As a result of the deep thinking that was necessary for the 10-in-a-row competition, I have been examining a number of trading approaches that I used to use in the 1990s for long-term trading. I have converted some of these into general models of certain types of market more and have loaded this with masses of stop loss and limit exit tests. The results have been remarkable - for instance, a pair of systems with a combined Sharpe Ratio of 19.5.
Apparently the worst day in a three month period would have been -45 pts, while the average profit is 71 pts per day. This seems ludicrous on first view. For instance, if we are willing to lose, say, £1,000 on a losing day, then the average day would be +£1,600 and more than 9 out of 10 days would be profitable.
I am setting a tentative roll out for this in the first week of Jan 2011 with last testing throught the Xmas period - but what are the implications for the very good (but not ludicrously good) current systems? On first go, it would seem that the newest ideas completely dominate the existing systems. And given that they have a number of other features that are attractive, it might be that we switch a considerable weighting to the newer stuff
As always, much work is required to assess if the profit is real or has some hidden problem lurking in its actual workings.