Back testing is a form of scenario analysis in which historical data are input into a financial forecasting model.
The idea is to identify - according to the model being used for the back testing - what the future outcomes would be if the selected historical scenarios were to recur in the future.
Another use of back testing is to test the validity and accuracy of the forecasting model itself, in cases where actual historical data is available in relation to the output of the model.
In this form of back testing historical data are input into the forecasting model.
The outputs from the model being back tested are then compared with the related actual data for the variable being forecast by the model.
Any differences identified may be the result of errors or limitations in the forecasting model.