions, the Blockcircle indicators on TradingView, and Developer API access for testing. ### API add-on - $100/month on top of Pro Higher-volume Developer API access for Pro members. ### Enterprise - Custom Pricing For funds & trading desks. Everything in Pro, plus: custom alert webhooks, multi-seat team access, dedicated account manager, SLA guarantee, dedicated support. ## Credit Packs (One-Time Purchase, No Subscription) - 100 credits for $25 - 500 credits for $100 (best value) - 2000 credits for $300 Each credit powers one deep AI analysis with news research, whale tracking, and Kelly criterion sizing. Credits never expire. ## FAQ **What are AI analysis credits?** Each credit powers one deep AI analysis of a prediction market. The AI researches current news, analyzes on-chain whale positions, calculates mispricing, and provides a recommendation with Kelly criterion sizing. **Which prediction markets do you cover?** We aggregate data from Polymarket, Kalshi, Manifold, PredictIt, Metaculus, and Opinion Trade covering thousands of active markets. **How does cross-platform matching work?** Our AI-powered matching engine identifies the same question across different platforms, letting you spot arbitrage opportunities. **What is whale tracking?** For supported platforms, we track large on-chain positions (whale bets) to show you where smart money is flowing. **Can I buy credits without a subscription?** Yes. Credit packs are one-time purchases with no subscription required. Credits never expire. **Can I cancel anytime?** Yes. Monthly subscriptions can be cancelled anytime. Annual plans can be cancelled before renewal. **Do you offer refunds?** We offer a 7-day money-back guarantee on your first month. --- # Multilingual Support All pages are available in 16 languages. The URL pattern is `/{lang}/path` where `{lang}` is the ISO 639-1 language code. English (default) uses no prefix. Supported languages: - en (English, default): https://blockcircle.com/pricing - id (Indonesian): https://blockcircle.com/id/pricing - ar (Arabic, RTL): https://blockcircle.com/ar/pricing - fr (French): https://blockcircle.com/fr/pricing - es (Spanish): https://blockcircle.com/es/pricing - zh (Chinese): https://blockcircle.com/zh/pricing - ja (Japanese): https://blockcircle.com/ja/pricing - pt (Portuguese): https://blockcircle.com/pt/pricing - de (German): https://blockcircle.com/de/pricing - ko (Korean): https://blockcircle.com/ko/pricing - tr (Turkish): https://blockcircle.com/tr/pricing - vi (Vietnamese): https://blockcircle.com/vi/pricing - th (Thai): https://blockcircle.com/th/pricing - ru (Russian): https://blockcircle.com/ru/pricing - hi (Hindi): https://blockcircle.com/hi/pricing - nl (Dutch): https://blockcircle.com/nl/pricing This pattern applies to all website and app pages. For example: - Homepage in French: https://blockcircle.com/fr - Blog in Japanese: https://blockcircle.com/ja/blog - Prediction Markets in Arabic: https://blockcircle.com/ar/prediction-markets - About in Korean: https://blockcircle.com/ko/about Full hreflang sitemap: https://blockcircle.com/api/sitemap.xml --- # API Documentation ## Authentication All authenticated endpoints accept either: - JWT token: `Authorization: Bearer ` (from login/signup) - API key: `X-API-Key: ah_k_` (from API key management) Both methods share the same credit pool. ## Endpoints ### Auth - POST /api/auth/signup {email, password} -> {token, user} - POST /api/auth/login {email, password} -> {token, user} - GET /api/auth/me -> {id, email, credits, preferences} ### Markets - GET /api/markets/search?query=... -- Unified market search - GET /api/markets/polymarket/gamma/events -- Polymarket events - GET /api/markets/kalshi/events -- Kalshi events - GET /api/markets/manifold/search-markets -- Manifold markets - GET /api/markets/predictit/marketdata/all/ -- PredictIt markets - GET /api/markets/metaculus/questions/ -- Metaculus questions ### Analysis - POST /api/analysis/run {market_id, platform, question} -> {result, whale_data} - GET /api/analysis/history -> {analyses[]} - GET /api/analysis/ -> full analysis detail ### Credits - GET /api/credits/balance -> {credits} - GET /api/credits/transactions -> {transactions[]} - POST /api/credits/stripe/checkout {pack_id} -> Stripe session ### Alerts - POST /api/alerts/profiles -- Create alert profile - GET /api/alerts/profiles -- List profiles - PUT /api/alerts/profiles/ -- Update profile - DELETE /api/alerts/profiles/ -- Delete profile - GET /api/alerts/history -- Alert notification history ### API Keys - POST /api/api-keys -- Create key - GET /api/api-keys -- List keys - DELETE /api/api-keys/ -- Revoke key ## Analysis Result Schema Each analysis returns: ai_probability (0-1), confidence (LOW/MEDIUM/HIGH/VERY_HIGH), mispricing_direction, mispricing_magnitude, expected_value_per_100, kelly_fraction, whale_signal (BULLISH/BEARISH/NEUTRAL), recommendation (STRONG_BUY_YES/BUY_YES/HOLD/BUY_NO/STRONG_BUY_NO/AVOID), key_factors, risk_factors, catalyst_events, reasoning, time_horizon. ## MCP Server Model Context Protocol server for AI agent integration. Tools: search_markets, analyze_market, get_whale_data, get_recommendations, check_credits. ## Rate Limits - Web (JWT): 120 requests/minute - API Key: configurable (default 60 req/min) --- # Blockcircle Games Free skill games built on real market, insider, congressional and prediction-market data: guess the chart, call the regime, size your risk. Anyone can play without an account. A free account keeps your streak on every device, ranks you on the leaderboards and earns Arcade Credits. Games hub: https://blockcircle.com/games - **The Tape** (https://blockcircle.com/games/the-tape): Five short-dated binary questions a day. State a probability, not a guess. - **Guess the Chart** (https://blockcircle.com/games/guess-the-chart): One masked chart a day, eight guesses, a new clue after every miss. - **Blockcircle Arena** (https://blockcircle.com/games/arena): Blind bar-by-bar replay against a deterministic AI persona. - **Beat the Indicator** (https://blockcircle.com/games/beat-the-indicator): Trade a hidden window head-to-head against a Blockcircle engine. - **Ghost Duel** (https://blockcircle.com/games/ghost-duel): Race another player's recorded run on the exact same seed. - **Whale or Noise** (https://blockcircle.com/games/whale-or-noise): Ten masked whale prints. Did price move, and how sure are you? - **Insider Tell** (https://blockcircle.com/games/insider-tell): Redacted insider filings. Did the insider's stock beat its sector over 90 days? - **Who Traded This** (https://blockcircle.com/games/who-traded-this): A masked legislator trade. Name the committee, chamber or party. - **Beat Congress** (https://blockcircle.com/games/beat-congress): Your virtual draft against a legislator's real disclosed book. - **The Draft** (https://blockcircle.com/games/the-draft): Five assets, one salary cap, priced off relative strength. - **The Gauntlet** (https://blockcircle.com/games/the-gauntlet): A simulated evaluation: profit target, hard drawdown cap, minimum days. - **Payoff Puzzle** (https://blockcircle.com/games/payoff-puzzle): Build the options structure that matches the target payoff. - **What Happens Next** (https://blockcircle.com/games/what-happens-next): A real historical setup, narrated. State your probabilities. - **Academy Quest** (https://blockcircle.com/games/academy-quest): The 16-week curriculum as a skill tree, with playable boss rounds. - **Autopilot Arena** (https://blockcircle.com/games/autopilot-arena): Tune a rule set in-sample. Ranked on a hidden hold-out window. - **Still Alive** (https://blockcircle.com/games/launch-or-rug): Six masked token launches from the last three years. Still alive today, or dead? - **Follow the Flow** (https://blockcircle.com/games/follow-the-flow): Ten masked options prints. Call whether the underlying followed the flow over the next hour. - **Call the Regime** (https://blockcircle.com/games/call-the-regime): A real macro dashboard with the date torn off. Price the next quarter, not guess it. - **Fair Value** (https://blockcircle.com/games/fair-value): A masked company's fundamentals as they stood. Call the verdict, then see what the market did. Arcade Credits are a loyalty reward with no cash value. They are not currency, cannot be bought, sold or transferred, and are redeemable only for Blockcircle plan discounts, subscription time and consulting sessions. All games are free to enter and decided by skill. Nothing here is a wager, and nothing here is investment advice. Games use historical and simulated data; they never place real orders. --- # Blog Posts ## The Revision Problem in On-Chain Data and How It Breaks Backtests URL: https://blockcircle.com/blog/onchain-data-revisions-backtests Published: 2026-10-21 Tags: backtesting, on-chain data, lookahead bias, data quality, advanced On-chain metrics get restated when providers find new wallet clusters, so the exchange-reserve series you backtest today is not what you saw live. Here is how that bias creeps in. I pulled an exchange-reserve series a while back to test a simple idea, that outflows leading a rally are worth fading into and inflows are worth respecting as supply. The backtest looked great. Then I re-pulled the exact same series from the same provider a couple of months later, same date range, same asset, and the numbers were different. Not off by a rounding error. Different in the places where the trades actually happened. That is when it clicked that on-chain data is not a fixed historical record the way a candle chart feels like it is. It gets revised, quietly, all the time, and if you are not accounting for that your backtest is quietly reading tomorrow's newspaper.Why on-chain numbers move after the factRaw chain data does not revise. A transaction that happened at a given block happened, and the hash is not going to change. What revises is everything built on top of the raw data, and almost every metric people actually trade off of is built on top of the raw data.The key word is entity. Most useful metrics are entity-adjusted. Exchange reserves, netflows, miner balances, whale holdings, stablecoin supply on exchanges, all of these depend on a provider having correctly labeled which wallets belong to which entity. And entity labeling is a moving target. Exchanges spin up new deposit addresses constantly. They rotate cold storage. They rebrand, merge, or get acquired, and the clustering that connects a loose pile of addresses to a single named exchange is a heuristic that improves over time.So when a provider discovers that a cluster of addresses it had been treating as unknown actually belongs to a large exchange, it does not just start counting them from that day forward. It typically backfills. It rewrites history so that those addresses were part of the exchange all along. Which is correct, in the sense that it makes the historical series more accurate. It is also exactly the thing that ruins your backtest, because the series you are now testing on incl [Content truncated. Read full post at the URL above.] --- ## How Futures Roll Costs Erode Carry and Trend Strategies URL: https://blockcircle.com/blog/futures-roll-costs-carry-trend Published: 2026-10-21 Tags: futures, roll-costs, contango, carry, cross-asset Rolling a futures position is a recurring cost or credit hiding in your P&L. Here is how contango and backwardation set it, and how to annualize the drag before you deploy. The thing that trips people up about futures is that a position you never touch still bleeds. You buy a contract, you hold your view, price does exactly what you thought it would, and somehow the return you booked is smaller than the move on the screen. The gap is the roll. Every futures contract expires, and if you want to keep the exposure past that date you have to close the expiring one and open the next month. The price you pay to do that is almost never zero, and over a year it adds up to a number that quietly changes whether a strategy is worth running at all.I ignored this for longer than I should have. When I was mostly trading spot and perps, roll was somebody else's problem. Then I started backtesting trend and carry ideas on quarterly futures, and the same signal that looked great on the underlying spot price looked mediocre once I forced it to actually roll every contract. Nothing about the signal changed. The mechanics of staying in the trade ate the edge.What the roll actually isA futures contract is an agreement to transact at a set date. As that date approaches, the contract stops being useful for anyone who wants ongoing exposure, so you sell it and buy a further-dated one. The two contracts trade at different prices, and that difference is the whole story.If the further-out contract is more expensive than the one you are leaving, the market is in contango. You are selling low and buying high every time you roll, so the roll is a cost. If the further-out contract is cheaper, the market is in backwardation, you sell high and buy low, and the roll pays you. That credit is roughly what people mean when they talk about a positive carry or a roll yield.The important part is that this has nothing to do with your directional call. You can be dead right on the direction of the underlying and still lose to the roll if you are long a contango market, or pick up a tailwind you did not earn if you are long a backwardated one. The move in the underlying and the [Content truncated. Read full post at the URL above.] --- ## Reading Rolling Volatility Before You Reset a Vol Target URL: https://blockcircle.com/blog/rolling-volatility-reset-vol-target Published: 2026-10-21 Tags: institutional, portfolio-risk, volatility, vol-targeting, regime A 30-day volatility reading carries a standard error of roughly 13 percent of itself, and one five-sigma day lifts it by a third. Read the 30, 90 and one-year lines together before a vol-target reset changes gross. The rolling volatility panel on the Distributions tab has a three-button toggle for 30D, 90D and 1Y, and on the capture I am working from the 30-day line reads 69.3% at its latest point, over a window running from 2026-04-11 to 2026-07-11. The Risk and Statistics block on the overview tab of the same page reports an annual volatility of 42.30% over 86 observations spanning 96 days. Those two figures are not in conflict. They are different windows on the same book, and the gap between them is the entire subject of this piece, because a vol target that resets off the first number and a vol target that resets off the second run at materially different gross exposure. Before you change the multiplier on a live book, you owe the process a decision about which estimator you are trusting and why. What a reset actually does to the book Volatility targeting scales positions by the ratio of target to realised. If realised doubles, gross halves. That is the whole mechanism, and it is well behaved when realised volatility is measured accurately and badly behaved when it is not, because the estimator's noise is transmitted directly into turnover. Assume iid returns for a moment, which is wrong but tractable. The sampling standard error of an estimated standard deviation from n observations is approximately the estimate divided by the square root of twice n. Run that on the three settings the panel offers, on a daily series: 30 observations gives a standard error of about 12.9% of the estimate. On a 69.3% reading that is roughly plus or minus 9 volatility points at one sigma. 90 observations gives about 7.5%, or roughly 5 points on the same reading. 