AlgoScoreAlgoScoreAlgoScore analyzes six deeper dimensions behind every strategy and compresses them into one score you can understand instantly.
From raw data to final score in four transparent steps.
Daily returns and account statistics pulled from verified Myfxbook track records via API every day at 01:00 UTC.
Six risk-adjusted metrics computed across two time windows: all-time and recent (365 days).
Ljung-Box significance testing on autocorrelation. Penalties applied only when statistically confirmed (p < 0.05).
Bayesian-inspired shrinkage toward neutral for short records. Longer proven history rewarded with diminishing returns.
Each metric targets a different weakness that simple returns can't reveal.
Measures return relative to downside deviation only. Unlike the Sharpe ratio, upside volatility is not penalized, making it ideal for trend-following or momentum strategies.
Annualized return divided by maximum drawdown. Answers a direct question: how much return does this strategy generate per unit of worst-case loss?
Captures both the depth and duration of drawdowns from the equity peak. A strategy that stays underwater for months scores worse than one that recovers quickly.
Detects left-tail fat in the return distribution. Strategies with occasional large losses hidden behind steady small gains will show elevated negative skewness.
Tests for serial dependency in weekly returns using the Ljung-Box statistic. High autocorrelation can indicate grid trading, martingale layering, or artificial equity smoothing.
Measures persistent floating losses relative to account balance. Catches strategies that hold losing positions open for extended periods to avoid realizing losses on the equity curve.
A single score derived from four weighted components.
AlgoScore ranges indicate the overall quality of a strategy.
