|
PureBytes Links
Trading Reference Links
|
Mark Brown wrote:
>
> JOHN, THEY SAY A PICTURE IS WORTH....http://www.markbrown.com/bug.htm
>
> Here you will find the biggest bug that Omega has! It has cost many, many
> people their hard earned money. This feature has no business being in TS at
> all! Some here will argue its the users responsibility to know about the
> faults of these settings.
Mr. Brown, you are providing valuable information and warnings in this
post that people should utilize as they develop/test systems. As a
system and trading newbie, I learn from many of the posts on this list.
The question is, what should TS do if presented with a system and stop
settings that make it exit with a small profit that might or might not
be actually tradeable? Should it not permit that? If you are
entering on the close and would in fact move your stop during the next
day, and assuming you very rarely get stopped out at your stoploss type
exit, I believe this would give you accurate results *to the extent you
could enter the orders on the floor with enough cancel/replaces to trail
your stop by 10% after the contract moved up $200* (the first ex. at
your website). Here, of course is the problem. You cant' realistically
do that, at least not in the big s+p, unless you are actually in the
pit (If it were corn, maybe you could put a stop in to preserve 4c of
profit). Clearly, no-one could do several of the other examples like
the $1 trailing stop. But where should TS draw the line? Conceivably a
pit trader is at the keyboard after hours, testing some strategy that he
*could* do. It would be the same problem testing some very active
strategy with 1 min. bars. Logically the system may work but can you
implement it realtime? Depends on your order placement circumstances.
It would appear this is the user's responsibility not TradeStation's,
because only he/she knows.
Perhaps TS could come with a diagnostic that issues warnings like, "Your
profit target is only 5% of the average bar's range". Or, "Warning: 10%
of trades exited on same bar based on assumed OHLC sequence". The
latter one would be a warning that either your stop is too tight, or you
need more detailed (intraday) data. In fact, here's a development
suggestion: develop "expert commentary" based on the system stats. It
could comment on "sources of error" and possibly even other related
commentary. Also, it seems it would be helpful to include the system
parameters, including TS stop settings on the results summary display.
Also, the literature/video that accompanies TS might point some of these
topics out (and they may, I don't remember).
This one seems more like a potential pitfall than a bug. Thanks for
pointing it out, though, to those of us less experienced.
Conrad Bowers
|