The predictive performances of the developed models
were evaluated using the untouched out-of-sample data
(second period). This is due to the fact that the superior in-
sample performance does not always guarantee the validity
of the forecasting accuracy. One possible approach for
evaluating the forecasting performance is to investigate
whether traditional error measures such as those based on
the RMSE or correlation (CORR) between the actual out-of-
sample returns and their predicted values are small or highly
correlate, respectively.However, there is some evidence in
the literature suggesting that traditional measures of
forecasting performance may not be strongly related to
profits from trading (
Pesaran and Timmermann, 1995
).An
alternative approach is to look at the proportion of time that
the signs of excess stock returns (SIGN) are correctly
predicted.