Quantile Thresholds for Mean-Reversion Spread Trading

Summary

This strategy turns changes in a spread series into long and short entry thresholds. It separates historical spread changes into positive and negative values, then calculates a chosen upper quantile of positive changes and a lower quantile of negative changes. A predicted spread change is compared with these thresholds: crossing the upper boundary signals long, while crossing the lower boundary signals short. The prediction can come from a user-supplied model or a time-series method such as ARIMA or a neural network.

The allocation logic also uses the previous signal and an exit threshold to decide whether to keep a position or return to cash. The document describes the mechanics, but provides no performance results, transaction-cost analysis, or guidance on choosing quantiles and exit levels. Thresholds are fitted from historical spread changes, so their stability depends on the data and period selected. The method also relies on a predicted spread change, whose accuracy and timing are not evaluated here.

Key ideas

• Positive and negative spread changes are modeled separately to set long and short entry thresholds.
• Predicted spread change is compared with those quantile thresholds to generate trade signals.
• An exit threshold and the previous position signal influence whether a trade is held or closed.
• The document gives implementation mechanics but no evidence of profitability or robustness.

Tags

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