The Statistical Arbitrage Trading Strategy

The statistical arbitrage trading strategy is an approach to equity trading that uses data mining systems and automated trading, and attempts to make a profit based on a concept similar to pairs trade strategy.

Pairs Trade Strategy

Stocks are known to shift up and down over time and so a pairs trade strategy tries to take advantage of these shifts by predicting the next ones based on what has already happened recently. Two stocks that are in the same market might be considered a pair, because you might assume that they will do similarly over a long period of time, very generally speaking. A trader might notice a divergence between two stocks in a pair, for example, stock A goes down and stock B goes up. Using a pairs trade strategy, that trader might purchase some of stock A and short some of stock B with the assumption that their prices will eventually shift towards each other again. The strategy assumes that a stock that has recently gone down is more likely to go back up, and a stock that has recently gone up is more likely to go down soon.

Statistical Arbitrage

Statistical Arbitrage basically takes that concept and applies it to a portfolio of a hundred stocks or more. The portfolio also applies diversification by taking stocks from different industries and regions to eliminate beta exposure and other risks. The construction of this portfolio is automatic and has two steps to its process. The first step is scoring, where each stock in the portfolio is assigned a numeric ranking of desirability based on how well or poorly it has done recently. Continuing out of the pairs trade concept, stocks that have done well recently are given a number that represents low desirability and stocks that have done poorly recently are ranked as being more desirable. This concept is also known as mean reversion, which is that prices will generally always shift back to their historical average. The second step of portfolio construction is risk reduction, where the desired amounts of each stock are computed based on optimizing the risk of the portfolio. There are other types of statistical arbitrage that incorporate different concepts such as lead/lag, psychological barriers, corporate movement, and also momentum.


One of the risks involved with taking up a statistical arbitrage strategy is the availability of trading time and liquidity. For example, let’s say you have an automatic system that accepts 10 quarters at a time and flips the quarters for you, and every time a quarter lands on heads the machine gives you $1. You decided to use this machine because it is known to flip quarters to heads more often than tails. Therefore, you know that overtime that the machine will generate a profit for you. What happens if you run out quarters before that happens? That is one of the main basic risks of statistical arbitrage, which is that you could have losses over the short term. Having more available funds means having higher chances of profit.

blog comments powered by Disqus