When to retire a setup that used to work

[Shadow Trader]

Edges decay. Market participants adapt, volatility regimes change, and a pattern that worked for a year may stop working. The difficulty is distinguishing a normal drawdown from a genuine decline in edge. I set criteria in advance. For example, if the setup's expectancy over the last fifty trades is below zero after costs, and its drawdown exceeds what my backtest suggested was typical, I move it to a probation status with reduced size. If the next thirty trades do not recover, I retire it. I also check whether anything structural changed, such as a shift in the pair's volatility or a change in my broker's spreads. Having predefined thresholds prevents two errors: abandoning a good setup after a few ordinary losses, and clinging to a broken one out of attachment. I keep a short log for each setup, listing when I started it, its performance by quarter, and notes on market conditions. That history gives perspective. A setup is a tool, not an identity, and replacing a worn tool is part of the job.

[FXS HongKong Scalper]

Writing the retirement criteria before the drawdown is the only way it works. Mine: if expectancy over the last 40 trades falls below zero and the drawdown exceeds the worst one in the backtest by half, the setup goes to quarter size for 20 trades. Only after that does it get dropped or restored.