Drawdown is an experience
Maximum drawdown measures the fall from an equity peak to a subsequent trough. It does not tell you how long recovery takes. Consider depth, duration, and the concentration of losses. Two strategies with the same total return can create very different experiences along the way.
Exposure changes the whole problem
Lower exposure often lowers variability, but also changes capital use and trading costs. Leverage can magnify losses and liquidity stress. The public lab keeps exposure between cash and a fully invested long position so that the basic relationships remain easy to inspect.
Put friction into the rule
Fees, spreads, and slippage affect results. Our example subtracts a one-way basis-point cost from each absolute change in exposure. Real execution also requires capacity, latency, minimum order sizes, and market impact. A simplified cost model should not be mistaken for a real execution estimate.
Define a review trigger in advance
Before looking at performance, record the conditions that require a review: missing data, costs above the budget, or behavior inconsistent with the hypothesis. A predefined process reduces the temptation to change the standard after losses appear. The public examples support learning and method development.