Methodology
The audit does not create random fantasy markets. Synthetic worlds are designed to remain close to the original market while introducing controlled path perturbations.
Why single-path backtests are fragile
One historical path can reward rules that depend too heavily on the exact sequence that produced the result.
Synthetic market worlds
Alternative paths are generated to remain statistically close to the original while changing the sequence enough to test fragility.
Statistical validation
Synthetic worlds are validated before being used as audit material, including distribution and temporal-structure checks.
Metrics analyzed
Profit Factor, Expected Payoff, Maximum Drawdown %, Recovery Factor / Recovery Ratio and Trade Count.
Robustness Score
A 0–100 score summarizes path bias, consistency, downside behavior and viability.
Warning labels
Flat distribution, near-identical trade count across worlds, excessive drawdown dispersion and original-result isolation.
Standard audit context
Standard AntiOverfit Audit work is versioned. Each audit belongs to a specific EA version, standard, symbol, timeframe, data window and methodology.
This keeps public results comparable and avoids mixing incompatible test conditions.
Interpretation limits
The objective is not to forecast future profits. The objective is to measure how stable the system remains when the exact historical path is perturbed under controlled conditions.
That is why the output is an audit result, not an investment recommendation.
Synthetic-world acceptance is based on preserving the original market’s statistical structure, including distribution similarity and temporal structure. The purpose is controlled perturbation, not fantasy-market generation.
How to read the method without overcomplicating it
AntiOverfit is not trying to predict the future. It asks whether the EA depends too much on one exact historical path.
What is a synthetic world?
A synthetic world is an alternative plausible path generated from the same market structure. It is not a random fantasy market; it is a stress variation of the tested history.
What is compared?
The EA is compared on the original path and on accepted synthetic paths using metrics such as Profit Factor, Expected Payoff, Drawdown, Recovery and Trade Count.
What makes a result strong?
A stronger result keeps acceptable behavior across alternative paths. A weak result deteriorates when the exact historical sequence changes.
What do warnings mean?
Warnings flag structural patterns that need caution, even when the numerical score is high.
Why not a profit forecast?
The score measures robustness under tested path variation. It does not predict live profitability or guarantee future performance.
Simple reading
High clean score: stronger evidence. High score with warning: caution. Limited or failed result: do not rely on the backtest alone.