[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re[2]: RE Subject:(01) futures contract rollover issues, (02) optimization


  • To: "Gerald Marisch" <gm@xxxxxxxxxxxxx>
  • Subject: Re[2]: RE Subject:(01) futures contract rollover issues, (02) optimization
  • From: Ivo Karindi <ivo@xxxxxxxxx>
  • Date: Wed, 2 Jul 2003 22:08:02 -0700

PureBytes Links

Trading Reference Links

GM> With all the e data available today, its a simple matter to simultaneously
GM> download data for all an issue's contracts daily, throughout the year.  This
GM> obviously maintains the "integrity" and accuracy of each contract's data.

GM> I think this "integrity" is important and would be lost if past data were
GM> "adjusted" for a continuous contract, rendering the data suspect for
GM> accuracy and thus useless.

GM> I'm sure there are purposes for a continuous contract, but I fail to see
GM> them.  Comments?

In my opinion, meddling with the "integrity" of the data IS THE POINT
to avoid curve-fitting your system to that perfectly precise
historical data you have collected.

Here is a (partial) list of ways to meddle with your historical
data to see if your system can swallow it:

1) cut your data into pieces and shuffle the pieces around.
2) Add/subtract a substantial number from the price data
3) Reverse your data, either vertically (in price), horizontally (in
time) or both.
4) Add some random noise to the data.
5) Filter out some noise from your data by using smoothing functions.
6) Use several, drastically different ways to backadjust your data.
etc.

Future will always be different from the past and by meddling with the
historical data there's more certainty that the system will perform in
the future when the future is meddling with your system. ;)

Ivo Karindi