|
PureBytes Links
Trading Reference Links
|
First, let me thank those who have offered suggestions
on this issue. Bob, I am going to respond to your
post first and if you don't mind post it to the whole
list as well. I will post a summary of possible
reasons that have been offered soon.
--- Bob Fulks <bfulks@xxxxxxxxxxxx> wrote:
> We would need more information to guess what the
> problem might be.
>
> > Are you entering on a stop, on close of a bar,
> or at market?
>
As I stated in my initial e-mail, both the system and
corresponding indicator (functions which incidentally
do not know anything about stops) showed "incorrect"
results.
> > How many bars in a holding period?
10-100
>
> > Are you exiting on a stop, on close of a bar, or
> at market?
>
> Entering on a stop will result in different behavior
> "within-the-bar" since in backtesting, TradeStation
> assumes certain behavior with the bar. You say
> "Becktesting resolution is unchecked". That means it
> does not test within the bar. Have you tried
> checking this?
No. But please remember, this is not a difference
between backtested performance vs. actual performance.
It is a difference between realtime chart at time x
and the same realtime chart at time x+delta after
closing tradestation and restarting it with the same
exact chart.
>
> If the number of bars in the holding period is less
> than about 10, you are likely to get significant
> errors because of the differences caused by the
> different behavior within-the-bar.
Why? Why should the mathematical indicators be one
scintilla different after I restart Tradestation?
>
> If you happen to be entering and exiting on stops on
> the same bar, the results will be totally bogus
> since the behavior within-the-bar will certainly be
> different that that assumed by TradeStation.
Right, this issue is well known. Though I doubt it
could be relevant, thanks again for reminding me of
it.
Jack
|