There is a version of every backtest that looks amazing, and it is the cropped screenshot of the equity curve. I get sent those a lot. The line runs bottom left to top right, the sender asks what I think, and the honest answer is that the panel they cropped in is the one I trust least on the entire sheet. A full tear sheet has eight or ten panels, and when there is a deal-breaker, it usually lives in the ones nobody screenshots. So this is how I read a tear sheet top to bottom, what each chart is actually for, and the short list of things that make me close the tab.
The panels everyone reads first
Start with the equity curve, but start with the axis before the line. If the strategy compounds over several years and the chart is on a linear scale, the early period gets visually flattened and the recent period looks explosive, which hides early drawdowns and makes late-sample luck look like skill. Insist on a log scale for anything longer than a couple of years. Then do the hand test. Cover the final quarter of the chart and ask whether you would still be impressed, because a surprising number of curves are one favorable regime wearing a track record. The last thing I check is smoothness. A curve with almost no wiggle is either a genuinely rare market-neutral machine or, far more often, a strategy that sells insurance in some form and has simply not been billed yet. Smoothness is a question to investigate rather than a comfort.
The drawdown chart, usually drawn as an underwater plot, tells you what living with the strategy would have felt like. Two numbers matter, depth and duration. Depth is the maximum drawdown everyone quotes. Duration is how long the strategy sat below its previous high, and it is the number that actually breaks people, because nobody abandons a system during a two-week dip but almost everyone abandons one after a year underwater. The red flags here are drawdowns that recover implausibly fast, which usually means optimistic fills or lookahead bias creeping into the simulation, and a maximum drawdown that looks tiny next to the annual return. A common rule of thumb is to assume the live drawdown will be meaningfully worse than the backtested one. Mentally doubling it is a reasonable starting point.
The monthly return table is where I check for concentration. Find the best two or three months and ask what the track record looks like without them, because plenty of multi-year backtests are two lucky months plus noise. Then scan the losers. If the losing months cluster in one stretch, the strategy has a regime it cannot handle, and you want to know exactly what that regime was before assuming it will not repeat.
The panels people skip, where the deal-breakers live
The return distribution histogram is the panel beginners skip most reliably, and it is often the single most informative chart on the sheet. What you want from it is the skew. A strategy with a high win rate and negative skew makes many small gains and takes rare large losses, which is the classic short-volatility shape. The dangerous part is that the rare large loss may not appear in the sample at all, so the backtest shows years of premium collection and omits the claim. Historically, plenty of strategies that blew up looked wonderful on every other panel and showed their hand only here. Positive skew with a mediocre win rate, the typical trend-following shape, tends to be a more honest profile even though the monthly table looks uglier. Neither shape disqualifies a strategy on its own, but not knowing which one you are holding does.
The rolling Sharpe panel exists because the headline Sharpe is an average, and averages hide decay. A strategy that did brilliantly in the first third of the sample and roughly nothing since will still print a respectable full-period number. Look at a six or twelve month rolling window and compare the recent third of the chart to the early third. A steady downward staircase usually means the edge is getting crowded or the market structure it exploited has changed, and you would be paying live costs to trade someone else's former edge. I would much rather see a modest, stable rolling Sharpe than a spectacular average built on ancient history.
Rolling beta is the lie detector on the sheet. If the description says market neutral and the rolling beta chart spends every bull market around 0.6, the strategy owns beta with extra steps, and an index fund would give you the same exposure for less. The exposure block deserves equal suspicion. Check time in market, because a system that is invested a fifth of the time and quotes returns on committed capital is quoting a different number than you think you are reading. Check gross leverage. And check turnover, because turnover multiplied by a realistic cost per trade tells you whether the stated returns survive contact with an actual exchange. If the sheet never states its cost and slippage assumptions, that alone ends the conversation for me, since a high-turnover strategy with free fills can make almost anything look like an edge.
The order I actually read them in
Put together, this is the pass I run on any tear sheet, in order, with the point where each step can end the process.
- Fine print first. Sample period, cost and slippage assumptions, and whether any out-of-sample split exists. Missing cost assumptions, or a sample that conveniently starts right after a crash, and I stop here.
- Drawdown chart. Depth, duration, recovery speed. Deep drawdowns that recover in days suggest simulation artifacts rather than resilience.
- Monthly table. Remove the best two or three months in your head. If nothing interesting remains, stop.
- Return distribution. Identify the skew and decide whether the tail risk is survivable at your intended size. High win rate plus negative skew plus a short sample is the classic trap.
- Rolling Sharpe. Compare the last third of the sample to the first. A clear decay trend means the edge probably left before you arrived.
- Rolling beta and exposure. Test every neutrality claim against the chart, and sanity check turnover against realistic costs.
- Equity curve last, on a log scale, as a summary of what you already learned rather than a first impression.
The ordering matters because the equity curve is persuasive while the rest of the sheet is informative, and you want the informative parts to set your expectations before the persuasive part gets a vote. This is also roughly how we ended up laying out results in the Blockcircle backtester, mostly because I kept watching people make decisions off the one chart that is easiest to fool.
One habit worth stealing before your next tear sheet. Write down your disqualifiers in advance, a maximum drawdown you can live with, a minimum sample length, a beta ceiling, and only then open the report. Reading a tear sheet with your walk-away conditions already written is a very different exercise from reading one while hoping it passes, and the second kind of reading approves almost everything.