Friday, 2 December 2011

No more trading innovations . . .

Having spent a couple of days on a new research project, I have had a change of plan. Instead of looking for another market that trades in a way that is uncorrelated with what we currently do, I think it might be better to have no new markets, or other major changes to our current trading for the time being. Too often we have been distracted from the consistent application of a method we have extensively researched, have pursued something else that initially looked good, only to find that it under performed our original systems in the subsequent period. Instead, I am planning to focus on issues of position sizing and estimates of the so-called "Kelly fraction". Also I will be re-reading the literature on fractional Kelly methods, starting with the complex Thorp paper I read a year or so ago.

So my current plan for the trading is that we will trade as we are now till the new year, then increase leverage by 50% by end of Feb, then assess again.

Yet I do have one other research project that I would like to look at! Most market traders use data that is structured in a way that is a factor of one hour - that, afterall, is why an hour is split into 60 minutes, as 60 factors by 2,3,4,5,6,10, 12, 15, 20 and 30. But most of these time periods results in trading decisions being based on the data at the hour or half hour mark. My thought is that it might be better to be on a time stamp that is not one of these times. For instance, a 29 minute period would hardly ever line up with the other time stamps. If we traded our two markets with different, "unusual" time stamps, it would also be easier to do the trading as well as only one market would require calculations at a time. It might even be possible to trade several different programmes in each market if they were all on different time stamps.

But our charting package doesn't allow back testing of this idea so I am having to look at it in real time. Early data is quite promising, (as it so often is) but it will be several weeks before I have any real results.

No comments: