Evidence triangulation across leaderboards
strat.ninja's backtest ranking and freqst's reported results, side by side — see where independent evidence sources converge or diverge before trusting a headline rank.
AI · strat.ninja · freqst
Quantitative due diligence for ranked crypto strategies.
Inspect Freqtrade strategies through public rankings, source-code risk analysis and research reports with stated test windows. Static audits describe code; backtests describe historical tests. Neither is live validation or a promise of returns. Check each report's dates, sources and limitations before drawing conclusions.
Decision framework
Evidence before confidence in a rank
AI static audits of strategy source
Entry and exit logic, risk controls, and embedded assumptions are read directly from available source code — observable strengths and weaknesses, explained in plain language.
A risk audit that flags the traps
Overfit parameter clusters, disabled stop-losses, look-ahead bias, and win-rate traps — the failure modes that make a backtest look better than live trading — flagged per strategy.
Monthly research engine
Not a return leaderboard. A layered filter for fragile strategies.
Candidates are frozen before backtesting to avoid picking winners after the fact. Each month, only ten AI-selected candidates enter the expensive empirical stage, reducing compute while preserving an auditable selection path.
Source-quality prefilter
Entry logic, exit strategy, risk management, technical indicators, backtesting robustness, market adaptability, code quality, innovation edge, practical implementation, and risk-adjusted-return analysis.
AI monthly basket selection
Enforces category and timeframe diversity inside a high-quality clean pool; indicator and risk metadata are considered only when available, while month-seeded tie breaks and a recent-selection penalty promote rotation.
Cost-adjusted empirical validation
Fixed multi-asset, multi-year windows with fees and fee-equivalent slippage; measures Sharpe, profit factor, maximum drawdown, sample depth, and cross-asset breadth.