|
PureBytes Links
Trading Reference Links
|
Mike,
Yes, I have played around with this backtesting
resolution quite a bit. You said it exactly right. It
does help eliminate much of the bouncing ticks
problem. I almost always set it to 1 minute when
testing any system and generally provides very
accurate results. Although, recently I did discover a
strange occurance on a system running on a 2 min chart
and when I set it to 1 min resolution, it made a
couple of inaccurate asussumptions of the way the tick
moved inside the bar and ended up showing a profit
target hit twice on the same bar. Kind of hard to
explain in detail, but somehow TS got confused on the
1 min resolution setting.
Hope that helps some.
Sincerely,
Todd Hoff
<Forwarded Message
Date: Sun, 08 Dec 2002 11:26:17 -0800
From: "Mike Gossland" <mga@xxxxxxxx>
To: omega-list@xxxxxxxxxx
Subject: Backtesting resolution
Plain Text Attachment [ Save to my Yahoo! Briefcase |
Download File ]
Has anyone had experience playing with the backtesting
resolution in
either TS2000i or TS 6?
It's a property of a strategy. TS documentation of the
feature is
pretty sparse,and I haven't gotten down to trying it
out myself yet. I am
guessing that this offers a decent solution to the
"bouncing ticks"
problem, where TS can't know the order of ticks within
a bar to determine
where a % risk trailing stop would have been hit, or
whether a stop or
limit is hit first within a bar. Seems to me if you
set the backtest
resolution to 5 minutes, on a daily bar system, TS
should be able to
generate a much more accurate history of what would
have happened.
Is this an accurate description? If so, does TS
internally deliver the
higher bar interval in the smaller resolution units ?
Any thoughts or
comments?
TIA,
Mike Gossland
|