252 observations gives about 4.5%, or roughly 3 points. So a 30-day estimator that moves from 55% to 69% has not necessarily told you anything. That move is inside a one-sigma band of an unchanged process. Fat tails and volatility clustering make the real band wider than the iid figure, not narrower, so treat 12.9% as a floor on you [Content truncated. Read full post at the URL above.] --- ## A Swing Trading Framework Built Around the 4-Hour Chart URL: https://blockcircle.com/blog/4-hour-chart-swing-trading-framework Published: 2026-10-21 Tags: swing-trading, 4-hour-chart, timeframes, trading-framework, setups A full single-timeframe swing playbook on the 4-hour chart: daily for bias, 4H for structure and entries, with stop placement and a two-target exit you can paper trade today. The 4-hour chart is the one I keep coming back to for people who have a job. Not because it is magic, but because it produces a manageable number of decisions. A 5-minute chart asks you to babysit it, and a daily chart moves so slowly you start inventing reasons to act. The 4H sits in a spot where a new candle prints six times a day for crypto, roughly twice during your waking hours, and you can check it at lunch and after dinner and not miss much. That cadence is the whole reason the framework works. You are trading with the trend over a few days, not scalping, so you do not need to be at the screen when the candle closes.What I want to give you here is a complete ruleset. Bias from the daily, structure and entries from the 4H, where the stop goes, and a two-target exit that takes money off the table without forcing you to nail the top. You should be able to paper trade this the moment you finish reading, which is the only honest test of whether a framework is actually a framework or just vibes.Daily for bias, 4-hour for the tradeStart on the daily and answer one question. Are price and the trend pointing up, down, or sideways? I keep this crude on purpose. A 20-period and 50-period moving average, or just where price sits relative to the last few swing highs and lows. If the 20 is above the 50 and price is holding above both, the bias is long and I only take long setups on the 4H. If it is the mirror image, short only. If the two averages are tangled and price is chopping across them, the bias is nothing, and I sit out. Most losing weeks I have ever had came from taking a clean 4H setup against a daily that was clearly against me, or worse, against a daily that had no direction at all.Then drop to the 4H. This is where you actually find the entry, mark structure, and place the stop. The daily tells you which direction you are allowed to trade. The 4H tells you when. Keeping those two jobs separate is most of the discipline. When I catch myself justifying a trade b [Content truncated. Read full post at the URL above.] --- ## Flipping Trending From 24 Hours to 30 Days Changes the Story URL: https://blockcircle.com/blog/trending-window-changes-ranking Published: 2026-10-21 Tags: retail, prediction-alpha, attention, leaderboards, time-windows On the 24 hour board I captured, nine of the ten top volume rows were single fixtures resolving within a day. The 7d and 30d toggles exist to tell you which of those was news and which was a market people keep coming back to. The trending tab defaults to 24 hours, and 24 hours is the window that flatters the wrong contracts. A single fixture with a two hour betting window can post more volume in an afternoon than a policy question does in a month, and the default view will rank it first every time. Sitting immediately to the right of that default are two toggles, 7d and 30d, and the useful thing about them is not either board on its own. It is the difference between them. What was actually on the 24 hour board Here is the top ten by 24 hour volume from the capture I have, in order: a Chelsea fixture at $4.1M, the Clarity Act contract at $1.9M, a Counter-Strike series at $1.7M, a League of Legends series at $1.1M, a US Open qualifying match at $1.0M, a Dota 2 playoff series at $953.0K, then three rows for the same Natus Vincere against Fnatic match at $872.7K, $872.2K and $859.0K, and a Fulham fixture at $856.4K. Count the categories. Nine of the ten are sports or esports fixtures resolving within about a day. One, the Clarity Act, is a policy question with an end date more than a year out. That is what a 24 hour attention window looks like on this asset class, and if you only ever open the default view, that is the entire universe you will ever be shown. The 24h, 7d and 30d toggles sit at the top left, above six separate leaderboards. Notice the right hand board: all ten Most Liquid rows are the same 2028 nominee market on the same venue, listed once per candidate. Ten rows is not ten opportunities, and that is true of the volume board too, where rows seven, eight and nine are one esports match sliced into Game 1, Game 2 and the series result. Ten rows are rarely ten events Before you compare anything across windows, deduplicate. The trending page tracks sixty markets, ten across each of six boards, and on the capture those sixty rows contained substantially fewer than sixty distinct questions. On the volume board, rows seven, eight and nine are the Natus Vincere against Fnatic fixture l [Content truncated. Read full post at the URL above.] --- ## Logging for Trading Systems: What to Record So You Can Reconstruct Any Trade URL: https://blockcircle.com/blog/trading-system-logging-trade-reconstruction Published: 2026-10-21 Tags: logging, observability, trading-bots, debugging, intermediate A trading bot did something surprising and you have three months to explain why. What you record now decides whether that is a two-minute query or a two-day forensic dig. A bot did something last quarter that I could not explain, and the log I reached for was a wall of plaintext lines that told me the order got sent and filled but not one word about why. I could see the trade. I could not reconstruct the decision behind it. That gap, between having a record of what happened and having a record of why it happened, is the whole reason I now treat logging as a first-class part of a trading system rather than something you bolt on after the strategy works.The uncomfortable part is that most logging in trading code is written for the happy path. You log that the signal fired, you log that the order filled, and it reads fine when you skim it live. Then something weird happens months later, you open the log to do forensics, and you realize every line was written to be glanced at, not queried. So this is the schema I have settled on, the correlation-ID discipline that ties it together, and the boring retention math that keeps the whole thing affordable when you are producing millions of events.The five stages you actually need to recordA trade is not one event. It is a chain of them, and the reconstruction only works if you capture each link separately and can walk from any point to any other. I think of it as five stages, and I log each one as its own structured record.Signal. The raw input that woke the strategy up. A price cross, a funding-rate shift, a new disclosure landing in your feed, whatever it is. Record the input values as they were at that instant, not a summary. If the signal came from a data source, record the source and the timestamp the data itself carried, which is often different from when you received it.Decision. The strategy's verdict and, more importantly, the reasoning inputs. Which rule matched, what the position sizing math produced, what risk checks passed or were skipped, and what the strategy chose not to do. The decision record is the one everybody forgets, and it is the one you will want most.Order. The intent [Content truncated. Read full post at the URL above.] --- ## Testing Whether a Political Trade Cluster Beats Random Grouping URL: https://blockcircle.com/blog/testing-cluster-significance Published: 2026-10-21 Tags: institutional, political-alpha, statistical-significance, null-model, cluster-validation Three members filing the same mega-cap inside ten days is roughly what pure arrival rates predict. Here is the null model that gives you the co-occurrence count a cluster has to beat before it is evidence of anything. A cluster card is a persuasive object. It puts a ticker at the top, a member count under it, a date window under that, and then the individual names in a table, and the visual grammar of the thing says these people acted together. Sometimes they did. Most of the time what you are looking at is the arithmetic of a busy filing calendar meeting a stock that a lot of people own. The discipline that separates the two is boring and it is the only thing that works. Compute how many co-occurrences you would expect by chance, and only treat the excess as information. Until you have that number, a cluster of three tells you nothing, because you do not know whether three is remarkable or whether three is Tuesday. What the card asserts and what it leaves unlabelled The Clusters tab opens with two view buttons, All Clusters and Bipartisan Only. Each card carries the ticker and company name, a bare number in the top right corner, a summary line reading member count, direction and USD volume, a window with explicit start and end dates, and then a table with columns Party, Politician, Type and Amount. On the card in front of me that reads: Microsoft, 45 in the corner, three politicians, Sale, 56.50 K USD volume, window of ten days. The rows underneath carry a party letter, a name, a Buy or Sell type and an amount. Three things about that construction matter for a significance test. The corner number is unlabelled, so I do not know what it scores or how it is scaled, and I am not going to treat it as a p-value or as a confidence measure just because it is a number in a prominent position. The direction on the summary line reads Sale while the visible rows include both Sell and Buy, so the header direction is a summary of a mixed group rather than a description of every row. And the Amount column reads as a dash on those rows, which means per-row size is not always populated even though the card reports a group volume. The summary line reads three politicians, Sale, but the rows unde [Content truncated. Read full post at the URL above.] --- ## Supercore, Trimmed Mean, and Sticky Price CPI Explained for Traders URL: https://blockcircle.com/blog/supercore-trimmed-mean-sticky-cpi Published: 2026-10-21 Tags: supercore, trimmed-mean-cpi, sticky-prices, inflation, federal-reserve Headline CPI is the last thing to move. Supercore, trimmed mean, and sticky-price gauges strip out the noise, and Fed officials watch them for a reason. The month a Fed governor stops saying the word "transitory" and starts saying "supercore services" is the month you should have already been reading the alternative inflation gauges. Headline CPI is what the news anchor reports and what your relatives argue about at dinner. It is also the slowest, noisiest, most backward-looking number in the whole release, and by the time it confirms a turn, the people who set policy have been talking about that turn for a while. If you only watch the top-line print, you are watching the scoreboard instead of the game.So this is a walk through the three gauges that actually show up in Fed speeches, what each one throws away on purpose, and how to use them to be early rather than surprised. None of this is exotic. The data is public and free. The edge is just in knowing which series leads and which one confirms.What each gauge is trying to fixEvery alternative measure exists because plain CPI has the same problem. A handful of wild, volatile categories can swing the whole number in a given month and tell you almost nothing about the underlying trend. Gas prices spike, used-car prices crash, airfares do something insane, and suddenly the headline moves half a point in a direction that reverses next month. The alternative gauges are all different answers to the question of how to see through that.Supercore is core services excluding housing. You start with CPI, strip out food and energy to get core, then strip out shelter too, and you are left mostly with labor-heavy services like medical care, transportation services, and recreation. The logic is that these prices move with wages, and wages move slowly, so supercore is closer to a read on domestic, demand-driven inflation than anything tied to commodities or the lagged housing data.Trimmed mean, published by the Cleveland Fed for CPI and by the Dallas Fed for PCE, does something more mechanical. Instead of deciding in advance which categories are noise, it throws out the most extreme [Content truncated. Read full post at the URL above.] --- ## Solo Staking vs Pooled Staking vs Exchange Staking URL: https://blockcircle.com/blog/solo-vs-pooled-vs-exchange-staking Published: 2026-10-21 Tags: staking, ethereum, validators, defi-basics, yield The base ETH staking yield is roughly the same everywhere. What actually differs is who holds your keys, who eats the slashing risk, and how fast you can get out. The thing nobody tells you about ETH staking is that the yield is the least interesting part of the decision. The protocol pays roughly the same base reward to everyone who helps secure the network, whether you are running a validator in your closet or clicking a button on an exchange. That number moves with how much total ETH is staked and how busy the network is, but the headline rate is not where the three routes diverge. Where they diverge is everything that happens around the yield. Who holds your keys, who eats the penalty when something goes wrong, how long it takes to get your ETH back, and how much you needed to start. Those four things are the actual product, and they are wildly different across solo, pooled, and exchange staking.So it helps to stop thinking about staking as one activity you are choosing to do, and start thinking about it as three different custody arrangements that happen to pay similar rewards.What each route actually isSolo staking means you post the full validator stake, which on Ethereum is 32 ETH, and you run the software yourself. You hold the withdrawal keys. You are the operator, which means uptime is your job and the penalties for messing it up land on you directly. Nobody is between you and the protocol.Pooled staking, usually via a liquid staking token, means you deposit any amount of ETH into a protocol that aggregates deposits, spins up validators run by professional node operators, and hands you a token that represents your staked position plus accruing rewards. You can hold that token, sell it, or use it as collateral elsewhere while the underlying ETH keeps earning. The big names issue a token that tracks your balance so you are never locked out of liquidity, at least in theory.Exchange staking means you already have ETH sitting on a centralized exchange and you toggle staking on. The exchange runs the validators, takes a cut, and credits rewards to your account. You do almost nothing. You also give up almost everything in [Content truncated. Read full post at the URL above.] --- ## Attributing Sleeve Profit to Entry System Not to Symbol URL: https://blockcircle.com/blog/entry-system-profit-attribution Published: 2026-10-21 Tags: institutional, momentum-trading-engine, attribution, entry-systems, trade-log The trade log groups by instrument, so a sleeve review reports that Ford did well rather than which of eleven entry systems paid. Rebuilding the other axis needs a key the log does not carry. After a decent quarter the useful question is not which positions made money. It is which of the mechanisms in the sleeve made money, because mechanisms are what you allocate to next quarter and positions are not. If your attribution comes back saying that Ford and silver did the work, you have produced a list of outcomes rather than an explanation, and the sleeve will be sized next quarter on the basis of two tickers that happened to trend. The Momentum Trading Engine documents eleven distinct entry systems spanning momentum, breakout and mean reversion, running across five synchronised timeframes. That is the axis a review should be cut on. The obstacle is that the trade log is not organised that way, and the join key you need is not in it. The log's grain is instrument, not mechanism The trades log tab presents rows with these columns: Asset, Strategy, Direction, Status, Timeframe, Order, Time in UTC, Entry, Current, PnL, Run-up, Drawdown, Source and Duration. Above the table sit three filter pills, All, Active and Closed, and a counter that at capture read 357 trades with six strategies holding nothing. The Strategy column looks like it might be the mechanism axis, but read the values. They were F MTE 1day, XAG/USD MTE 4h, BK MTE 1day and NDX MTE 1day. Every one of those names an instrument and a timeframe. The naming convention is symbol-scoped, which means grouping by Strategy is grouping by ticker with extra steps. You can produce a perfectly clean pivot table off that column and it will still answer the wrong question. Read the Strategy column rather than the Asset column and notice they carry the same information: F MTE 1day, XAG/USD MTE 4h, BK MTE 1day. There is no entry-system field anywhere in