PureBytes Links
Trading Reference Links
|
> However, I've noticed that trailing stops get triggered even when
> the position is a loser. Consequently, assuming they're accurate,
> they may not produce the intended result without modification.
The big danger with trailing stops (as I understand it anyway) is
when a significant portion of your profits (or losses?) come from a
single bar. More precisely, you have to be careful if the Maximum
Favorable Excursion and the trailing stop are hit in the same bar.
In that case, TS doesn't know what happened inside the bar, and it
"assumes" the best possible case.
For example, let's say you have a 2pt stop, and the bar has an OHLC
of 106/110/100/104. Since the open is closer to the High than the
Low, TS assumes the price went smoothly from 106 to 110 to 100 to
104. It reports you got stopped out at 108 with a trailing stop.
However, in reality the market might have gone from 106 to 100, and
you'd have gotten stopped out at 104.
If this trade had been open for some time, and 4pts was a small
percentage of your open equity, the difference between 104 and 108
wouldn't make a huge difference to your reported results. E.g. if
you went long at 50, 4pts is a reasonably small amount.
But what if you went long at 106? Then 104 vs. 108 makes a big
difference.
This behavior can be exploited to create bogus results. If your
system has a lot of 0- or 1-bar trades, you can't trust the results.
Basically you're using this TS behavior to "pretend" you could have
squeezed the best possible results out of each bar's volatility. But
in reality you almost certainly couldn't have traded it that way.
If you're going to use trailing stops, make sure the trail amount is
larger than the typical bar. That way it will usually take several
bars to go from the MFE price back to the stop, and you can be sure
that the price really DID move like that.
Gary
|