Thursday, 9 December 2010

Latest trading ideas

The main focus of my trading research over the past few weeks has been equity curve management. This can produce some pretty odd "meta" principles for trading. For instance, there may be principles of trading at the meta-equity level that are not applicable at the non-meta, market level. I remember having a long talk with one of my hedge fund investor friends about this and us getting very confused. Back then, I had applied a trend-following trading system to the results of trading a portfolio of markets using a trend-following system and was claiming that I could avoid some of the drawdowns that apply to trend-following by using the meta-principle to reverse our approach - so we would be trend followers then the equity curve was up and counter-trend traders when it was down. But this is a pretty confusing overall methodology and is very hard to set up as you have to record what you would have done trading the way you currently aren't doing.

One idea that does seem valid for counter-trend trading is a limit exit for each day's trading. For instance, if we were 75 pts up for the day, it seems that the most common outcome is not that we add 30 pts more, but that we lose some proportion (even more than 100% on occasions) of the day's gains. So a limit exit locks in a profit for the day.

A second idea has been that of a limit exit decay curve - something like the time value for an option. So at 3:00pm, we still seek x points, but by 7:00pm, our limit exit would be lower. This might stop us being fully invested as the evening progresses to the close, when there are often odd moves to do with position-squaring.

Finally, I have just acquired a copy of Ralph Vince's The Mathematics of Money Management, which is a long work on Kelly Optimal f trading. I expect to put in quite a bit of time reading this over the coming weeks

There was a show on TV last night called The Joy of Stats - I hope to see this later on today. I can see that quite a lot of my future research will be to do with statistical testing. Maybe it is time I had a refresh of my knowledge of basic statistics e.g. how to test whether two samples come from the same overall population.

No comments: