See important Research Disclaimer.
Using correlation in trading
Correlation is a form of analysis that looks to examine and identify whether two variables have any relationship, and if so, how strong that relationship is. In trading, correlation is used to identify assets that may move in a particular fashion, with any change in the correlation providing a possible trading opportunity, based on the idea that the change will be followed by a reversion to the mean. This is known as ‘pairs trading’.
Negative correlation vs positive correlation
A perfect negative correlation exists when two assets move in opposite directions – i.e. when asset A moves up, asset B moves down. By contrast, a positive correlation exists when the two assets move in unison. Completely random movements have no correlation.
In reality, assets are usually neither perfectly negatively nor positively correlated. Instead, traders can look for those assets that have a high degree of correlation, and trade them when price action moves outside of a given statistical boundary. A strong correlation is one of 0.8 or above, and those below 0.5 are weak.
Look for a strong correlation to spot trading opportunities
Merely picking two random instruments to trade would not be correlation trading. Any brief correlation would simply be a move away from the statistical norm, and would not provide a trading edge over time. Their movements would not be predictable with any degree of accuracy, offering no insight for the trader. Instead, those with a strong correlation can provide interesting opportunities, since there will be times when the correlation weakens and provide a possible trade opportunity.
Trade all markets and asset classes
Correlations can be found within many markets, or across various asset classes. A common example would be the relationship between gold and the USD. When the dollar rises, the price of gold tends to fall, since gold is priced in dollars, and a higher price makes gold more expensive, decreasing demand. Two companies in the same sector can be closely correlated too, for example BP and Shell in the oil sector, or two supermarkets.
How to find correlations
Finding correlations requires a tool that utilises price data to evaluate correlation values, which will require price data. It is important to note that correlations can break down over time, or go against the expected movement for very long periods, so risk management is still crucial – whichever system is used requires a risk management approach that ensures only a small amount of your trading capital is risked on each position.