Improving Volatility Forecasts with Mean Reversion Adjustments
Summary
The post investigates why strong risk-adjusted trading forecasts can have weaker subsequent outcomes than a linear relationship would imply. Since forecasts divide expected return by recent volatility, a strong signal can reflect unusually low volatility as well as a large expected return. If volatility then rises, realized risk-adjusted returns are reduced even when return forecasts are otherwise accurate. The author tests this explanation by grouping observations by forecast strength and comparing future volatility with the volatility forecast available at the time.
A second analysis conditions forecast errors on current volatility relative to a slow historical average. Its results indicate mean reversion: forecasts tend to be too low in quiet regimes and too high in volatile ones. The proposed response is to blend current and slow-moving volatility measures, then use this revised estimate in risk adjustment. The post reports that this reduces, but does not eliminate, forecast-conditioned volatility errors. It uses historical, pooled observations and illustrative plots; the supplied text is truncated before the final evaluation, so the improvement’s full effect and robustness cannot be determined.
Key ideas
- Strong risk-adjusted forecasts may partly arise from unusually low recent volatility rather than larger expected returns.
- When volatility rises after a quiet period, realized risk-adjusted returns can disappoint despite accurate return forecasts.
- Conditioning future-to-forecast volatility ratios on current volatility provides evidence of mean reversion.
- Blending current volatility with a slow historical measure reduces the reported forecast bias but leaves residual error.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.