this table, which is why the regrouping described here has to be done against a mapping you maintain yourself rather than against a column you can select. Building the key the log does not carry The mapping from a deployed row to an entry system has to come [Content truncated. Read full post at the URL above.] --- ## Reference Class Forecasting for Prediction Market Traders URL: https://blockcircle.com/blog/reference-class-forecasting-prediction-markets Published: 2026-10-20 Tags: forecasting, base-rates, probability, methodology, prediction-markets Before you look at the price, build a base rate from historical analogs. It tells you whether you actually disagree with the market or just don't know enough yet. The mistake I made for a long time was reading the market price first. I would pull up a contract, see it trading at 62, form an opinion, and then go looking for reasons my opinion was better than 62. That is backwards, and it took me a while to notice why. Once you have anchored on a number, everything you read after that gets bent toward defending or attacking it. You stop forecasting and start arguing.The fix I stole from planning researchers is boring and it works. Before you look at the price, you build your own number from the outside. You find the reference class, which is just the set of past events that look like this one, and you count how often the thing happened. Incumbents running for re-election. Announced product deadlines. Drug approvals on first review. Central bank moves that were already telegraphed. Each of these has a rate that history has already paid out, and that rate is your starting point. Only after you have that number do you allow yourself to look at what the market thinks.Why the outside number beats your gutThe reason this helps is that your gut is very good at the inside view and very bad at the outside view. The inside view is the story of this specific case. This particular candidate, this particular scandal, this specific delay in this specific supply chain. The story feels rich and it makes you confident. The outside view ignores almost all of that detail and asks a dumber question. Of the last hundred things that looked roughly like this, how many resolved yes.The dumb question wins more often than it should, because the story you build about the specific case is where your biases live. You overweight the vivid detail you just read. You assume this time is different, and most times are not that different. The base rate does not know your story and does not care, which is exactly why it is useful as an anchor you did not choose.Here is the ordering I try to hold to, and I fail at it often enough that writing it down helps:Name the [Content truncated. Read full post at the URL above.] --- ## The Short Swing Profit Rule and Why Insiders Cannot Flip Shares URL: https://blockcircle.com/blog/short-swing-profit-rule-section-16b Published: 2026-10-20 Tags: section 16b, short swing rule, insider trading, holding period, sec rules Section 16(b) makes insiders hand back profit on any round trip inside six months. Read that way, every open-market insider buy is a six-month-minimum bet, which changes how you should copy it. The thing that took me a while to understand about insider buying is that the person doing it is not playing the same game you are, and not because they know something you don't. They might, but that is not the interesting part. The interesting part is that they are trading under a legal constraint you are not, and that constraint quietly shapes the timeframe of everything they do in their own stock. Once I started reading insider buys with that constraint in mind, a lot of them stopped looking like noise.What Section 16(b) actually doesSection 16 of the Securities Exchange Act covers company insiders, which for this purpose means officers, directors, and anyone who owns more than ten percent of a class of the company's equity. Section 16(a) is the part that makes them file their trades, so you can see them at all. Section 16(b) is the part that most people skip over, and it is the one that matters here.The rule is short swing profit disgorgement. If an insider buys and sells, or sells and buys, the same company's stock within a six month window, any profit that can be matched between those trades belongs to the company. Not a fine, not a penalty in the usual sense. The insider simply does not get to keep the gain. The company can sue to recover it, and in practice shareholders and their lawyers are the ones who go looking, because there is a fee in it for them. It is close to strict liability. It does not matter whether the insider actually had material nonpublic information. It does not matter if the timing was innocent. If the trades match inside six months and there is a profit, it gets clawed back.The matching is the part people get wrong. It is not first in first out and it is not average cost. Courts use a rule designed to produce the maximum possible recoverable profit. They pair the lowest priced purchase with the highest priced sale inside any six month window, then the next lowest with the next highest, and so on. You can lose money on paper across a peri [Content truncated. Read full post at the URL above.] --- ## What the Blackout Window Does to the Policy Signal URL: https://blockcircle.com/blog/fed-blackout-window-policy-signal Published: 2026-10-20 Tags: retail, macro-risk-scorecard, fed-policy, blackout-window, event-risk For roughly ten days before every rate decision the officials stop talking, so the policy read goes stale while positioning keeps building. Here is what that does to risk you are holding into the meeting. There is a stretch before every rate decision when the people who will make it stop speaking in public. No speeches on policy, no interviews, no answering the question at a conference. The window opens more than a week before the meeting and lifts the day after the decision. The exact dates are published on the central bank's own calendar and you should look them up rather than trust my paraphrase, because knowing precisely when the quiet starts is the entire point of this article. What matters for a portfolio is not the courtesy of the arrangement. It is that during those days the flow of new information about policy drops close to zero while the amount of money positioned around the outcome keeps climbing. That is an unusual combination and it has a predictable effect on how markets behave. An information-starved component is not a calm one The temptation is to read the quiet as calm. It is the opposite. Think about what a policy view is made of. It is made of the data prints that have arrived since the last meeting, plus the running commentary of officials reacting to those prints in public. During the blackout the first source keeps flowing and the second stops entirely. So every new piece of economic data lands without any official reaction to interpret it. The market has to guess how the committee read the print, and there is nothing to check the guess against. Guesses without a correction mechanism drift, and drift in a crowded position is how disorderly moves start. This is why a strong or weak data release during the blackout period often produces a larger move than the identical release would produce a month earlier. Nothing has changed about the data. What changed is that the usual dampener, an official saying in public that the print does or does not change their thinking, has been switched off. What the scorecard can and cannot tell you about this The Macro Risk Scorecard has a Policy tab, sitting fifth in the row after Dashboard, Sentiment, Regime and [Content truncated. Read full post at the URL above.] --- ## How TVL Gets Inflated and How to Adjust for It URL: https://blockcircle.com/blog/tvl-inflation-adjustments Published: 2026-10-20 Tags: tvl, defi, metrics that lie, protocol analysis, intermediate TVL is easy to inflate and easy to misread. Here is how the four common tricks work and a DefiLlama workflow to compare protocols on real outside capital. The number that gets quoted in every DeFi thread is total value locked, and I have watched it move a token more than any actual product update. Which is strange, because TVL is one of the softest metrics in the whole space. It looks like a bank balance. It reads like proof that real people parked real money here. A lot of the time it is closer to a headline the protocol wrote about itself, and once you understand how it gets inflated you stop reading it the way you used to.I am not saying TVL is useless. It is a fine first glance. But most of the ways it goes wrong are boring mechanical double counts, not fraud, and that is exactly why they survive. Nobody is lying. The accounting is just generous, and generous accounting compounds.The four ways the number gets bigger than the moneyThe first one is recursive lending. Someone deposits a stablecoin into a lending market, borrows against it, deposits the borrowed asset back, borrows again, and loops. Each turn of that loop counts as fresh TVL even though it is the same original dollar walking in a circle. A protocol built around leverage loops can show a TVL several times the outside capital that actually arrived. The money is real once. The counting happens on every lap.The second is a protocol counting its own token. If a project lets you stake its native token and then reports the dollar value of that staked pile as locked value, you are looking at a number that is mostly a function of the token price the project itself controls the supply of. Price doubles, TVL doubles, and no new capital showed up. When the token is thin and mostly held by insiders, this is less a measure of trust and more a measure of the current mark on their own inventory.The third is wrapped assets re-counted across chains. You lock ETH on the mainnet to mint a wrapped version, bridge it, and put the wrapped version to work somewhere else. Depending on how the trackers attribute it, both the original lock and the downstream deployment can show [Content truncated. Read full post at the URL above.] --- ## Adding Hysteresis So the Regime Flag Stops Flickering Weekly URL: https://blockcircle.com/blog/regime-flag-hysteresis-dwell-time Published: 2026-10-20 Tags: institutional, market-regime, hysteresis, turnover-control, signal-smoothing A dual-threshold band plus a dwell clock over the raw regime scores removes most of the switch chatter. The break-even inequality between turnover saved and lateness paid is four terms long. A regime overlay that changes its mind every few weeks is not an overlay, it is a transaction cost generator with a research paper attached. The failure is rarely in the classifier. It is in the decision rule bolted on top of it, which is almost always the argmax of a set of scores, evaluated fresh at every update with no memory of what it said last time. The fix is old and it is boring. Two thresholds instead of one, and a clock. What is worth doing carefully is the parameterisation, because the band and the dwell are not free and the arithmetic that decides whether they pay is short enough to write down. Where the flicker actually comes from The Market Regime Detector publishes raw scores alongside its classification. At the capture I am working from, the visible bars read Expansion 75, Alt Season 55 and Liq Rally 45, and the card carried a Risk-On Expansion label at 75 percent confidence. A twenty point gap between the top two scores is a comfortable margin. The problem is not the state the screen is in, it is the states it passes through. If two scores converge to within a few points, the argmax becomes a function of noise, and a rule that reads the argmax will switch on the noise, switch back, and switch again. Every one of those switches carries a full portfolio reweight if the regime drives allocation. There is a second and less obvious source. The engine fuses three scorecards, so the scores inherit the update cadences of all three. A component that refreshes on a different schedule from the others will drag the fused score across a boundary and back purely as an artifact of arrival timing, with no change in the underlying environment at all. Before building anything, check whether the module's own timeframe control already solves your problem. The screen offers 1D, 1W and 1M toggles, and reading the slower one is a form of smoothing that costs nothing to implement. Measure the switch rate on each toggle over your own snapshot history first. If the 1M [Content truncated. Read full post at the URL above.] --- ## Payment for Order Flow: How Commission-Free Trades Actually Get Filled URL: https://blockcircle.com/blog/payment-for-order-flow-explained Published: 2026-10-20 Tags: pfof, equities, brokers, market-structure, execution-quality Zero commission does not mean free. Here is how a retail equity order actually gets filled, why brokers sell your flow, and how to read the execution reports that tell you what it cost. The thing that took me a while to internalize is that when you hit buy on a retail brokerage app, your order almost never touches the exchange you are picturing. There is no floor, no matching engine on the NYSE quietly pairing you with a seller. Your order gets sold, in bulk, to a firm you have probably never heard of, and that firm decides what price you get. This is not a scandal exactly, and it is not free money for you either. It is just the plumbing, and once you see how the plumbing works, the zero-commission headline stops being the number you care about.Let me trace one order from the moment you tap the screen, because the path is more interesting than the marketing.Where your order actually goesSay you buy 100 shares of some liquid large-cap stock. Your broker does not run to the exchange. Instead it hands your order to a wholesaler, also called a market maker or an internalizer. A handful of firms dominate this business, and they pay your broker for the privilege of receiving that flow. That payment is the PFOF, payment for order flow. Fractions of a cent per share, sometimes a bit more, sometimes rebated in other forms, but it adds up to real money across millions of orders.The wholesaler now has your order and a decision to make. It can fill you from its own inventory, it can match you against another retail order flowing in at the same moment, or it can route you out to an exchange. In practice it fills you itself most of the time, because retail flow is the flow it wants. Retail orders are what people in this business call uninformed, which sounds insulting but just means you are not a hedge fund with a reason to think the stock is about to move. You are noise, statistically speaking, and noise is safe to trade against.Here is the part that confuses people. The wholesaler is required to fill you at or better than the National Best Bid and Offer, the NBBO, which is the tightest publicly quoted spread across exchanges at that instant. So the wholesaler [Content truncated. Read full post at the URL above.] --- ## Exchange Counterparty Risk: How Much to Keep on Each Venue URL: https://blockcircle.com/blog/exchange-counterparty-risk-allocation Published: 2026-10-20 Tags: counterparty-risk, exchange-risk, custody, portfolio-risk, capital-preservation The balance you see on an exchange is really an unsecured loan you made to the venue. Here is how I size that loan per exchange and sweep the rest to self-custody on a schedule. The thing that took me too long to internalize is that the number on your exchange dashboard is not your money in the way a bank balance is your money. It is a claim. You handed the venue your coins, and in return you hold a promise that they will give them back when you ask. Most of the time that promise is good. When it is not, you find out that you were an unsecured creditor the whole time, standing in line behind everyone else with the same promise. So the useful question is not whether an exchange is safe. It is how much unsecured credit you are willing to extend to it, and whether you are getting paid anything to take that risk. Usually you are not.Once you frame a balance as a loan you did not choose to make, the budgeting gets a lot clearer. You would not lend a random counterparty a third of your net worth with no interest and no collateral just because their app has a nice chart. But that is exactly what a lot of people do by leaving everything on one venue because