The method in the title is a real method and it is the right one. Take each of your own executions, locate those exact prints in the venue's tape, strip them out of the surrounding print series, and compare the price path around your activity against the path the rest of the tape describes. Whatever decays back after your last fill is temporary impact. Whatever does not decay back is permanent. That decomposition is the only honest basis for a participation cap, because a cap set without it is a number somebody liked the sound of.
It requires a print tape in which your own executions are identifiable. Blockcircle's Market Trades page is not that page, and I would rather say so in the second paragraph than build five sections on a premise that collapses the moment a quant on your desk opens the module.
What the Market Trades page actually carries
The page describes itself, directly under the heading, as trade signals and positions from Blockcircle analysts, presented as a live execution feed across crypto, stocks and commodities. The phrase "live execution feed" is what does the damage. It reads like time and sales. It is a stream of posts.
The controls at capture were a Templates button, a search box that takes a keyword, ticker or symbol, a subscribe action, and a row of horizon tabs reading ALL, SHORT-TERM, MID-TERM and LONG-TERM. Each post carries an author, a timestamp and its own trade action. The content of a post can be detailed: one close-long card on silver laid out an exit price of 64.27 dollars, a result of plus 8.02 percent, a duration of 28 days and 4 hours, a run-up of 8.28 percent and a drawdown of 8.00 percent, with a stop loss of 11.29 percent and a strategy performance block reporting a 91.2 percent win rate over 34 trades and an average hold of 19 days and 1 hour, as displayed on that day.
Notice what is absent from the screenshot. No size column. No venue column. No aggressor flag. No sequence of prints at all. A post describes an idea and its outcome. An impact study consumes a print series and your own fill records joined to it. These are different objects and no amount of filtering turns one into the other.
The three joins that make own-print isolation possible
If you want the study, here is what has to be assembled before anyone writes a line of analysis. None of it comes from a signal feed.
First, your own execution records as the venue reports them, keyed on the venue's execution identifier rather than your internal order id. This is the join that fails quietly. An internal parent order id will not match anything in a public tape, and reconstructing the mapping after the fact from price and quantity produces false matches on exactly the busy intervals you most want to study.
Second, the venue's own trade tape for the instrument, covering a window that starts before your first child order and runs well past your last.
Third, one clock. Use venue timestamps for both sides of the join and record the resolution in the report header. If the venue publishes to the millisecond and your execution records land to the second, your isolation is only as good as the coarser of the two, and you should say that in the methodology note rather than let a reader assume otherwise.
There is a structural caveat worth establishing before you commit engineering time. Public tapes differ in how they aggregate. Some publish one print per aggressing order across multiple resting levels, some publish per level, some collapse further. Whether your child fills appear individually or merged with somebody else's is a property of the specific venue's feed, not a general truth, so check the format for each venue you care about before designing the report around an assumption.
Separating temporary from permanent without over-claiming
The measurement itself is simple arithmetic and the difficulty is entirely in what contaminates it.
Fix a reference price at arrival, compute the volume weighted price of your own fills, then sample the market price at a set of horizons after your last fill. Something like one minute, five minutes, thirty minutes and the next session open gives you a decay curve rather than a single point, and a curve is what tells you whether your chosen horizon is doing any work. The displacement that persists at the far end of the curve is your permanent estimate. Total displacement minus permanent is temporary.
Three contaminants will each produce an answer that is confidently wrong.
Market drift is the easy one. Measure the residual against a contemporaneous benchmark, an index or a peer basket, so that a market that moved while you were working does not get booked as your impact.
Selection is the hard one and it never fully goes away. You traded because a signal fired, and if the signal has any predictive content then the price was going to drift in your direction whether or not you participated. Raw post-trade drift therefore overstates permanent impact for any book with genuine alpha. The long-standing treatment in the market microstructure literature is to model the two components jointly rather than to attribute the whole residual to trading, and the honest institutional version is to report permanent impact as a bounded estimate with the signal-decay assumption stated, not as a measured constant.
Sample size is the third. A single parent order tells you nothing. Impact estimates are dominated by noise at the individual order level, and the unit of analysis has to be a population grouped by instrument and by size bucket, with your own non-trading days serving as the control for what ordinary price behaviour looks like in that name.
Turning the estimate into a cap the scheduler enforces
An impact curve is an input, not a policy. The policy step is to state a target and read the cap off the curve.
Fit measured impact against participation rate, defined as your executed volume over market volume in the same interval. Then the investment committee, not the desk, states the target: something of the form "temporary impact at or below a stated basis point figure at the seventy fifth percentile of orders in this liquidity tier". The number is a choice about how much cost the strategy can carry, and it should be argued on the strategy's expected edge rather than borrowed from a peer.
Two implementation details decide whether the cap is real. It has to bind per interval rather than per day, because a daily cap is trivially satisfied by a schedule that puts everything into the first twenty minutes. And it has to be a parameter the scheduler consumes, not a line in a document the trader is expected to remember. A cap that lives only in a policy PDF is a cap that binds during audits and at no other time.
The output of this exercise is also your capacity number, which is the thing the committee actually wants. If the cap holds impact under target at a given participation rate, then strategy capacity in that name is that rate multiplied by the volume you are willing to assume persists. Present it with the volume assumption visible, because that assumption is where the estimate is fragile.
What the methodology note says when the tape is not available
Plenty of desks cannot get a usable tape for some of the venues they trade, and pretending otherwise is how a fabricated decomposition ends up in a client pack.
The defensible fallback uses only your own fills. Arrival price shortfall gives you a total cost figure that is unarguable because every input is yours. A reversion test on your own fill sequence, comparing the price at your last child fill against the price some fixed interval later, gives you a directional read on whether your execution is leaving a footprint, without claiming to size it. Interval VWAP comparisons tell you whether your schedule was worse than a naive one.
What none of those produce is a temporary and permanent split. If you cannot isolate your prints, do not publish the split. Report total cost, report the reversion diagnostic, and write one sentence in the methodology note saying that impact decomposition is unavailable for these venues and why. An allocator who reads that sentence will trust the rest of the pack more, not less, and the desk that quietly interpolated a decomposition is the one that gets taken apart in the operational due diligence meeting.