# On-Line Portfolio Selection with Moving Average Reversion

**Authors:** Bin Li, Steven C. H. Hoi  
**arXiv:** [1206.4626](https://arxiv.org/abs/1206.4626)  
**Date:** 2012-06-18  
**Summary page:** https://www.summarizepaper.com/en/paper/1206.4626v1/

## Key points

- The article proposes a new on-line portfolio selection strategy named "On-Line Moving Average Reversion" (OLMAR) that exploits the Moving Average Reversion (MAR) phenomenon by applying powerful online learning techniques.
- Stock prices tend to follow the mean reversion phenomenon, where high and low prices are temporary and likely to revert to their average over time.
- Existing mean reversion strategies often make the single-period mean reversion assumption, which is not always satisfied in some real datasets, leading to poor performance when the assumption does not hold.
- To overcome this limitation, the authors propose a multiple-period mean reversion approach using MAR and apply it to OLMAR.
- From their empirical results, they found that OLMAR can achieve significantly better results than existing algorithms, especially on datasets where existing algorithms failed.
- OLMAR also runs extremely fast, making it practical for a wide range of applications.
- The proposed strategy provides a promising solution for investors looking to maximize their returns while minimizing risks in an ever-changing market environment.

## Summary

The article "On-Line Portfolio Selection with Moving Average Reversion" by Bin Li and Steven C. H. Hoi proposes a new on-line portfolio selection strategy named "On-Line Moving Average Reversion" (OLMAR) that exploits the Moving Average Reversion (MAR) phenomenon by applying powerful online learning techniques. The authors note that stock prices tend to follow the mean reversion phenomenon, where high and low prices are temporary and likely to revert to their average over time. While existing mean reversion strategies have shown good empirical performance on many real datasets, they often make the single-period mean reversion assumption, which is not always satisfied in some real datasets, leading to poor performance when the assumption does not hold. To overcome this limitation, the authors propose a multiple-period mean reversion approach using MAR and apply it to OLMAR. From their empirical results, they found that OLMAR can overcome the drawback of existing mean reversion algorithms and achieve significantly better results, especially on datasets where existing algorithms failed. In addition to superior trading performance, OLMAR also runs extremely fast, making it practical for a wide range of applications. The article highlights the increasing interest in machine learning and AI communities towards on-line portfolio selection. The proposed strategy provides a promising solution for investors looking to maximize their returns while minimizing risks in an ever-changing market environment. Overall, this research contributes to advancing our understanding of financial markets and developing effective investment strategies using machine learning techniques.