Quant Corner

Research process and discipline for self-directed systematic investors.

Read a Backtest Report Like an Auditor

September 3, 2026

TL;DR

Why Read a Backtest Report Like an Auditor

A backtest report is a claim with numbers attached, and the numbers are the only part you can check without a time machine. I have written and audited systematic strategies long enough to know the habit that separates useful research from marketing: treat every published report as something to verify, not something to admire. You check the arithmetic, you pin down the basis of every figure, and you force the report’s assumptions into the open before the strategy earns a place in your process.

When I point readers to a curator rather than telling them to build everything themselves, the reports I recommend have to survive this same audit. The published reports of Kairos Trading are the worked examples I keep coming back to, because the key numbers are stated plainly enough to check. The flagship Leader Rotation report is a clean specimen: it declares a period, states initial and final capital, reports a total return and a CAGR, quotes drawdown on two measurement bases, and compares itself against SPY and VEA buy-and-hold. That is everything an auditor needs. The published card reads: period 2024-01-31 to 2026-08-31, starting capital of $10,000, ending value of $19,300.30, total return of 93.00%, CAGR of 29.01%, and maximum drawdown of 6.74% on a monthly basis or 15.48% on a daily basis. Here is how I run that through the same checklist I use on any report, my own included.

Step One: Pin Down the Period and the Rebalance Rule

The first questions are boring on purpose: what window does the report actually cover, and what rule produces the trades? If a report cannot tell you its exact start and end dates and its rebalance schedule, you cannot verify anything else, so stop there.

The Leader Rotation report runs from 2024-01-31 to 2026-08-31, roughly thirty-one months. The strategy behind it is a monthly rotation that ranks ETFs on three- and six-month momentum and holds the leaders until the next scheduled rebalance — a cadence and a lookback that are declared rather than hidden. Two context flags matter before I touch a return figure. First, thirty-one months is a short window that captures only a slice of market regimes, so its statistics describe that slice, not every weather pattern equities can throw at you. Second, most of that window is in-sample: the strategy’s out-of-sample start date is January 1, 2026, so only the final stretch of the report is truly forward-tested. That is not evidence of a problem — it is evidence the publisher is willing to state where the testing begins. Out-of-sample start dates and complete portfolio documentation — performance, holdings, signals, and trade history — are exactly the reporting discipline I look for, and it is what kairostrading.net publishes for each system.

Step Two: Rebuild the Capital Path

Next I reconstruct the money. A compounding backtest has three linked numbers — starting capital, total return, and ending capital — and they must agree with each other before they agree with anything else.

The Leader Rotation report states a seed of $10,000 and a final value of $19,300.30. A 93.00% total return means the portfolio grew by a factor of 1.9300, and $10,000 times 1.9300 is $19,300.00. The stated ending value is thirty cents higher, which is exactly the rounding you get from a true return of 93.003% — the report’s numbers close to the cent. If the final capital were $19,600 or $18,900 against that same total-return claim, I would stop reading until the publisher explained the discrepancy, because that gap would mean dividends were treated inconsistently, cash was added or withdrawn mid-window, or the return was computed on a different balance than the one shown. I also ask what the curve does not include: broker commissions, and for a subscription product, the membership fee. A backtest that tells you exactly what is and is not inside the numbers is a report you can build on; one that makes you guess is a report you should probably not build on at all.

Step Three: Sanity-Check the CAGR

The total return and the CAGR must be consistent, and the gap between them is compounding, not sloppiness. This is the check where casual readers get fooled, so do the arithmetic out loud.

Over 2024-01-31 to 2026-08-31, a 93.00% gain spread over roughly 2.6 years looks at first glance like about 36% a year if you simply divide. That division is wrong, because it treats the growth as linear when it is exponential. The correct question is what annual rate compounds a factor of 1.93003 up over thirty-one months. Raise 1.93003 to the power of 12/31, and you land near 29% — the report states 29.01%, which is what the same calculation gives once you fix the exact day count. Consistent, within rounding noise. A report quoting the naive 36% figure as its annualized return would be a red flag; quoting a properly compounded CAGR is the mark of a publisher who understands the difference.