moving it is annoying.Treat each venue as a credit line you are extendingStart by writing down, honestly, why any given balance is sitting on an exchange at all. There are only a few good reasons. You are actively trading it. You need margin posted against open derivatives positions. You are parked in something that pays a yield you actually understand and have decided is worth the risk. Or you are mid-transfer. Everything outside those buckets is idle credit you are giving away for free, and idle credit is the first thing to sweep.For derivatives specifically, the discipline is to keep only the margin the position requires plus a buffer, and nothing more. It is tempting to over-collateralize so you never get a margin call at three in the morning, and a buffer is fine, but a fat buffer is just a larger unsecured loan to the venue sitting there doing nothing. I would rather hold that buffer in self-custody and top up the margin account when a position actually needs it. The top-up takes a few min [Content truncated. Read full post at the URL above.] --- ## Defending an Insider Sleeve to an Investment Committee URL: https://blockcircle.com/blog/defending-insider-sleeve-investment-committee Published: 2026-10-20 Tags: institutional, insider-alpha, investment-committee, due-diligence, attribution The evidence pack an allocator needs before approving a filings driven sleeve, built only from what the attribution panel substantiates. Capacity, turnover, compliance, and the failure modes you name before they name you. An insider sleeve is easy to pitch and hard to approve, and the gap between those two is almost never about whether the signal works. It is about whether the person proposing it has done the unglamorous half of the work, which is capacity, turnover, compliance and a written account of how the thing fails. This is the pack I would want in front of me if I were sitting on the other side of the table, assembled from what the module can actually substantiate rather than from what a strategy deck usually asserts. Lead with the measurement convention, not the return The first slide should not contain a performance number. It should contain the definition of the number, because every question that follows depends on it and a committee that discovers the definition on slide nine will discount everything before it. The Insider Alpha performance view states its own convention on the page. For every past insider filing it tracks the actual forward return at 7d, 30d, 90d and 1yr and compares it to SPY over the same window, describing itself as proof of edge with no cherry picking and every scored filing counted. That is a population statistic over filings. It is not a portfolio, and the distinction is the single most important thing to establish before a discussion of returns begins. Say what follows from that plainly. There is no equity curve on the page, no drawdown, no turnover, no position count and no holding period. A forward return attribution answers the question of whether the raw material contains any edge. It does not answer whether a portfolio built from that material would have been ownable, and those are the two separate approvals you are asking for. What the panel substantiated on the day, stated with its date attached Give the committee the reading verbatim, with filters and date, and resist the temptation to round it into a headline. At capture, with the range on last 1yr, the side selector on All and the size dropdown on Any size, the header read N filin [Content truncated. Read full post at the URL above.] --- ## How to Backtest Prediction Market Strategies URL: https://blockcircle.com/blog/backtest-prediction-market-strategies Published: 2026-10-20 Tags: prediction-markets, backtesting, binary-contracts, odds-data, calibration Prediction market contracts resolve, expire, and jump on news, so equity-style backtesting breaks. Here is how I source odds data, handle resolution, model thin-book slippage, and score on calibration. The first time I tried to backtest a prediction market strategy, I did the lazy thing and treated a contract's price like a stock's price. Load the series, generate signals, mark to the close, plot the equity curve. The curve looked great. It was also completely fake, and it took me longer than I want to admit to see why. A prediction market contract is not a share of anything. It is a bet that pays one dollar if some event happens and zero if it does not, and almost every assumption baked into equity backtesting quietly breaks when you point it at a binary contract that resolves. So this is the methodology I ended up with. It is less about clever signals and more about not lying to yourself with the mechanics. Get them right and a mediocre edge shows up honestly. Get them wrong and a nonexistent edge shows up looking like a fortune. Odds data is not price data, and sourcing it is half the work The cleanest way to think about a binary contract is that its price is a probability. A contract trading at 0.62 is the market saying roughly a sixty-two percent chance. That framing changes what you need out of your historical data. You do not just want the last trade. You want the full order book, or at least a decent bid and ask, at every timestamp you plan to act on, because the spread on these things is where most of your imagined profit lives or dies. When you pull historical odds, get greedy about metadata. For every contract I want the resolution date, the actual resolution outcome, the market's open and close timestamps, and whatever unique identifier ties the contract back to the event so I do not accidentally stitch two different questions into one series. A lot of public datasets give you a midpoint tick history and nothing else, which is barely enough to fool yourself with. If you can get top of book with sizes, take it. The full book puts you in a much better position to model what your order would actually have done. One quiet trap here is survivorship. If your [Content truncated. Read full post at the URL above.] --- ## When Liquidity Expands but Price Falls, Trust Which One URL: https://blockcircle.com/blog/liquidity-expands-price-falls Published: 2026-10-20 Tags: retail, global-liquidity-scorecard, divergence, price-action, conflict-resolution The composite is rising and your position is not. The Regime tab keeps net flows and asset momentum in separate fields for exactly this reason, and there are three conditions where price is the better guide. The uncomfortable version of a liquidity framework is not when it is wrong. It is when it is right and you still lose money. The composite climbs, the policy field says easing, the driver tags say the Fed is adding liquidity, and the thing you bought on the back of all that is down eleven percent. At that point you have to decide which of the two things in front of you is the better guide, and the honest answer depends on which of three specific conditions you are in. The screen already separates the two, which is the first clue Look at the strip on the Regime tab. LIQUIDITY reads NEUTRAL with a sub-label of "Net flows". Four fields to its right, MARKETS reads NEUTRAL with a sub-label of "Asset momentum". Those are two different fields with two different sub-labels, sitting on the same row. That layout is telling you something before you read a single value. The module does not treat liquidity conditions and asset momentum as the same measurement. It measures them separately because they can and do disagree, and a design that expected them to move together would have collapsed them into one number. On the capture I am looking at they happen to agree, both printing NEUTRAL underneath a composite of 85 and a RISK-ON classification. The point is the structure, not today's values. When they diverge, you will see it as one field moving while the other does not, and you will see it there before you see it anywhere else. Two fields, four apart in the same row. LIQUIDITY measures net flows, MARKETS measures asset momentum. Both read NEUTRAL here, but they are deliberately separate measurements, and the moment they stop matching is the divergence this post is about. Why liquidity can genuinely rise while price falls The mechanism is not mysterious and it is worth having in your head before you need it. Liquidity is a condition, not a bid. More money existing in the system does not create a buyer for the specific thing you own at the specific price you paid. It changes [Content truncated. Read full post at the URL above.] --- ## The Crypto Basis Trade: What Cash and Carry Pays After Fees and Funding Flips URL: https://blockcircle.com/blog/crypto-basis-trade-cash-and-carry Published: 2026-10-19 Tags: basis-trade, cash-and-carry, funding-rates, delta-neutral, crypto-derivatives Long spot, short perp, collect funding. The headline rate looks great until you net out taker fees, borrow, spread, and the stretches where funding flips negative on you. The pitch for the basis trade is almost too clean, which is usually a sign you should slow down. You buy spot, you short the perpetual future against it, and you sit there market-neutral collecting the funding rate that longs pay shorts. When funding is running hot, the annualized number you see quoted on some dashboard looks like free money, and people size into it as if it were. Then they look at their actual account a month later and the yield is a fraction of the headline, or negative, and they cannot quite explain where it went. It went into fees, borrow, spread, and the days nobody warns you about when funding flips the wrong way.I want to walk through what the trade actually pays once you subtract all of that, because the gap between the quoted rate and the realized rate is the whole story here.The legs, and why the entry already costs youThe structure is long spot plus short perp on roughly the same notional so your net delta is close to zero. Price goes up, your spot gains and your short loses about the same. Price goes down, the reverse. What you are left holding is the funding rate, which perpetuals use to keep their price anchored to spot. When the perp trades above spot, longs pay shorts, and as the short leg you collect that. That collection is the yield you are after.The first thing that eats into it is getting in. You are crossing the spread and paying taker fees on both legs unless you are patient enough to post maker orders, and most people are not patient at entry because they are chasing a funding rate they can see right now. Say you pay taker on both the spot buy and the perp short. Round-trip that, because you also have to close both legs eventually, and you are looking at four fee events plus two spread crossings over the life of the trade. On a position you hold for a couple of weeks, that drag alone can swallow a meaningful chunk of what a modest funding rate would have paid.The rule of thumb I use: before you put the trade on, write down th [Content truncated. Read full post at the URL above.] --- ## Non-Custodial DeFi Automation: Bots, Keepers, and Session Keys URL: https://blockcircle.com/blog/non-custodial-defi-automation-keepers Published: 2026-10-19 Tags: defi, non-custodial, automation, smart-accounts, security Automate rebalances, limit orders, and liquidation protection on-chain without handing a private key to a server. A threat-model comparison of keepers, session keys, and approval hygiene. The first thing anyone building on-chain automation runs into is a very specific fork in the road, and most people take the wrong branch without noticing. You want your positions to rebalance, or a stop to fire when a liquidation threshold gets close, or a limit order to fill while you sleep. The obvious way to do that is to write a bot that holds your key, watches the chain, and signs transactions when a condition is met. It works on day one. It also means there is now a server somewhere with a hot private key that can move all of your money, and that server is the entire security model. Everything below is about keeping the automation and getting rid of that server.The reason this matters is that the failure mode is not subtle. A hot key on a box that holds your funds is a single point of total loss, and it does not have to be hacked in some clever way. A leaked environment variable, a dependency compromised upstream, a misconfigured log that prints the key, a backup in the wrong bucket. Any one of those and it is gone, all of it, in one transaction. So the real question is never whether to automate. It is how to automate in a way where the worst case is bounded to something you can live with.Keepers do the watching so you do not have to hold a keyThe cleanest mental model is to separate two jobs that people usually smush together. There is deciding what should happen and under what condition, and there is being the thing that is online at 3am to actually submit the transaction. Keeper networks exist to take the second job off your hands. You register an upkeep, which is really just a condition check plus an action, and a decentralized set of nodes competes to call it when the condition goes true. Chainlink Automation and Gelato are the ones most people reach for, and the pattern is the same.What makes this non-custodial is that the keeper never gets your key. Your funds live in a smart contract, usually a vault or a smart account, and the keeper is only allowed t [Content truncated. Read full post at the URL above.] --- ## Comparing Print Quality Across Venues Before You Commit Flow URL: https://blockcircle.com/blog/print-quality-across-venues Published: 2026-10-19 Tags: institutional, execution, market-trades, venue-selection, order-flow Market Trades posts analyst ideas rather than prints, so a venue scorecard has to come from venue tapes and your own fills. Here are the five statistics worth scoring and the half of venue quality no tape shows. A desk that allocates flow on relationship history and a vague sense that one venue "feels better" has no answer when a mandate holder asks why. Scoring venues on their printed tape is the right instinct, because a tape is evidence: it exists whether or not it flatters anyone, it covers the intervals you did not trade as well as the ones you did, and it can be recomputed by a sceptic. The scorecard is worth building. It just cannot be built from the module in the title, and the reason is worth thirty seconds of your attention before you brief anyone to go and do it. The page you would reach for, and what it holds instead Blockcircle's Market Trades page describes itself as a live execution feed across crypto, stocks and commodities. That phrasing points at time and sales. What the page actually carries is trade signals and positions posted by Blockcircle analysts: authored, timestamped entries with their own trade action, filtered by horizon rather than by venue. The filters on this page sort by holding horizon, not by venue, and each row is an authored post rather than an executed print. A venue scorecard needs size, timestamp and venue on every row, and none of those three columns exists anywhere on this surface. Look at the filter row in the screenshot: ALL, SHORT-TERM, MID-TERM, LONG-TERM. Those are horizon tabs. A venue study needs to slice by venue, and there is no venue dimension on this page because the page is not organised around execution locations. The posts themselves can carry rich detail about an idea, including exit prices, durations and strategy statistics as displayed on the day, but a post is a description of a trade thesis and a print is a record of a transaction. The scorecard consumes the second kind of object. So the sourcing is the first project. Venue tapes come from the venues, through their own market data products or their public feeds, and your own executions come from your execution records keyed on venue identifiers. That is the [Content truncated. Read full post at the URL above.] --- ## Stagflation Positioning Across Crypto, Commodities, and Bonds URL: https://blockcircle.com/blog/stagflation-positioning-crypto-commodities Published: 2026-10-19 Tags: stagflation, inflation-regimes, commodities, portfolio-construction, macro What actually held up in the 1970s regime, where those analogs break down today, and an honest watchlist for spotting stagflation risk before your portfolio finds out the hard way. Every time growth softens and prices refuse to cool at the same time, someone asks me the