|
PureBytes Links
Trading Reference Links
|
> When you are finding a workable system idea you want to try lots of
> ideas quickly. You want perfectly accurate data that covers many
> contracts so that you can quickly see how the system idea you are
> testing has worked over time.
To quote from your excellent paper Bob,
"All of these factors (caused by Continuous Adjustments) can affect the
performance of a trading system and could make the results of backtesting
differ from trading a single contract. For a long-term system, one that
stayed in the market for many months, using such data would be very
convenient since it would eliminate the need to ever worry about expirations
in backtesting. For a medium-term system, say one that stayed in the market
for many days to weeks, such distortions would be insignificant. For
short-term trading, say under a few days, they become important. For a
day-trading system they could be serious sources of error."
It looks like, combined with my response to Gary, that using continuously
adjusted data must be very carefully considered with the kind of system you
are developing, no matter what stage you are in. My comments here were
merely trying to point out that indeed, this kind of historical data is not
the "universal" answer, and that we must be mindful of not inadvertantly
replacing one error with a larger one.
> If you hold positions overnight, you get gaps and your system results
> include the effects of the gaps. But if you use un-adjusted
> contracts, the resulting gaps at rollover can make your backtesting
> either better or worse than what will happen in real trading and will
> distort the results.
So can different historic data sources... I have personally seen results on
systems using different sources of ND data over just one contract length
vary more widely than the effect of hard spliced data. Obviously, this is
dependent on the system. Again, if a system is too sensitive to gaps to
being with, then it better be changed, no matter what it's tested on.
>If you run an optimization with such gaps, some
> trades might span the gaps at one parameter value and not with a
> small change in parameter value. This can cause you to select
> non-optimum parameters.
I worry that this approach seems to suspend for the moment all the perils of
over-optimization, and might focus too tightly on these specific 4 price
moves in one year.
>
> Back-adjusting contracts has been around for a long time and for
> good reasons. <snip> If they weren't necessary, people wouldn't be doing
it.
>
Sorry, I'm not going there... too much like the Dodo's in the movie "Ice
Age"... ;~)
Thanks for the chance to debate these ideas...
Best regards,
Gene Pope
|