Revision for “Mean Reversion Strategy” created on April 16, 2015 @ 13:12:50 [Autosave]
Mean Reversion Strategy
Numerical Method’s mean reversion strategy is loosely inspired by the following papers.
<li><a class="ext-link" href="http://numericalmethod.com/papers/course1/lecture4.pptx"><span class="icon"></span>Optimal Pairs Trading: A Stochastic Control Approach. Mudchanatongsuk, S., Primbs, J.A., Wong, W. Dept. of Manage. Sci. & Eng., Stanford Univ., Stanford, CA.</a></li>
<li><a class="ext-link" href="http://numericalmethod.com/papers/course1/lecture5.pptx"><span class="icon"></span>Pairs trading. Elliott, van der Hoek, and Malcolm. Quantitative Finance, 2005</a></li>
Very briefly, a good mean reversion strategy depends on its ability to:
<li>estimate the current true/hidden price (the mean) so that we can estimate the disequilibrium</li>
<li>manage the risk by keep the "optimal" position based on the disequilibrium</li>
<h2 id="SyntheticAssets">Synthetic Assets</h2>
In practice, most single assets do not naturally exhibit mean reversion behavior (except maybe over a short horizon). We would thus have to create synthetic assets that mean revert. For example,
One good starting point is cointegration. However, the cointegration criterion, namely stationarity, is a very restrictive concept. It is very difficult to achieve practically. For the sake of making money, this is also unnecessary. What we really need is a concept of "close enough to stationarity so that we can make money".