Wednesday 25 August 2010

A statistical puzzle

Yet again I am returning to a statistical puzzle in the centre of the trading plans. The issue seems clear. The "theoretical" gain per day from the combined trading strategies is about 175 pts. So if £1 = 1pt, how much capital should be set aside to trade this combination of systems?

The trading margin requirement is straightforward; it is £700 for the full position at this trading size. And one can take this number as being something that the broker will have thought long and hard about. Margin is to protect the broker, so if they are happy with £700 this is because they think that is a suitable measure of risk.

So far the worst day-to-day drawdown has been -448 pts, so this suggests that the very minimum required is £700 plus £448 as this would enable you to keep trading after the worst drawdown, about £1150. An additional rule-of-thumb might be to double or treble this figure giving say £2500 - £3000.

But now the problem. If the systems are as profitable as indicated then they will build capital quickly and higher positions can then be put on. One goal of risk management is to enable the account to grow as quickly as possible. Assuming a new position of £1 per point was put on every time the systems have generated £3000, then the position size and the profits from trading grow exceptionally rapidly - too rapidly. Ludicrously high profits result. Can this really be the case?

So bumb down the daily profit forecast to, say, half the calculated number and then assume, say, £5000 per unit. Then what happens? Answer - enormous wealth, but it takes slightly longer to produce.

Of course part of the answer to these issues relates to the fact that as our position size grows, so we will eventually have some market impact - but I can't really see that being a problem in anything other than the very long term, by which time I won't be trading anyway.

So really the question becomes - how close to the theoretical return can we get and are we putting enough capital into the model to protect against something really odd happening? Roll on September 6th, the start of large scale trading and the beginning of an answer to this question!

No comments: