V.10:4 (173-177): Measuring An Indicator's Forecasting Ability by George R. Arrington

Measuring An Indicator's Forecasting Ability by George R. Arrington

Indicators are used to forecast a market. Traders have their favorite indicators. How can a trader determine whether an indicator leads the market? How can we tell whether one indicator is superior to another? Arrington explains how to measure a technical indicator's ability to forecast prices.

Many traders use a variety of technical indicators to forecast changes in the prices of stocks and commodities, including such diverse indicators as stochastics, relative strength, on-balance volume, sentiment index, weather patterns, candlestick patterns, the Arms index and several pattern recognition techniques such as wave and cycle theories. In addition, many traders have one or two indicators of their own devising. A good indicator leads changes, and its signal is helpful when forming a trading strategy. Different technical indicators are useful at different times and in different markets.

With all the varied technical indicators we have available, how do we evaluate whether and how much an indicator leads the market? How do we compare one indicator to another? And how do we determine how much confidence we can put into a particular indicator? A personal computer and spreadsheet software with the ability to run simple regressions can help us find out.

Regression analysis is a powerful statistical technique that shows the relationship between two variables (one dependent, one independent). It can be used to assess the forecasting ability of a particular indicator (the independent variable) by measuring the strength of the relationship between the indicator and subsequent values in the variable being forecast (the dependent variable). Regression analysis can tell us whether the independent variable leads the dependent variable; if so, the length of the lead; and how much confidence we can have in the relationship.