stagflation question. What do I hold. And the honest answer starts with a warning, because the only clean stagflation we have real data for is the 1970s, and half the assets people want to position with today did not exist back then. Crypto is the obvious one. So before anyone builds a portfolio around a regime, it is worth being precise about what the 1970s actually taught us, and where those lessons quietly stop applying.What actually held up in the 1970sStagflation is the ugly combination of stalling growth and stubborn inflation, and it breaks the standard playbook because the two problems ask for opposite responses. Central banks that fight inflation with tighter policy make the growth problem worse. Central banks that support growth with easier policy feed the inflation. There is no comfortable middle, and asset prices reflect that discomfort.The assets that held their real value in that decade were mostly the ones with a physical claim or a supply that could not be printed. Commodities broadly did well, energy in particular, because the inflation of that era was partly an energy shock and the underlying goods were exactly what was getting more expensive. Gold did well once it was allowed to float, for the same reason people reach for it now, a monetary asset nobody can issue more of. Real assets with pricing power, think producers of things people still buy when times are tight, generally beat the broad market.The assets that got hurt are just as instructive. Long duration bonds were the disaster. When inflation runs hot and rates climb to chase it, a bond that pays a fixed coupon for the next twenty years loses value fast, and you also lose purchasing power on the coupons themselves. Nominal growth stocks, the long duration equity of their day, struggled because a higher discount rate punishes cash flows that sit far in the future. The rough rule that came out of the de [Content truncated. Read full post at the URL above.] --- ## How Flash Loans Work and What They Are Actually Used For URL: https://blockcircle.com/blog/how-flash-loans-work Published: 2026-10-19 Tags: flash-loans, defi-basics, smart-contracts, arbitrage, lending-markets Flash loans let you borrow millions with zero collateral, as long as you pay it back in the same transaction. Here is how the atomicity works and what people actually do with it. The first time someone explained flash loans to me I assumed I was missing a step. You borrow millions of dollars, no collateral, no credit check, no signature from anyone, and the lender is fine with it. That does not exist anywhere in traditional finance, and my instinct was that it must be some kind of trick. It is not a trick. The catch is the whole design, and once it clicks, most of the scary headlines about them start to make more sense.The catch is time. You have to pay the loan back inside the same transaction that created it. Not the same day, not the same block in some loose sense, the same atomic transaction. If the money is not back in the lending pool by the last line of code, the entire thing reverts as if it never happened. Your borrow gets undone, whatever you did with the funds gets undone, and the lender is exactly where they started plus a small fee. That is the piece that makes the whole construct safe for the person lending, and it is worth sitting with for a second because it is genuinely different from how any loan you have ever taken out works.Why atomicity makes an uncollateralized loan safeOn most blockchains, a transaction is all or nothing. Every step inside it either succeeds together or fails together. There is no state where half of it happened. A flash loan leans entirely on that property. The lending contract hands you the funds, then calls back into your code to let you do whatever you planned, then checks one thing at the end. Did the pool get its money back plus the fee. If yes, the transaction settles and everyone moves on. If no, the contract throws an error and the chain rolls the whole transaction back to the starting point.So the lender is never actually exposed. From the perspective of the blockchain's final state, either the loan was repaid in full or it never happened. There is no scenario where you walk away with the money, because walking away requires the transaction to complete, and the transaction cannot complete unl [Content truncated. Read full post at the URL above.] --- ## Reconciling the Stock Rank With a Restricted and Watch List URL: https://blockcircle.com/blog/rank-reconciliation-restricted-list Published: 2026-10-19 Tags: institutional, asset-outperformer-engine, compliance, restricted-list, controls The board ends every row in a TRADE button and knows nothing about what your firm may not own. Here is the daily reconciliation that sits between the two, and the suppression log that keeps attribution honest. The Stocks tab ranked 522 names at capture and marked 100 of them as outperforming the benchmark basket. The engine runs a full scan every six hours, so that shortlist changes overnight without anyone at your firm deciding it should. Somewhere between the ranking and an order there has to be a step where the firm's own constraints are applied, and on this board that step does not exist. That is not a criticism of the module. A screening engine cannot know what your firm is restricted on, and it would be worse if it pretended to. What matters is that the absence is explicit in your workflow rather than assumed, because the last column on every scored row is an Action column containing a TRADE button, and the distance from a rank to a live order is one click. The board has no state for a name you may not own Walk the row from left to right. Asset, Type, Price, Mkt Cap, Vol 24h, Score, Conviction, Phase, then the return columns, then Action. There is nowhere for a suppression flag to live. A restricted name looks identical to an unrestricted one and sorts to exactly the same place, which means the restriction cannot be discovered by reading the board more carefully. It has to be applied from outside. The toolbar carries controls that partially narrow the universe. There is an exchange selector reading All Exchanges at capture, a country selector reading All Countries, a search box, a Filters panel and a Templates control. Those are venue and universe filters. They are not entitlement filters, and treating them as though they were is the specific mistake that puts a restricted ticker in front of a portfolio manager at eight in the morning. The practical shape of the control is therefore a reconciliation, run on a schedule, that takes the engine's output and the firm's lists as two inputs and produces a third artefact that is the only list anyone on the desk is permitted to read. Every scored row terminates in a TRADE button in the Action column. That is the control the [Content truncated. Read full post at the URL above.] --- ## Trading Event Contracts Through a Broker vs Directly on Kalshi URL: https://blockcircle.com/blog/event-contracts-broker-vs-kalshi-direct Published: 2026-10-19 Tags: kalshi, brokers, fees, market-access, event-contracts Your brokerage app now sells event contracts, but it routes them to a regulated exchange and quietly trims the toolkit. Here is what you give up for the convenience. A friend texted me a screenshot of an election contract sitting right there inside his regular brokerage app, next to his stocks and his cash balance, and asked whether that meant Kalshi was pointless now. Fair question. The contract in his app was the same contract, listed on the same regulated exchange, settling to the same outcome. But the two ways of reaching it are not interchangeable, and the gap shows up in exactly the places a serious trader cares about. So it is worth walking through what actually differs once you go through a broker instead of holding a native account. The short version is that most brokerage integrations act as an introducing broker. Your order still lands on the exchange's central limit order book, because that is where all these contracts have to trade to stay inside the regulatory framework. What changes is everything wrapped around the order. The broker sits between you and the book, and that middle layer is where fees, market selection, order types, and data get shaped to fit a mass-market app rather than a trading terminal. Where the money leaks: fees and spreads Native exchange pricing on event contracts is usually a per-contract fee that scales with how close the contract is to a coin flip, since a contract priced near the middle carries the most risk to the exchange and the most notional exposure per dollar. It is transparent and it is the same number the exchange charges everyone. When you route through a broker, you are often paying that exchange fee plus whatever the broker layers on top, and the broker markup is the part that varies wildly and is not always spelled out on the confirmation screen. Some brokers advertise commission-free event contracts, which should make you more suspicious rather than less. Free almost always means the cost moved somewhere you cannot see it. The usual place is the spread. If the broker is filling you a tick or two worse than the top of book, or routing in a way that does not chase the best ava [Content truncated. Read full post at the URL above.] --- ## What a Rising On-Chain Score With Flat Prices Is Telling You URL: https://blockcircle.com/blog/rising-onchain-score-flat-prices Published: 2026-10-19 Tags: retail, altcoin-market-scorecard, divergence, token-unlocks, accumulation Usage climbing while price goes nowhere has three honest explanations, and only one of them is good news. Each has a follow-up check you can finish in an evening. The setup that gets people into trouble is not a bad chart. It is a good one that has stopped working. Usage is climbing, the on-chain readings are strong, the network by every measure is busier than it was, and the price has gone sideways for two months. The instinctive read is that the market has not noticed yet and you have. That read is available. It is also the least likely of the three explanations, and taking it without checking the other two is how people end up holding the bag on a token that was doing fine operationally the whole way down. There are three honest explanations for usage up and price flat. Someone is selling into the strength. The usage was bought rather than earned. Or accumulation is happening quietly ahead of a repricing. Each has a specific check, and the checks are cheap. The score is a composite, so start by opening it up Before you interpret a divergence you need to know what is on the other side of it, and a composite score will hide that from you by design. The On-Chain tab of the altcoin scorecard leads with an On-Chain Health Score of 84, labelled Very Strong, described as being based on hash rate, active addresses, transaction volume and miner revenue trends. Underneath are four bars. Hash Rate reads 100, Tx Volume 100, Miner Rev. 100, and Active Addr. 36. A health score of 84 sounds like broad strength until you read the components. Three of the four bars are pinned at 100 and the only one with room to move in either direction is Active Addr. at 36, so a rise in this composite can come from a very small part of the underlying data. That matters for the divergence question. If three of four inputs are already at the top of their scale, then a rising composite is being driven by whatever is left, and the honest statement is not "on-chain activity is rising" but "one input moved". Do that decomposition first, every time, because a divergence between price and a composite is often a divergence between price and one narrow series [Content truncated. Read full post at the URL above.] --- ## Signal Decay and Crowding in Disclosure Trading Strategies URL: https://blockcircle.com/blog/signal-decay-disclosure-strategies Published: 2026-10-19 Tags: signal decay, crowding, alpha erosion, disclosures, advanced The post-disclosure drift that made congressional trade copying famous has quietly compressed. Here is how to measure decay in your own signal and where room still exists. The first time I backtested a naive congressional-disclosure strategy, the results looked too good, and that was the tell. Buy what the well-connected buy, hold for a couple of months, and the equity curve slopes up in a way that makes you want to trust it. What the clean backtest hides is that the edge you are measuring was mostly harvested in an era when almost nobody was watching. Once trackers, ETFs, and a few loud social accounts turned those filings into a spectacle, the drift they were feeding on got a lot thinner. The strategy did not stop working overnight. It just quietly stopped paying what the backtest promised.This is the crowding problem, and disclosure trading is one of the cleaner places to see it, because the signal is public, timestamped, and easy to copy. When a signal is easy to copy, it gets copied, and the excess return it used to carry gets competed away. The interesting question is not whether decay is happening. It is how you measure it in your own signal, and what you do once you accept that reaction speed now matters more than the thesis.How the drift gets compressedThe whole premise of a disclosure strategy is post-disclosure drift. A filing lands, the information is technically out, and yet the price keeps moving in the disclosed direction for days or weeks afterward because attention diffuses slowly. That drift is the alpha. It exists because the market is inefficient about processing a firehose of scattered filings.Crowding attacks the drift from both ends. On the front end, faster participants front-run the slow diffusion. If a hundred bots and three ETFs all read the same filing within minutes, the price adjustment that used to take a week now happens in a day, sometimes in an hour. On the back end, the crowd changes the character of the disclosed trades themselves. When people know they are being watched and mirrored, some of them get more careful, some get performative, and the average informativeness of the signal drops. You end u [Content truncated. Read full post at the URL above.] --- ## How Tokens Climb the Listing Ladder From Mid-Tier to Top-Tier URL: https://blockcircle.com/blog/listing-ladder-venue-tiers Published: 2026-10-19 Tags: institutional, alpha-hunter-suite, token-listings, exchange-tiers, sequencing Venue sequence is testable as a feature, but only after you fix the denominator, handle censoring and stop treating a dead token as a non-event. The construction, the confounds, and the number nobody has measured. The ladder story is one of the most durable pieces of folklore on a crypto desk. A token appears on a couple of second-tier venues, then a derivatives venue, then the majors, and everyone treats the early rungs as an early read on the late ones. Stated that way it is a story. Stated properly it is a conditional probability with a lag distribution attached, and the difference between those two framings is the difference between a feature and a narrative you are paying for. What follows is the construction. I am not going to hand you a conversion rate, because the module does not publish one and I am not going to invent one. What I can do is specify exactly which number you should be computing, and name the four things that will make your first attempt at it wrong. Turning the claim into something falsifiable Write the hypothesis so that it can fail. The version worth testing is this. Conditional on a token receiving its first listing at a venue in tier B during month t, the probability of a first listing at a tier A venue within h days is materially higher than the corresponding probability for tokens matched on age, chain, market cap and liquidity that did not receive the tier B listing. Three things are now nailed down that the folklore version left loose. There is a conditioning event with a date. There is a horizon. And there is a control group, which is the part almost everyone skips, because without it a high conversion rate might just be telling you that tokens which get listed anywhere are bigger and more liquid than tokens which do not. You also need a tier definition, and it will not come from the data. There is no tier field in the feed. Write your tiers down as a static mapping with a rationale, freeze it before you run anything, and version it. A tier map edited after