|
PureBytes Links
Trading Reference Links
|
This is an interesting thread!
After some thought, I'm in the camp that says, "why throw those big
winning/losing trades out?" I'm not sure I understand the reasoning that
arbitrarily says certain trades are not valid simply because of the size of the
win/loss. As pointed out by another list member, trend following systems, by
design, will try to catch and stay w/ the big moves. That's what LeBeau's
Chandelier exit is about.
What may be most relevent in determing the worth of a system (related to large
trades) is comparing the distribution of prices against the distribution of
trading profits/losses. If the price series itself contains lots of large moves,
then (I'm just throwing this out there) a system that is able to capture some
percentage of them should be valid. The fact that the p/l statement shows a few
large gains/losses is just reflecting the action of the market.
If, on the other hand, the price series doesn't contain "lots" (dont ask me) of
large moves, then a system that relied on catching a few of those moves may be
less reliable in real time.
So, I wouldn't throw the system out just because there were some large trades.
I'd look at the price series to see if these large moves could be expected.
Just some random thoughts.
Regards,
Bill Vedder
Chris Edwall wrote:
> > It is a good practice, when analyzing trading systems, to subtract out the
> > top 5% & the bottom 5% of the all trades. This will help you determine if
> > there is real consistency in the system. The truth will be known then.
> > There also was a article in Omega Magazine a few months back that outlined
> > some great thoughts to keep in mind when building or testing trading
> > systems. Sorry can't recall the issue. Anyone know?
>
> The new TS2000i system report (using Rina Systems methodology) excludes
> "outlier" trades that are greater than 3 standard deviations from the mean,
> I believe. I'd say this new system report is one of the better features of
> TS2000.
|