Monitor Strategy Quality Through Expectancy, Frequency, and Risk

Summary

This article explains how win rate, payoff ratio, trading frequency, and position risk work together to determine a strategy’s realized results. It frames trade quality as net expectancy after fees, slippage, and funding, and recommends measuring outcomes in units of planned risk (R). It also distinguishes signal, order, and completed-trade frequency, and warns that apparently separate positions may share the same market exposure.

The practical framework calls for recording entry and exit details, planned risk, net profit and loss, costs, and execution slippage, then tracking both recent and longer rolling samples. A simple breakout example illustrates how to inspect the four dimensions; it is a measurement exercise rather than proof of an edge. The article also proposes separate responses to statistical deterioration and execution failures, and recommends checking cost sensitivity, parameter robustness, and live reproducibility. Its numerical examples are illustrative, and the suggested thresholds are not universal. Averages and historical samples can obscure tails, dependence, and changing market conditions, so the metrics support monitoring rather than guarantee future performance.

Key ideas

• Net expectancy combines win probability, average win and loss, and trading costs.
• Risk-normalized R multiples make outcomes comparable across trades with different planned losses.
• Signal frequency, order activity, and completed trades differ, and more activity can add cost without adding independent opportunities.
• Position size should follow a risk budget, with limits for single trades, correlated exposures, and the full portfolio.
• Track rolling performance and execution separately, and distinguish strategy deterioration from system faults.
• Backtests should test realistic costs, parameter sensitivity, and whether measured behavior can recur in live trading.

Tags

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.