seeing results is not a feature, it is a fit. What the feed supplies, and what it withholds The listings module carries three streams, upcoming, listed and delisting. With all three enabled [Content truncated. Read full post at the URL above.] --- ## Why Active Address Counts Mislead and What to Track Instead URL: https://blockcircle.com/blog/active-address-count-misleads Published: 2026-10-18 Tags: active addresses, on-chain metrics, data quality, network activity, intermediate Active addresses is one of the easiest on-chain numbers to fake and one of the most quoted. Here is how it gets inflated and what to watch when you want the real signal. I keep coming back to active address count as the metric that fooled me the longest. It looks like the cleanest read on a network you can get. Count the unique addresses that sent or received value in a day, plot the line, and you have a proxy for how many humans are actually using the thing. Except an address is not a human, it never was, and almost every fun way to inflate that number costs the attacker close to nothing. So you end up with a chart that spikes, a headline that says adoption is exploding, and a token that does not move because the people who trade it already know the spike was manufactured.The core problem is that the chain sees addresses, not entities, and not intent. One person can spin up ten thousand addresses in an afternoon. One exchange can settle for a million users through a handful of hot wallets. The mapping between addresses and real economic actors is many-to-many and it drifts constantly, so any raw address count is measuring the wrong noun.The four ways the number gets fakedMost of the distortion I have seen falls into a small set of patterns, and once you know them you start seeing them everywhere.Spam and dust campaigns. Someone sends a tiny amount of a token to hundreds of thousands of addresses, often as a phishing setup or a cheap marketing stunt. Every one of those recipients now counts as an active address for that block, and none of them asked for it or did anything.Inscription and mint waves. When a new inscription standard or a hyped mint goes live, you get a flood of near-identical transactions from freshly created addresses whose entire purpose is to grab a slot. It is real on-chain activity in the sense that gas got paid, but it tells you nothing about durable usage of the network for its actual purpose.Exchange and custodian batching. This one cuts the other way and is sneakier. A large venue might net internal transfers or route millions of user actions through very few addresses, so a huge amount of genuine economic ac [Content truncated. Read full post at the URL above.] --- ## How Your Quote Currency Choice Changes Execution Costs URL: https://blockcircle.com/blog/quote-currency-execution-costs Published: 2026-10-18 Tags: trading-pairs, quote-currency, liquidity, trading-costs, stablecoins The same coin trades against USDT, USDC, BTC, and fiat, and those pairs are not equally deep. A quick habit for checking pair-level liquidity before you click buy. I watched someone lose real money on a trade that was correct in every way except one. They wanted a mid-cap alt, they had the right thesis, they sized it sensibly, and then they bought it against USDC because USDC was the balance sitting in their account. The coin barely traded against USDC on that venue. The USDT book was ten times deeper. They ate a spread and a chunk of slippage for the privilege of not moving money between two stablecoins first, and the whole cost was invisible because the fill went through and the position opened. Nothing errored. It just cost more than it had to.This is the part of trading almost nobody checks, because the interface makes it feel like a settled question. You search for a coin, the exchange surfaces a pair, you buy. But the ticker you are handed is a routing decision the venue made for you, usually to show you the most active pair, sometimes just the first one alphabetically or the one they want to promote. The asset is the same. The market you are actually trading in is not.Why the same coin has four different marketsWhen you trade a coin, you are never trading the coin in the abstract. You are trading it against something specific, and each of those somethings has its own order book, its own set of market makers, and its own depth. A large cap like ETH will have thick books against USDT, USDC, BTC, and a couple of fiat currencies all at once, and the difference between them is small enough that you can mostly ignore it. As you move down the size curve, the pairs start to diverge fast.The general shape holds across most venues. USDT is usually the deepest quote for crypto-to-crypto flow, because that is where the volume has historically pooled. USDC tends to be thinner for the same asset, sometimes dramatically so, even though the two are close substitutes in your head. BTC-quoted pairs are their own animal, priced in satoshis and moving with two things at once, so your fill quality depends on the BTC market as much as the al [Content truncated. Read full post at the URL above.] --- ## Sizing Altcoin Positions Around Exit Liquidity and Market Depth URL: https://blockcircle.com/blog/altcoin-position-sizing-exit-liquidity Published: 2026-10-18 Tags: liquidity-risk, position-sizing, altcoins, market-depth, slippage For small caps the real constraint is not your risk percentage, it is whether you can actually get out. Here is the liquidity screen I run before sizing any thin position. The position sizing math everyone learns first is about risk. You pick a percentage of your account you are willing to lose on a trade, you set a stop, and you back out the size from the distance to that stop. It is clean and it works fine for anything liquid. Where it quietly falls apart is small-cap altcoins, because the whole calculation assumes you can exit at your stop price. On a thin coin you often cannot, and the moment you most need to exit is exactly the moment the book is emptiest.Depth is not a constant, and it hates you when the stop hitsHere is the part that catches people. Order book depth is not a fixed property of a coin. It is a snapshot of how many resting bids and asks happen to be sitting there right now, and that number collapses during the exact conditions that trigger your stop. When price starts dropping fast, market makers widen their spreads or pull quotes entirely, other holders are hitting the same bids you want, and the liquidity you eyeballed in calm conditions is simply gone. You sized the position assuming the book you saw on a quiet afternoon. You are exiting into a book that has maybe a quarter of that depth, with a dozen other stops firing alongside yours.So the stop you drew on the chart is a price you hoped to get. What you actually get is that price minus however much slippage it takes to eat through a book that just thinned out. On something illiquid that gap can be enormous, and it turns a planned two percent loss into something a lot worse. Your risk model said one number. The exit gave you another. The difference was never about your discipline, it was about depth you never checked.The cap rule: a fraction of average daily volumeThe first thing I do before sizing a thin coin has nothing to do with the chart. I look at average daily volume, and I cap the position at a small fraction of it. The reasoning is simple. If you hold a chunk that is large relative to how much trades in a normal day, there is no clean way out. You be [Content truncated. Read full post at the URL above.] --- ## Sorting Whale Consensus by Count or Volume Changes the List URL: https://blockcircle.com/blog/consensus-sort-count-versus-volume Published: 2026-10-18 Tags: retail, whale-alpha, consensus, sorting, prediction-markets The Consensus board has three sort buttons and each one answers a different question. Here is what the same five markets look like under each, and which sort belongs to which decision. The Consensus tab on Whale Alpha has three buttons above the list. Total Volume, Whale Count, Conviction. They look like a preference, the way a sort dropdown on a shopping site is a preference. They are not. Each one answers a genuinely different question, and if you read the top of the board without knowing which question you asked, you will act on the wrong row roughly a third of the time. The board defaults to Total Volume. That means the list you see when you land on the tab is answering the question where is the most whale money parked, which is a fine question and almost certainly not the one you had in mind. Three buttons, three different questions Total Volume ranks by dollars. It tells you where the size is. Whale Count ranks by how many distinct tracked wallets are in the market. It tells you how many independent opinions exist. Conviction ranks by how lopsided the money is between the two sides. It tells you how one-sided the money that is there happens to be. Those three properties come apart constantly, because one large wallet produces high volume with a whale count of one, and five medium wallets produce a lower volume with a count of five. Whichever button is lit decides which of those two rows you see first. The sort buttons sit top left and Total Volume is the lit one, which is why a 14 whale market at $18.6M heads the list. Look down the rows at the whale counts, 14 then 2 then 4 then 1, and you can see the ordering scramble the moment you press Whale Count instead. The same five rows, three different orders Here is the top of the board as captured, under the default Total Volume sort, with the dollars per whale worked out because that column is the one that shows you what is about to happen. MarketWhalesVolumePer whaleConviction JD Vance 2028 presidency14$18.6M$1.33MNO 62% 2026 Balance of Power, D Senate D House2$6.7M$3.35MNO 51% Dota 2 playoff match4$6.3M$1.58MNO 100% Nikki Haley 2024 presidency1$3.7M$3.70MYES 100% Now press Whale Coun [Content truncated. Read full post at the URL above.] --- ## Portfolio-Level Backtesting vs Testing Strategies in Isolation URL: https://blockcircle.com/blog/portfolio-level-backtesting Published: 2026-10-18 Tags: portfolio-backtesting, multi-strategy, correlation, capital-allocation, advanced Ten strategies that each pass their solo backtest can still sink together. How to test them jointly on shared capital and allocate by real behavior. The moment that made me stop trusting individual tear sheets was watching ten strategies that each looked fine on their own drop into the same drawdown on the same day. Every one of them had a clean solo backtest. Positive expectancy, reasonable Sharpe, drawdowns I could live with. I stacked them onto one account expecting the losses to cancel out the way the math promised, and instead they all leaned the same direction at the same time and the account took a hit none of the individual sheets had warned me about. Nothing was broken. Each strategy did exactly what it was designed to do. The problem was that I had never tested what they do together.Testing a strategy in isolation answers a narrow question. Does this rule set make money on its own, with its own capital, with no one else competing for the same fills. That is a fine question to start with, and it is where most people stop. The trouble is that you never actually trade a strategy in isolation. You trade a book. And a book has behaviors that none of its parts have.What breaks when you put them on one accountThere are roughly three ways a collection of individually healthy strategies goes wrong once they share capital, and they are worth naming because you can test for each.The first is correlated drawdown. Two strategies can look uncorrelated on average and still be tightly correlated in the tail. A momentum system and a breakout system feel like different ideas, but on the day a trend snaps they both bleed at once, because they are both long the same regime. Average correlation hides this. You care about what happens in the worst five percent of days, and that is usually much uglier than the headline number suggests.The second is margin contention. Each strategy sizes its positions assuming it has the capital available. Run them together and they draw on the same margin pool. When several want to add risk in the same window, you either can not fund all of them or you fund them and quietly run at a leverage [Content truncated. Read full post at the URL above.] --- ## Underwriting Pre-Profit Companies With Cohort Economics URL: https://blockcircle.com/blog/underwriting-preprofit-cohort-economics Published: 2026-10-18 Tags: institutional, company-valuation-engine, loss-making, cohort-analysis, underwriting A revenue multiple is a terminal margin assumption you have declined to write down. Cohort retention, payback period and contribution margin turn it into a number the committee can actually interrogate. Every revenue multiple contains an implied terminal margin. Eight times forward sales at a twenty five times exit earnings multiple implies a terminal net margin of thirty two percent, and if you did not intend to assert that, you asserted it anyway by choosing the multiple. The reason revenue multiples survive in memos is not that they are informative. It is that they let the analyst avoid writing down the assumption that is doing all the work, which means the committee cannot argue with it and the attribution afterwards cannot test it. Cohort economics is the alternative, and its real advantage is procedural rather than analytical. It relocates the argument from a multiple, where disagreement is aesthetic, to three quantities that can be tracked quarterly and either confirmed or falsified inside the holding period. Define the population before you define the method Read the headline against the counter directly beneath it. The title says composite scoring across 8,000 plus listed names while the COMPANIES tile reads 4,420, with 4,432 fully scanned underneath, and the three verdict buckets below sum to exactly the 4,420. Two different populations are on screen at once, an addressable universe and a scored universe, and the gap between them is where pre-profit names tend to sit, because a composite built on fundamental ratios needs ratios to exist. For an investment committee this is the first thing to pin down, since a hit rate quoted against the wrong denominator is not a hit rate. That distinction is not a criticism of the tool, it is the ordinary condition of any scored universe and the reason population definition belongs at the top of the memo rather than in an appendix. The board carries Price, Mkt Cap, P/E, a composite, a verdict, three sub-scores and a mispricing percentage. There is no revenue line, no gross margin, no customer count and no cohort disclosure anywhere on the view, and the only earnings-based ratio present is a price to earnings ratio, w [Content truncated. Read full post at the URL above.] --- ## Why Martingale Position Sizing Fails and What to Use Instead URL: https://blockcircle.com/blog/martingale-position-sizing-risks Published: 2026-10-18 Tags: martingale, position-sizing, risk-of-ruin, money-management, strategy-design The equity curve looks beautiful right up until the sequence that ends the account. Here is the ruin math on doubling into losers, and what to size by instead. The thing that gets people about martingale is that it works most of the time, and most of the time is exactly the wrong sample to judge it by. You double your size after every loss, so a single win recovers everything you lost plus one unit of profit. On a chart, that produces the prettiest equity curve you will ever see. It goes up in a nearly straight line, small and steady, day after day. Then one afternoon it goes to zero, and the drop is so vertical it barely registers as part of the same picture. The smoothness is not a side effect. It is the bait.I want to walk through why the ending arrives sooner than intuition says, because the gap between how it feels and how the math actually behaves is the whole story. Once you have seen the sequence written out, grid bots and recovery bots stop looking clever and start looking like a countdown.Run the sequence outSay you risk one unit and double after each loss. Trade one is one unit, trade two is two, trade three is four, then eight, sixteen, thirty two. Your total exposure after n losses in a row is not n units, it is roughly two to the power