Where the bodies are buried
Profit = (win rate x average win) - (loss rate x average loss) - costs. To push win rate past 90% you must make losing trades vanish. Markets don't cooperate, so these strategies delay, hide or shift losses until they explode all at once. Nearly every 90%+ win-rate strategy is one of three builds.
Build 1: no stop loss, hold and pray
Price goes against you? Refuse to take the loss. Hold until the market oscillates back to entry, then close flat or slightly green. Markets range 70-80% of the time, so this usually works — the backtest becomes a sea of tiny green bars. Then a real trend or news event arrives, price never comes back, and one forced liquidation erases years of small wins. Picking up pennies in front of a steamroller.
Build 2: martingale
Lose? Double the size and average down. One eventual winner covers the whole losing sequence, so the sequence never records a loss and the win rate prints 98%. Until the statistically inevitable losing streak arrives with position size at 8x, 16x, 32x — and the last trade takes the entire account.
Build 3: scalping with breakeven stops
The sneakiest one, because it looks professional: filter for high-probability micro setups, move the stop to breakeven the moment price ticks in your favour, take 8-15 ticks and run, 50-200 times a day. Most trades end as small wins or breakevens, so the win rate prints 90%+. But every trade pays 2-5 ticks in spread and commission, so the real reward-to-risk is negative — and the first volatile session strings 10-30 consecutive stops together and hands back months of grinding.
What professionals look at instead
Risk-adjusted return, not win rate: Sharpe, recovery factor, drawdown depth. Long-term survivors typically run 40-60% win rates with 2:1+ reward-to-risk — letting winners run and cutting losers fast. The equity curve is less pretty and far more honest. Survive first. Only then talk about consistent gains.