Online Mean-Reversion Methods for Multi-Asset Portfolios
Summary
The article surveys four online portfolio selection methods that seek to profit from mean reversion: Passive Aggressive Mean Reversion (PAMR), Confidence Weighted Mean Reversion (CWMR), Online Moving Average Reversion (OLMAR), and Robust Median Reversion (RMR). PAMR adjusts how actively it reallocates based on a threshold; CWMR incorporates covariance information; OLMAR predicts reversals from moving averages; and RMR uses a robust L1 median to reduce the influence of outliers.
Reported comparisons use six historical equity and index datasets and explore parameter sensitivity with optimization trials. The reported winners and preferred settings vary across datasets, while some methods show striking returns and others fare poorly. These are historical results, not evidence of live profitability: the article stresses that parameters chosen after examining data may not generalize, and that market structure matters. It provides no transaction-cost or live-trading validation, so the results should be read as exploratory rather than a reliable strategy ranking.
Key ideas
- PAMR uses a threshold to control when it keeps or changes portfolio weights.
- CWMR incorporates second-order covariance information when setting weights.
- OLMAR predicts reversals using moving averages, while RMR uses a robust L1 median.
- Historical performance and favorable parameter settings differ across the six datasets.
- Hyperparameter choices and market structure limit how well these results may transfer to live trading.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.