of n. After ten straight losers you are trying to put on a position more than a thousand times your starting bet, just to claw back to slightly positive. Most accounts and most exchange limits die well before that, which means you do not actually get to complete the recovery. You get stopped out mid-sequence, and the sequence is where all the losses live.Now the part people underweight. A run of losses long enough to end you is not a freak event you can wave away as bad luck. If each trade wins with probability p, a losing streak of length k shows up with probability roughly (1 minus p) to the power of k, and you get many chances at it over a few hundred trades. Even a coin flip strategy will hand you a run of eight or nine losers if you trade often enough, and it will do it a lot faster than the word rare suggests. The strategy is not betting that you avoid the bad streak. It [Content truncated. Read full post at the URL above.] --- ## Building a Comps Table for Altcoin Relative Valuation URL: https://blockcircle.com/blog/altcoin-comps-table-relative-valuation Published: 2026-10-18 Tags: relative-valuation, comps, token-valuation, fundamentals, screening Absolute valuation almost never works in crypto, but ranking peers on price-to-fees, market-cap-to-TVL, and FDV-per-user does. Here is the comps table I actually build. Every time someone asks me what a token is "worth" in dollars, I stall. I do not have a good answer and neither does anyone else, because a discounted cash flow on a protocol that might not exist in three years is theater. What I can do, and what actually holds up, is tell you whether a token is expensive or cheap relative to the five or six things it directly competes with. That is a much smaller question, and it is answerable. You build a comps table, you line up the peers, and you let the ratios do the arguing.The trick is that a comps table only works if the things in it are genuinely comparable, and in crypto they almost never are out of the box. Two tokens in the same sector can have wildly different float, different emission schedules, and different ways of counting the same activity. So most of the work is not picking the multiples. It is scrubbing the inputs until the multiples mean the same thing across every row.Pick a sector, then pick real peersStart narrow. A comps table across "DeFi" is useless because a lending market and a perp exchange and a liquid staking protocol do not share a business model. Pick one narrative sector where the tokens do roughly the same thing and earn money in roughly the same way. Decentralized perp exchanges. Liquid staking tokens. Onchain lending. Oracle networks. Layer-2 rollups. Inside one of those buckets, the peers actually compete for the same users and the same fees, so a multiple that is high for one is a fair question mark for another.Then be honest about who belongs in the set. I usually cap it at five to eight names. Include the clear leader even if it looks overpriced, because it anchors the range. Drop anything that is really a different business wearing the sector's clothing. If one "lending" protocol makes most of its money from a points program or a token launchpad, it is not a lending comp, and leaving it in will poison your median.The three multiples I trust, and what breaks themI keep the table simple. Thre [Content truncated. Read full post at the URL above.] --- ## The Gap Between News Publish Time and When You Could Trade It URL: https://blockcircle.com/blog/news-publish-time-tradable-time Published: 2026-10-18 Tags: institutional, research-workflow, market-analysis, event-study, timestamp-integrity A news record carries at least three clocks and a study can only honestly use one of them. Anchoring an event window on publish time books returns nobody on your desk was in a position to capture. Every news-driven event study on your shared drive has a t-zero in it, and in most cases that t-zero is whatever timestamp field came attached to the record. That choice is rarely discussed in the writeup, it is almost never defended in the appendix, and it is usually the single largest determinant of the abnormal return the study reports.The problem is not that publish time is a bad field. It is that a news record carries several clocks, they can sit days apart, and the one you can trade on is never the earliest of them. A study anchored on the earliest clock books returns that accrued before your desk had the information, and it books them as alpha. Three clocks in one row, and a fourth you have to derive Pull any news or research record apart and you will find at least three distinct moments compressed into what a downstream table treats as one date.The first is observation time: the moment the underlying inputs were sampled. For a computed product this is a model run time. For a filing it is the period end or the transaction date. This is the clock that describes the world the content is about.The second is publication time: when the item was released by its source. This is the field most vendors expose and the one most studies use, because it is the one that is populated.The third is arrival time: when the item landed in your systems, in your timezone, in a form a person or a process could read. Vendor latency, queue depth, parser failures and reprocessing all live in the gap between the second clock and the third, and none of them are visible in the record itself.The fourth clock is not in the row at all and you have to construct it. Tradable time is the first moment at which your desk could have transacted at a price you would accept, given your own reaction latency, your venue access and the liquidity available in that instrument at that hour. Everything before tradable time is someone else's return. Two stamps, eight days apart, on the same card This is [Content truncated. Read full post at the URL above.] --- ## Testing Trading Bots: Unit Tests, Exchange Sandboxes, and Chaos Drills URL: https://blockcircle.com/blog/testing-trading-bots-sandbox-chaos Published: 2026-10-17 Tags: testing, trading-bots, testnet, chaos-engineering, reliability A testing pyramid for trading systems, from sizing math to exchange testnets to chaos drills, plus a pre-deploy checklist for the bugs backtests never show you. The bug that scared me most wasn't in the strategy. It was a rounding error in position sizing that only showed up when the account balance crossed a specific decimal boundary, and it turned a modest long into something about ten times too big. The backtest never caught it because the backtest fed clean, pre-rounded numbers. The bot in production got the raw balance from the exchange, did its own arithmetic, and got it wrong. Nobody tests the arithmetic. Everyone tests the alpha.So this is a rough map of how I think about testing a trading bot, borrowed from the usual software testing pyramid but bent to fit the ways trading systems actually break. The short version is that unit tests catch your math, integration tests catch your assumptions about the exchange, and chaos drills catch everything else, which is most of it.Unit tests: the boring math is where the money leaksThe parts of a trading bot that most deserve unit tests are the least glamorous. Position sizing, fee and slippage math, order rounding to the exchange's tick and lot sizes, PnL accounting, and any conversion between quote and base currency. None of this involves the strategy. All of it can quietly move real money in the wrong direction.My rule of thumb is that anything that turns a signal into a number the exchange will act on gets a test, and the tests are mostly about edge cases rather than the happy path. What happens when the balance is zero. What happens when a fill comes back partial. What happens at the minimum notional, where the exchange rejects orders below some floor and your sizing logic has to either round up or skip the trade. What happens when you compute a stop that lands on the wrong side of the current price, which is a surprisingly common way to accidentally market-out of a position the instant you open it.A few cases I now write by default:Sizing at tiny balances, where rounding down to the lot size can produce a zero-size order or, worse, silently drop a decimal.Sizing at large [Content truncated. Read full post at the URL above.] --- ## Backtesting Cross-Asset Strategies Without Data Alignment Errors URL: https://blockcircle.com/blog/backtesting-cross-asset-data-alignment Published: 2026-10-17 Tags: backtesting, data-quality, cross-asset, quant, time-series, methodology The silent data bugs that make cross-asset backtests look better than they are, and a pre-flight checklist to catch them before you trust a single number. The first time I ran a backtest that mixed a crypto asset with an equity, the equity curve came out looking so clean that I almost believed it. Smooth, low drawdown, the kind of Sharpe you screenshot. Then I spent an afternoon pulling the thing apart and found that almost none of the smoothness was real. It came from how I had lined up the data, not from any edge in the strategy. That is the trap with cross-asset backtests. Single-asset tests fail loudly. Cross-asset tests fail quietly, and the failure usually points the wrong way, making a mediocre idea look like a good one.The reason is that the two assets do not live on the same clock, do not settle on the same calendar, and often are not even priced in the same currency. Every one of those gaps is a place where your data-alignment code can quietly fabricate returns. Below is how I think about the main ones, and a checklist I run before I trust any number from a mixed-asset test.The 24/7 versus session-bound clashCrypto trades continuously. A stock trades during a session, roughly six and a half hours on a US exchange, then goes dark until the next open. When you join those two series on a shared time index, you have to decide what the equity is doing during all the hours the market is closed. Most naive joins forward-fill the last equity price across the gap. That feels harmless. It is not.Forward-filling a closed market means your backtest can rebalance, or measure a signal, at a timestamp where one leg has a live price and the other has a stale one that nobody could have traded at. If your strategy reacts to the crypto move over a weekend and adjusts the equity leg, you are trading against a price that did not exist until Monday's open. The PnL you book is imaginary. Worse, it tends to be favorable, because the fill assumption is frictionless and instant.The rule I follow is that a leg can only transact when its own market is open. If I want a daily cross-asset series, I resample everything to the equity's tra [Content truncated. Read full post at the URL above.] --- ## Changing the Bin Count Changes What Your Distribution Says URL: https://blockcircle.com/blog/bin-count-changes-distribution-story Published: 2026-10-17 Tags: retail, portfolio-risk, distributions, histogram, binning The bin selector offers 10, 20, 30 and 50 over the same 87 trades. Three standard rules for choosing a bin count all land near the smallest setting, and here is the arithmetic plus the rule to fix it and stop. The distribution panel on the Performance page has a small control most people click without thinking about it. Four buttons labelled 10, 20, 30 and 50, which set how many bars the histogram is drawn with. It looks like a display preference, like choosing a chart colour. It is not. On a sample this size it changes what the chart appears to say, and if you click through all four settings before deciding which one to believe, you have already lost the argument with yourself. On the capture I am working from the panel header reads n equals 87, mean equals -0.53, median equals 0.00, and the PnL percent histogram is drawn at 20 bins running from about -17.34 on the left to about 1.00 on the right, with the count axis topping out at 66. Those numbers are enough to work out exactly what each of the four settings does to the picture, without clicking anything. The arithmetic of the four settings The data spans roughly 18.34 percentage points, from -17.34 to 1.00. Divide that span by each setting and you get the width of one bar. 10 bins gives bars about 1.83 points wide, and 8.7 trades per bar on average. 20 bins gives bars about 0.92 points wide, and 4.4 trades per bar on average. 30 bins gives bars about 0.61 points wide, and 2.9 trades per bar on average. 50 bins gives bars about 0.37 points wide, and 1.7 trades per bar on average. Those averages are misleading in a specific and important way, because the distribution is nowhere near uniform. At the 20-bin setting one bar holds 66 of the 87 trades. That leaves 21 trades to be spread across the other 19 bars, which is about one trade each. Read that again, because it is the whole point. Outside the central spike, every bar you are looking at represents one trade or two. The tail of that histogram is not a distribution, it is a list of individual trades drawn as rectangles. Push the setting to 50 and it gets worse: 21 trades spread over 49 bars, most of which are empty, and the eye reads the resulting picket fence as a w [Content truncated. Read full post at the URL above.] --- ## Human Capital and Crypto: Sizing Risk Around Your Career URL: https://blockcircle.com/blog/human-capital-crypto-allocation Published: 2026-10-17 Tags: human-capital, allocation, risk-capacity, financial-planning, diversification Your future paychecks are usually your biggest asset, and if a crypto or tech employer signs them you are already long the sector. Here is how to size the rest. The thing I keep coming back to when someone asks me how much of their net worth to put into crypto is that they are almost always answering the wrong question. They look at their brokerage balance, their coins, maybe a house, and they treat that pile as the whole picture. But for most people under fifty, the pile is not the whole picture. The largest asset they own is the stream of paychecks they have not collected yet. Economists call it human capital, and it dwarfs the portfolio for anyone who is more than a few years from retirement.Once you put that asset on the balance sheet, the allocation math changes in a way that is uncomfortable for a lot of people who work in this industry, myself included. If you draw a salary from a crypto exchange, a token project, a mining operation, or even a public company whose stock trades like a crypto proxy, you are already carrying an enormous long position in the sector before you buy a single coin. The paycheck is the position. And it is a leveraged one, because you cannot easily sell it, hedge it, or diversify it.Why your job is a hidden asset with a hidden betaThink of your career as a bond that pays a coupon every two weeks. The present value of all those future coupons is your human capital. For a young software engineer it can easily be the single most valuable thing they will ever own, worth more than the house and the 401k combined. Nobody puts it on a spreadsheet because you cannot log into an account and see the number, but it is there, and it behaves like an asset with its own risk profile.The risk profile is what matters here. A tenured professor has human capital that looks like a government bond, steady, and almost totally uncorrelated with markets. A commission salesperson at a startup has human capital that looks like a small-cap growth stock, high expected return, high variance. And someone paid by a crypto business has human capital that is correlated with the exact asset class they are thinking about buying [Content truncated. Read full post at the URL above.] --- ## Building a Depth Budget Across Politics, Sports, Crypto and Macro URL: https://blockcircle.com/blog/category-depth-budget Published: 2026-10-17 Tags: institutional, prediction-alpha, category-liquidity, capital-allocation, risk-budget Sports is 47,154 of the indexed listings and Politics is 12,898, but listing counts are not capacity. A depth budget allocates participation in resting books, which is the only constraint the screener will never show you. Every event contract book I have looked at closely concentrated somewhere its mandate never mentioned. Not because anyone decided to, but because the screener sorted by 24 hour volume, the top of that sort was sports, and the desk worked from the top. Six weeks later the exposure report says the book is a sports book with a