|
PureBytes Links
Trading Reference Links
|
In a message dated 98-07-12 09:23:22 EDT, K.Morimoto@xxxxxxxxx writes:
> say I train the net using the 10 year bond futures data between 1990 ~
> 1995. Then if I use that trained net on 3 year bond futures data for the
> same dates 1990 ~ 1995, would it be a proper test for the net? The
> markets will be highly correlated, but the data points would not be
> exactly the same. Its just that for me to train the net on 1 market (10
> year bond futures), I just don't have enough of test data.
The more correlated the data, the less valid the test. What you should do is
train on the 1990 - 1995 data, and then test on data from 1996 to present.
However no matter how good that may look, if the market changes character in
the future, you still have garbage. I wish you luck.
Pete
|