Balancing Signal Predictiveness, Stability, and Trading Costs
Summary
The document explains why a trading signal should be judged by more than its correlation with near-term returns. A highly predictive but jumpy forecast can trigger frequent trades, raising transaction costs and turnover. Smoother, more autocorrelated signals may be more practical even if their immediate predictive power is slightly lower. Exponential moving averages are offered as one way to smooth signals that naturally fluctuate quickly.
Signal evaluation should also examine how predictiveness changes across future holding horizons. A signal whose predictive power fades slowly allows more time to enter and exit efficiently; one that decays rapidly may require speed and leave the trader holding positions with little expected return while waiting for a cost-effective exit. The piece recommends plotting signal behavior, keeping analysis simple, recognizing that components interact, and using simulation to explore those interactions. It offers conceptual guidance rather than specific empirical results or a universal formula for selecting signals; the best trade-off depends on the trading strategy and its costs.
Key ideas
- A signal's predictive power must be weighed against its volatility and the turnover it creates.
- Smoother, more autocorrelated signals may be preferable when trading costs make frequent turnover expensive.
- Evaluate forecast relationships across multiple future horizons, not only the horizon used for immediate decisions.
- Slowly decaying predictiveness can reduce the need to enter trades quickly.
- Plots and simulation can help reveal how signal quality, holding periods, and costs interact.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.