politics label on it. The fix is not a rule about what to trade. It is a budget denominated in the thing that actually binds, which is how much resting depth you are willing to be in each category, and it has to be built before the screener is opened rather than reconstructed afterwards from fills. A listing count is not a capacity estimate The Prediction Alpha stats tab gives you the taxonomy in one bar chart, twenty buckets ranked by market count. Reading it down from the top: Sports 47,154, Other 33,257, Politics 12,898, Tech 4,610, Elections 4,164, Geopolitics 3,083, Crypto 2,822, Finance 2,255, Culture 1,510, Entertainment 1,325, then a tail running from Economics at 727 down to four buckets holding fewer than thirty listings each. If you allocate proportionally to that distribution you have built a sports fund. Sports is over forty percent of the universe by count, and the reason is structural rather than economic: every fixture, every leg, every game inside a series is its own listing. Volume follows the same shape. On the trending tab's 24 hour volume board as captured, nine of the ten rows were sports or esports fixtures, and three of those ten were the same match sliced into Game 1, Game 2 and the series result. Meanwhile the deepest single books on the platform sat in politics. The most liquid leaderboard topped out at $4.7M, and all ten of its rows were legs of one long dated nominee market on one venue. So the count distribution and the depth distribution point in opposite directions, and only one of them is a constraint on your book. The stats tab is where the taxonomy and the depth distribution both live. The Categories tile reads [Content truncated. Read full post at the URL above.] --- ## How to Recover After Blowing Up a Trading Account URL: https://blockcircle.com/blog/recover-after-blowing-up-account Published: 2026-10-17 Tags: account-blowup, recovery, post-mortem, risk-management, beginner The urge to redeposit right away is the enemy. Before you fund the account again, run the post-mortem, name the actual cause, and build a staged return gated on process, not profit. The dangerous moment after blowing up an account is not the blow-up itself. It is the next morning, when the loss has cooled from panic into a dull ache and some quiet part of you starts doing math. If I just put in a little more and catch one good move, I get most of it back. That thought feels like resolve. It is almost always the first move of the second blow-up. I have watched people go through this loop three and four times, and the pattern is consistent enough that I think of it less as bad luck and more as a process failure that nobody stopped to diagnose.So before you fund the account again, do nothing with money for a while. The recovery starts with a boring piece of paperwork, and skipping it is why most comebacks fail before they begin.Run the post-mortem before you touch the accountA blow-up feels like one event, but it is usually the last link in a chain, and the useful question is which link actually broke. Four causes account for most of them, and they do not share a fix. Get the cause wrong and you will treat the wrong disease.Sizing. Your individual positions were too large relative to the account, so a normal losing streak, the kind that happens by chance, was enough to cripple you. The strategy might have been fine. The bet size was the problem.Leverage. Related to sizing but worse, because leverage adds forced liquidation. Here the account did not die from being wrong. It died from being right eventually but getting stopped out or liquidated before the move arrived. A trade that would have worked on spot killed you on 10x.Strategy. The method itself had no edge, or had edge in one regime and quietly stopped working when conditions changed. No amount of good discipline saves a strategy with negative expectancy. Discipline just makes you lose more smoothly.Discipline. The plan was fine and the size was fine, but you did not follow either. You moved stops, you revenge-traded after a loss, you sized up on the one setup you felt emotional about. This [Content truncated. Read full post at the URL above.] --- ## Reading the Sector Heatmap Before You Pick a Political Trade URL: https://blockcircle.com/blog/sector-heatmap-before-picking Published: 2026-10-17 Tags: retail, political-alpha, sector-heatmap, screening, trade-selection Picking the ticker first means buying whatever filed loudest this morning. Start on the heatmap instead, choose the cell, then choose inside it, and on one capture that reduces ten sectors to two candidates. The default way to use a congressional disclosure feed is to open it, see which name is at the top, and think about buying that name. It feels like research because you are looking at real data. It is not research, because the ordering of the feed is driven by whoever filed most recently and by the size of their paperwork, and neither of those has anything to do with whether the trade is worth copying. The Heatmaps tab fixes the ordering problem by showing you the whole window at once instead of the most recent row. My rule is simple. Choose the cell before you choose the ticker. If you cannot say why a sector is interesting, you have no business picking a name inside it, and the heatmap makes that a thirty second decision rather than an hour of scrolling. What a single tile is telling you Each sector tile carries five things: a direction label in the corner, a buy count, a sell count, a USD notional, and five leading tickers on the bottom line. The window selector above the grid read 30 days on my capture, and the three view buttons beside it are Sector, Ticker and Committee Correlation. Take the Information Technology tile. 81 buys, 106 sells, 4.80 M USD, tickers GEN, IT, ORCL, QCOM and QLYS, labelled Sell. That is 187 separate disclosed transactions in one sector in one month, which sounds like a lot of conviction until you notice that more of them were sales than purchases. Now take Utilities: 7 buys, 9 sells, 152.50 K USD, tickers AEE, CMS, AWK, NEE, PEG. Sixteen transactions. A whole sector of the US equity market, one month, sixteen disclosed trades. Look along the direction labels in the top right of each tile before you read anything else. Eight of the ten read Sell. Only Communication Services at 30 buy against 24 sell, and Materials at 23 against 20, read Buy, and those two are where a long-only shortlist has to come from in this window. The reason the tile matters more than the row is that a single filing tells you one person did one thing. A tile tells [Content truncated. Read full post at the URL above.] --- ## The Dollar Milkshake Theory Explained and Stress-Tested URL: https://blockcircle.com/blog/dollar-milkshake-theory-explained Published: 2026-10-17 Tags: dollar-milkshake, dollar-strength, global-debt, macro-theory, bitcoin The dollar milkshake thesis says a tightening world gets forced into US assets. Here is the mechanism, where it breaks against history, and what a dollar spike means for bitcoin. The thing that finally made the dollar milkshake theory click for me was not any of the charts people draw. It was noticing how often a person who is bearish on the US ends up buying dollars anyway, and hating that they are doing it. That reluctant-buyer dynamic is the whole thesis in one sentence. When money gets tight everywhere at once, a lot of the world does not get to pick its favorite currency. It has to reach for the one it owes debt in, and an uncomfortable amount of that debt is denominated in dollars.The mechanism, stated plainlyThe theory, associated with Brent Johnson, starts from a boring plumbing fact. For decades the world borrowed in dollars. Companies outside the US, governments, banks, all of them issued dollar-denominated debt because it was cheap and deep and everyone accepted it. That debt does not care what you think about American fiscal policy. It has to be serviced in dollars, on a schedule, whether or not dollars are easy to find.So picture what happens when global liquidity tightens. Central banks pull back, credit gets expensive, risk appetite drains out of the system. Everyone who owes dollars still owes dollars, but now there are fewer of them sloshing around. They scramble for the currency to stay current on their obligations. That scramble is dollar demand, and it shows up precisely when you would expect the dollar to be weak, in a global slowdown. The milkshake image is the US drinking up the world's liquidity through the straw of dollar demand, pulling capital into Treasuries and US equities because that is where the deepest, most liquid dollar assets sit.The uncomfortable part for anyone who wants a clean bull or bear story is that the theory does not require the US to be healthy. It can be a mess. It just has to be the least-bad place to park forced dollar demand, and to have the only market big enough to absorb it. Ugly and dominant are not mutually exclusive.Where it holds up, and where I get nervousThe record is kinder to the [Content truncated. Read full post at the URL above.] --- ## Moving Crypto Off an Exchange for the First Time URL: https://blockcircle.com/blog/moving-crypto-off-exchange-first-time Published: 2026-10-17 Tags: self-custody, wallets, defi-basics, security, exchanges Your first self-custody withdrawal, done the boring careful way: pick a wallet, verify the address, match the network, send a test amount, then move the rest. The first time you move crypto off an exchange, the whole thing feels backwards. On the exchange your balance is just a number in a database, and moving it means pasting a long string of characters into a box and pressing a button that says the transaction cannot be reversed. No customer support line, no chargeback, no undo. That warning is not there for legal cover. It is genuinely how the thing works, and the reason beginners lose money on their first withdrawal is almost always that they treated an irreversible action like a reversible one.So the goal here is not to make you fearless. It is to give you a routine boring enough that the scary part becomes clerical. I do this the same way every time now, and I still do the boring version even for amounts I would not miss, because the muscle memory is the whole point. Let me walk through it the way I would if you texted me before your first send.Pick the wallet before you pick the coinYou need somewhere for the crypto to land, and that somewhere is a wallet you control. For a first transfer, a reputable software wallet on your phone or as a browser extension is fine. The distinction that matters is not brand, it is who holds the keys. A self-custody wallet gives you a recovery phrase, usually twelve or twenty-four words, and that phrase is the account. Whoever has it has the money. If an app never showed you a recovery phrase and never made you write it down, it is probably custodial, which means you have just moved from one exchange to another and gained nothing.Write the phrase on paper. Not a screenshot, not a notes app, not a password manager you also use to log into the exchange. The reason is boring and specific: the most common way beginners lose a self-custody wallet is not hacking, it is a phone that dies or gets wiped with the phrase living only on that phone. Paper in a drawer beats a screenshot in the cloud for exactly this failure mode.One more thing before you touch the exchange. Open the wallet and fig [Content truncated. Read full post at the URL above.] --- ## The Deployments That Left the Leaderboard Before You Looked URL: https://blockcircle.com/blog/vanished-leaderboard-deployments Published: 2026-10-17 Tags: retail, momentum-trading-engine, leaderboard, survivorship-bias, strategy-selection A live board shows what is still running, and nobody keeps a page for what got switched off. Three checks recover the missing denominator before you put money behind anything ranked on it. Every public leaderboard has the same design property, and it is not a conspiracy, it is just what a live list is. It shows the things that currently exist. When a strategy stops working and its owner turns it off, it does not move to a page called "strategies that stopped working". It disappears, and the board silently gets better looking without a single number changing. That is worth taking seriously before you copy anything from a ranking, because the check that costs you the most money is the one you never think to run. You are looking at the top of a list. The question is how long the list is, and how long it used to be. The most honest version of the board is an empty one When I captured the Momentum Trading Engine's leaderboard tab, it was showing nothing. Three scope tabs across the top, ALL TIME, CRYPTO and STOCKS. A RANK BY control on the right set to Overall Score, with the other five orderings being Total Return, Risk-Adjusted Return, Win Rate, Alpha and Profit Factor. In the middle, the words "No strategies found" and a suggestion to try a different asset filter. An empty board is the clearest possible demonstration of the point. The contents of that panel are entirely a function of who is currently running something. It is not a historical archive with a filter applied. There is no year in which it holds a fixed set of names. It holds whatever is live. Notice the mismatch between the top half and the bottom half of this page. Six tiles report 9 strategies, 350 backtested trades, a 96.6% average win rate, -11.3% average drawdown and a 35.06 average profit factor, while the ranked list under them contains zero entries. The tiles summarise the strategy library's backtests; the empty panel below is the ranking, and the two are not measuring the same thing. Three ways a name leaves without leaving a trace It helps to be specific about the mechanisms, because they are not equally bad and the checks that catch them are different. The first is the pla [Content truncated. Read full post at the URL above.] --- ## Building Implied Probability Distributions from Range Contracts URL: https://blockcircle.com/blog/implied-distributions-from-range-contracts Published: 2026-10-16 Tags: probability, analysis, kalshi, event-contracts, quant Bucketed markets like CPI-between-ranges quietly encode a full probability distribution. Here is how to stitch the buckets together, check them for coherence, and find the one that is actually mispriced. The thing I keep coming back to about range contracts is that most people trade them one bucket at a time, and that is the whole problem. You see a market for CPI coming in between 2.9 and 3.1 percent, you have a view, you buy or fade that single line, and you move on. But the exchange has usually listed the neighboring buckets too. Below 2.7, 2.7 to 2.9, 2.9 to 3.1, and so on up the scale. Once you look at all of them at once instead of one at a time, you are not staring at a handful of separate bets anymore. You are staring at a probability distribution for the underlying variable, drawn out for you by the crowd, and you can read it the same way you would read your own model output.That reframing changes what you are hunting for. You stop asking whether a single market is cheap. You start asking whether the shape is wrong.Turning the ladder into a distributionEach bucket price is roughly the market's probability that the outcome lands inside that range. A contract that settles at one dollar if CPI is between 2.9 and 3.1 trading at 22 cents is the crowd saying there is about a 22 percent chance of that. Line up every bucket on the same scale and you have a discrete probability mass function. The buckets are the bins, the prices are the heights of the bars.A few mechanical things to get right before you trust the picture. Prices on these venues include the spread, so use the midpoint between bid and ask, not the last trade, which can be stale or one-sided. Strip out fees if you are going to compare against real edge later. And be honest that a yes price is not exactly a probability. It is a probability under the market's risk preferences plus whatever premium sits in the spread. For most reading purposes that distinction is small, but it stops being small when you are deciding whether a two-cent gap is worth trading.Coherence checks come firstBefore you compare the implied distribution to anything, make it internally consistent, because it usually is not. The bucket [Content truncated. Read full post at the URL above.]