Portfolio Volatility Targeting While Preserving Forecast Strength
Summary
This article considers whether a trading system should maintain a fixed expected portfolio risk or allow risk to vary with signal strength. The author's system targets a long run average volatility, while its daily expected risk varies with both aggregate forecast magnitude and a relative correlation factor. Constant risk targeting is common in portfolios where positions are selected through thresholded long and short signals, but applying it to a forecast driven system may discard information in signal strength.
The proposed calculation estimates portfolio risk from position weights and the covariance matrix, then scales positions by the ratio of target risk to estimated risk. The author argues that this removes both correlation effects and relative forecast strength. A compromise multiplies the risk adjustment by aggregate absolute forecasts weighted by instrument weights, aiming to correct correlation changes while retaining stronger exposure when forecasts are stronger. The post illustrates aggregate forecast analysis and account curves but provides no clearly stated final conclusion in the excerpt. Results depend on forecasts, covariance estimates, portfolio construction, and the historical system examined.
Key ideas
- Expected portfolio risk can vary with both aggregate forecast strength and changes in correlations.
- Scaling positions inversely with measured portfolio risk can stabilize ex ante risk while removing variation due to forecast strength.
- A compromise scales by weighted aggregate forecast strength as well as the risk adjustment.
- The article uses covariance based portfolio risk and historical account curve comparisons to examine the alternatives.
- The approach's behavior depends on forecast quality and covariance estimates, and the excerpt does not state a definitive result.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.