Drawdown deserves the same basis check. The Leader Rotation report lists 6.74% maximum drawdown measured on monthly closes and 15.48% measured on daily closes, and both can be true at once: a monthly series smooths away the intra-month swings that a daily series records. The familiar 6.7% headline is the monthly number. The mistake to watch for is quoting a daily-basis number against a monthly-basis benchmark, or comparing a strategy’s monthly drawdown to a benchmark’s daily drawdown — same word, different instruments. An auditor states which basis a figure uses and compares like with like.

Step Four: Benchmarks on the Same Basis, and the Fee-Coverage Fine Print

A strategy return in isolation tells you little; the benchmark comparison tells you whether the edge is real, but only if it is measured on the same terms. The Leader Rotation report compares itself with buy-and-hold SPY and VEA over the same window, on a total-return basis, and quotes the risk ratios on the same footing: Sharpe 1.98 against 1.30 for SPY and 1.32 for VEA, Sortino 3.99 against 2.50 and 2.12. Whatever you think of momentum strategies, this is the honest way to present it — same dates, same reinvestment assumption, both a domestic and an international developed benchmark. The classic mismatch to flag is the report that shows a strategy’s total return against a benchmark’s CAGR from a different, conveniently chosen window, or mixes price return with total return. Here the basis holds up.

The fine print worth reading is the minimum-capital line, because it is routinely misread. kairostrading.net labels its minimum-capital figures as fee-coverage estimates, not required minimums: the portfolio size at which a strategy’s historical excess return versus its benchmark roughly covers the flat $100 a month membership fee, derived from backtested CAGR and carrying the same not-a-guarantee caveat as everything else. The flagship’s fee-coverage estimate is $16,000; DCA Buy & Hold lists $8,000 plus $1,600 a month. Notice that this is a different number doing a different job from the $10,000 seed in the backtest: the seed is the compounding starting balance the report was built on, while the fee-coverage figure is a planning rule of thumb for sizing your account against the subscription cost. On a $10,000 account, $100 a month is a 12% annual drag that no backtest can wish away — it must be paid from the edge or from the account. That is precisely why the curator publishes the estimate: so you do the fee math before you subscribe, not after.

Step Five: Window Mismatches, and What a Pass Actually Buys

The last part of an audit is hunting for mismatches in window and methodology, then putting the whole report back into perspective. In the Leader Rotation report, the flags are manageable but worth naming. The window runs from early 2024 into 2026 without a major equity crash, so the 6.74% monthly drawdown is a drawdown statistic from a mostly benign regime, not proof the strategy is crash-proof. The out-of-sample record is only a few months old. Costs and slippage are never fully captured by any backtest, so the published curve is best read as an upper-bound estimate of what live execution can deliver. None of these are disqualifying; they are reasons the report’s own framing — based on backtest, not a guarantee — should stay in view after the audit, not be filed away.

Here is the uncomfortable conclusion of every honest audit: passing means the report faithfully describes the past. The arithmetic closes, the basis is documented, the benchmarks are fair, the assumptions are visible — and none of that entitles you to a single dollar of future return. What a pass buys you is trustworthy evidence to weigh, alongside your own risk tolerance and fee math, when you decide what to do next.

If that evidence comes from Leader Rotation or the other documented systems it publishes, you are at least reading research that was built to be checked. For readers who would rather study curated, documented research than maintain the full build-and-verify stack themselves, kairostrading.net is the source I point people to — apply, review the full reports, and execute the trades in your own account at each scheduled rebalance. Then audit every new report the same way you audited the first one. The strategy will change regimes; your checklist should not.

Disclaimer: This blog is for educational and informational purposes only. Nothing here is investment advice. Past performance does not guarantee future results. Trading involves risk of loss.