Diagnosing Signal Performance Across Rebalancing Horizons

Summary

The document describes a long–short factor backtest that ranks assets into top and bottom quintiles. The signal has positive performance at monthly, quarterly, and annual rebalancing frequencies, with returns improving as rebalancing becomes less frequent. The author wonders whether the monthly results indicate excessive signal volatility and asks what diagnostics should precede changes to the signal.

The author reports that attribution finds little exposure to most Barra factors, but provides no performance statistics, transaction-cost estimates, or further tests. The pattern alone cannot establish that monthly signal volatility is the cause: slower rebalancing can also change turnover, trading costs, exposures, and the sample’s realized returns. The document is a research question, not a demonstrated method or conclusion, and it leaves signal-smoothing and validation approaches open.

Key ideas

• The backtest reports positive long–short quintile performance at multiple rebalance intervals.
• Performance rises as the portfolio is rebalanced less often in the reported results.
• The author suspects monthly signal volatility but does not establish it as the cause.
• Factor attribution shows little exposure to most Barra factors, while other diagnostics remain unspecified.

Tags

# Signal Quintile Performance


# Signal Quintile Performance












I’m currently backtesting the performance of a signal (top - bottom quintile). The signal shows positive performance over monthly, quarterly, and annual rebalance horizons. However, the performance strictly increases as I rebalance less frequently.

I’m assuming this indicates that my signal is too volatile on a monthly basis. How can I reduce volatility on a monthly basis? Are there other tests I should be conducting on top of performance before making changes to my signal?

For context, I’ve already run attribution on the long/short portfolio (top quintile - bottom quintile), and it has little to no exposure to most Barra factors.

Any insights or suggestions would be greatly appreciated!

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

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