Using correlation coefficients and beta
In deploying a dynamic risk management plan, credence to market correlations needs to be considered. Markets often don’t trade in their own little world for very long. If you are trading multiple markets, you probably already have an idea of correlation, but you must remember that correlation can be dynamic. Certain events can increase the correlation of certain markets, like last year when the Federal Reserve telegraphed its second round of quantitative easing. A more recent example occurred on May 5 when the dollar reversed more than 100 ticks (see "Strange bedfellows").
HFTs and algorithmic trade programs use a variety of factors in executing strategies. Most are top secret and can be based on many different factors. After speaking with programmers, I found that often strategies start with the use of "correlation coefficients" and the use of "beta."
A correlation coefficient is a statistical relationship between two or more random variables. As applied to investment vehicles, it implies the price dependence differing assets have on one another. The standard measuring tool runs on a scale from 1.0 to -1.0; 1.0 being perfectly correlated and -1.0 being perfectly negatively correlated.
Obviously, nothing in trading is ever correlated perfectly and you always must remember that correlation does not imply causation. Traders need to buy/sell for all different reasons, thus always creating market imperfections. Correlation coefficients can be run in real time, minute-by-minute or basically any time frame.
Another factor that has joined the main stream of trading is "beta." Beta can be defined as the relationship between percentage moves in one particular commodity as opposed to another. In the example used above, if the dollar falls 1%, beta could be used to measure an expected move in silver. If beta is running at 5, you would anticipate silver to rise 5%.
This is a very crude example. Coefficient/beta models can get very sophisticated. However, a basic understanding is important. Beta can not only be a good tool for determining how much heat you can take on a trade, it also helps understand how much profit is possible. Trade programs and quants are heavily into the use of these and often times they dictate the direction and flow of many markets.
When basing a trade on these factors, an exit strategy contingent on when these variables turn against you also should be considered. Know what your market is correlated to. Know also the corresponding beta. Stops should be used when these correlations dip below the desired level. Beta can tell you where the stop should be placed.