LossDays much longer than WinDays

StrataSearch will find what you tell it to, but what should that be? Will the same criteria work across different Sectors? Different time periods? Here we discuss the ins and outs of OneClick Searches.

LossDays much longer than WinDays

Postby rjay » Sun Jan 17, 2010 8:23 am

A couple of OC's I've run recently have all come up with similar results, namely that the figure for LossDays is much higher than for Windays (by a factor of at least 2.5 to 1) on otherwise profitable strategies.

Obviously this would seem to suggest that the longer a trade goes on the less likely it is to be successful, which is something I generally believed to be the case anyway. So what criteria might I add to my filtering or scoring, or to my autosearch to find strategies which are less likely to do this ? Simply using $daysheld seems somewhat of a blunt stick.

Any ideas ?
rjay
 
Posts: 116
Joined: Wed Jul 26, 2006 6:51 am

Re: LossDays much longer than WinDays

Postby Overload » Sun Jan 17, 2010 9:47 am

The first question is whether you want to improve such a trading system, or simply move on to another. To improve it, there are a couple charts in the Detailed Analysis worth looking at. First, the "Trade Returns by Days Held" can be helpful in identifying whether there is a logical cutoff for how long you should hold unprofitable positions. As an example, it's not uncommon to see all the profitable trades bunched up with the shorter number of Days Held, with nothing but losing trades after a certain number of days. In such a case, using $daysheld may make some sense. The probability of trades being profitable after a certain number of days becomes very low for some strategies, so it may make sense to exit at that point. But only additional back testing can tell for sure.

Another good chart in the Detailed Analysis is the "Maximum Adverse Excursion". If you see nothing but red arrows after a certain point moving horizontally across the Drawdown Percent axis, that can help identify a good stop. For example, if you know that no position that's ever had a drawdown of 20% or greater has ever been profitable, that might be a good place to exit. But, again, only additional back testing of the approach will tell you for sure. As well, this adjustment won't have as direct an impact on the number of days held.

Apart from the above (stops and $daysheld), I'm not sure of any other trading rules that will specifically help. But if you're running a OneClick Search, you can certainly add a new filter that will take care of it. On the Strategy Performance tab, add a new Evaluation Field and enter the following:

$C_WinDays - $C_LossDays

Then set your Best Rated Value to 1 and your Worst Rated Value to 0, while rejecting Strategies outside the Worst Rated Value. I would also set the Importance slider all the way to the left so that this is only a filter but doesn't affect scoring. This filter, when used, will bypass any strategies where the average days held for losing trades is greater than the average days held for winning trades.

Hopefully this helps.

Pete
Overload
 
Posts: 2248
Joined: Wed Nov 30, 2005 12:14 pm

Re: LossDays much longer than WinDays

Postby rjay » Sun Jan 17, 2010 12:56 pm

All great suggestions which gives me hours of homework !! Thanks very much.
rjay
 
Posts: 116
Joined: Wed Jul 26, 2006 6:51 am


Return to OneClick Searches and Scoring

cron