Backtesting I Generated 1,500 Trading Strategies. 114 "Passed" on Bitcoin. None Were Real. A hands-on demonstration of why in-sample / out-of-sample / holdout testing isn't enough to catch overfitting — and the statistics that actually are.
