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

Re: Optimization Question



PureBytes Links

Trading Reference Links

For what it's worth, quite some time ago I did an extended experiment with
the realtime version of OmniTrader.

I created a custom profile based on some sytems I had put together in TS,
let it rip on 11,000 bars of backtesting and 1000 bars of forward testing on
a basket of 30 active stocks. It would only reoptimize every subsequent 1000
bars.

Bottom line: no one stock would remain at the top of the "sort list" based
on forward testing Hit Rate or Profit % (or any other forward or backward
measure for that matter) for any stable or predictable amount of time.

JNPR would be good for perhaps 2 weeks, then crash and burn... CIEN would
repeat the pattern for 4 days, then AMCC would blow up immediately, and on
and on. High results in the OOS test had no reliable predictive quality
whatsoever.

Think it's beyond a dead issue.

Best regards,

Gene Pope