A systematic strategy, fictional, for the sample report. A measurement of the history supplied, under the method published by AiNT, with every assumption stated.
The claim as stated
Window
01.01.2023 to 31.12.2025
Window fixed
2026-10-01 09:00 UTC, before the measurement
Basis
Returns as given by the client. Not re-marked against independent prices.
Probability that the true Sharpe exceeds zero, one test, corrected for skew and kurtosis.
Trials
25, declared
As declared by the client. AiNT did not observe the search and cannot verify the count.
Deflated Sharpe ratio
9.4%
Probability that the true Sharpe exceeds what the best of 25 zero-skill trials would show (1.15 annualised). Variance of trial Sharpes: null assumption, 1 / (n - 1).
Minimum backtest length
25.7 years
For 25 trials at this Sharpe. Supplied: 3.0 years, which supports at most 2 trials. Closed-form bound: 41.4 years.
Haircut Sharpe ratio
0.00
Bonferroni adjustment for 25 tests, a 100% haircut. The strictest standard adjustment.
Probability of backtest overfitting
not computed
Requires the full set of trial series, which was not supplied.
By calendar year
Period
Return
Volatility
Sharpe
Drawdown
PSR
2023
−9.1%
13.1%
−0.67
−22.8%
25.0%
2024
14.6%
12.8%
1.14
−10.7%
87.8%
2025
9.6%
13.1%
0.76
−13.5%
76.8%
Before and after live trading began, 01.01.2025
The live period is the part of a record no backtest search could have shaped. Each part is measured as a single trial.
Period
Return
Volatility
Sharpe
Drawdown
PSR
Before live trading
2.1%
12.9%
0.23
−22.8%
62.8%
Live trading
9.6%
13.1%
0.76
−13.5%
76.8%
What this report does not cover
It does not verify accounts, custody, fees or the existence of the capital. It does not re-mark positions against independent prices, because the basis is returns as given. It does not assess the client, any manager or any instrument, and it contains no recommendation to allocate to, buy, sell or hold anything. A figure computed from a declared trial count is only as reliable as the declaration.
Method
Probabilistic and deflated Sharpe ratios after Bailey and López de Prado (2012, 2014). Minimum backtest length after Bailey, Borwein, López de Prado and Zhu (2014). Haircut Sharpe after Harvey and Liu (2015), Bonferroni bound. Probability of backtest overfitting by combinatorially symmetric cross-validation after Bailey, Borwein, López de Prado and Zhu (2017). Implementation kernel@1, tested against the published worked examples.