Of course, I’ll propose trading strategies based on the findings. In this post, I’ll explore mean reversion of individual time series and in the next post will investigate constructing artificial mean reverting time series. Note that stationarity does not imply a range-bound price series with variance independent of time, rather that the variance simply increases more slowly than that of a normal diffusion process (yes, I know there are more formal and robust definitions of stationarity, but since we are only interested in the practical application of the concept to trading, this will do for now and you can stop cringing). In this context, I will refer to data that tends to mean revert as stationary. Mean reversion also exists in, or can be constructed from, financial time series data. Performance can be thought of as being randomly distributed around a mean, so exceptionally good performance one year (resulting in the appearance on the cover of Sports Illustrated) is likely to be followed by performances that are closer to the average. My earlier posts about accounting for randomness ( here and here) were inspired by the first chapter of Algorithmic Trading. Ernie works in MATLAB, but I’ll be using R and Zorro.Įrnie cites Daniel Kahneman’s interesting example of mean reversion in the world around us: the Sports Illustrated jinx, namely that “an athlete whose picture appears on the cover of the magazine is doomed to perform poorly the following season” (Kahneman, 2011). I’m a big fan of Ernie’s work and have used his material as inspiration for a great deal of my own research. The book follows Ernie’s first contribution, Quantitative Trading, and focuses on testing and implementing a number of strategies that exploit measurable market inefficiencies. This series of posts is inspired by several chapters from Ernie Chan’s highly recommended book Algorithmic Trading.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |