|
PureBytes Links
Trading Reference Links
|
Funny as I was just comtemplating this very issue last night... i.e. how to
set a "stop price" entry on data1 using a stop price entry on data2 (or
whatever).
I've been thinking through two issues:
1) For backtesting purposes, how does TS "know" what your fill price is on a
stop when there is divergence between cash and futures price?
2) For entry, perhaps one could calculate a bar by bar "spread ratio"
between the two and create the stop price off the cash index times the ratio
of the previous bar... should be more accurate than simply "at close".
Especially since, the shorter the interval, the less likely with Oddball
that your results will match a longer compression. This, I believe, is due
to the "missing intervals", which, while taken care of by Bob's code, can
still result in a different result based on "rejected trades".
This entry method may take care of the backtest issue as well.
Thoughts?
Best regards,
Gene Pope
----- Original Message -----
From: "Bob Fulks" <bfulks@xxxxxxxxxxxx>
To: "David Folster" <mr_bond@xxxxxxxxx>
Cc: "BobR" <bobrabcd@xxxxxxxxxxxxx>; <omega-list@xxxxxxxxxx>
Sent: Saturday, March 23, 2002 11:26 AM
Subject: Re: Oddball modification
> At 10:22 PM -0800 3/22/02, David Folster wrote:
>
> >When I tested various versions of OB entering on a stop, with a view to
> >waiting until price was moving in the correct direction, I got great
results
> >on the cash index, but my results on futures were *very* poor. And that
was
> >not even taking into account any of the added real world costs. Maybe
your
> >results would differ from this.
> >
> >Just my $.02.
>
> I am sure you are correct, as we have been saying:
>
> >Remember that you cannot trade the $SPX and the results will be
> >poorer on futures or SPY.
> >
> >I take this to show that waiting until the price is moving in
> >the right direction before entry works better (not surprising).
>
> The significance of these tests is "that they show that waiting until
> the price is moving in the right direction before entry works better
> (not surprising)".
>
> The question of how to accomplish that trading futures is more
> complicated.
>
> One way we might try is to use $SPX as data3 and translate the system
> to 5-minute (or 1-minute. etc.) bars, still keeping trades on the
> hour. Then, we could check to see when the $SPX 5-minute bars cross
> the stop levels and use that to trigger a market order on the futures.
>
> Another way would be to filter the futures data some to remove noise
> and look for the filtered value to cross the stop level to trigger a
> market order.
>
> In any case, it is not real simple to turn this "discovery" into a
> trading system...
>
> Bob Fulks
>
>
>
|