Backtesting myth-buster: ten winning trades prove nothing. One hundred trades across regimes prove a little more. Sample size matters. #trading NFA