I recently went to dinner with several friends who are traders at large investment banks. All of them are smart successful guys with many years of trading experience. I was explaining to them some of the statistical techniques we use at Bigger Capital. None of them were familiar with any of the basic statistical measures used in statistical arbitrage. I’ve encountered the same thing talking to other traders on different occasions. Ninety nine percent don’t have any idea of the basic concepts of statistical trading. And yet none of these measures are new. Statistical Arbitrage has been used as a successful strategy since the 1980’s. So what are the basic measures used? We give brief definitions of the primary ones below. They are the same statistical values we provide subscribers to our SpreadTraderPro. These measures can be used as a standalone strategy, or combined with other methods to enhance returns. They are:
1. Cointegration – measures the degree of confidence that a stock pair which has diverged from its mean value will revert to that mean.
2. Z-Score – measures the distance the spread is from its mean value in standard deviations.
3. Half-Life – is the expected time it will take for the spread to revert half way to its mean.
4. Zero Crossing Rate – is the number of times the spread can be expected to cross the zero value for the defined time period. A higher number implies a shorter holding period and a greater likelihood that the spread will not continue to trend.
5. Sum of Least Square - is the squared distance over the selected period. The lower this number the tighter the spread is, and the better the chance that the price of leg one will not wander far from the price of the leg two.
Written by Norm Winer. Follow me on Twitter and StockTwits.