- concetto·7 min di lettura
Does the best backtest predict the future? We measured it 200 times
When you optimize a strategy and ship the best-Sharpe settings, you're betting the in-sample ranking of your parameters holds up out-of-sample. Across 200 strategy-coin grids, that ranking correlated just 0.37 with reality — and a quarter of the time it was no better than random.
Leggi →
- approfondimento·8 min di lettura
Which crypto strategy families survive out-of-sample? All 10, ranked
We tuned ten classic strategy families on twenty coins, then ran each tuned winner once on data it had never seen. Only two families kept a positive median edge. Here is the full ranking — including the three that were negative before fees even bit.
Leggi →
- approfondimento·10 min di lettura
We ran 11,440 backtests. The average 'winning' strategy lost half its edge on data it had never seen
Ten classic strategy families, 5,720 parameter configurations, twenty crypto pairs, one honest rule: tune only on the past, then test once on data the strategy never saw. Here is exactly how much of 'backtested profitability' survived.
Leggi →
- concetto·9 min di lettura
What is walk-forward optimization (and why TradingView strategy testers lie to you)
Backtest equity curves look beautiful right up until the moment you trade them live. Walk-forward optimization is the cure — here's how it actually works.
Leggi →