Today I want to talk about a subject that normally causes participants to doze – it’s a very unsexy subject. Unsexy but very important. So bear with me.
There are two issues I have with many commercial position sizing software:
- The software allows only testing on an instrument by instrument basis.
- It reports only on dollar results
There is no problem with testing on an instrument by instrument basis if you are trading one instrument. But if you are trading a basket of instruments, then serious problems rear their heads.
One of the main issues lies in the fact, that unless you can test on a portfolio basis, you cannot assess the damage a drawdown will do to the portfolio. For example, in a diversified portfolio, a drawdown in one instrument may be saved by a profitable run in another. Or, if all instruments suffer a drawdown simultaneously, then their drawdown would have a much greater impact on capital.
The commercial package I like best is the successor to Trading Recipes: Mechanica Standard Edition
I don’t own a copy (I had my own program written) but I have seen it work and I am impressed with it.
The second issue arises because all dollar results are treated equally. But in the real world of trading, this is clearly not the case. The volatility of the market plays an important role in assessing the reasonableness of the return. For example, a risk of $1000 would be very different in an instrument that has an Average True Range of $3000 to one that has an Average True Range of $200.
The Expectancy Formula I quoted in previous blogs is not the one I use for this reason. The formula quoted has been:
(Avg$Win x Win Rate) – (Avg $Loss x Loss Rate) = Exepectancy
I prefer to normalize the result by dividing the result on a one contract basis by the initiating price. Let’s say I bought gold at US$850 and sold it at US$930. The Initiating Price % result would be:
((930-850)/850)x 100 = 9.4%
What I am doing is substituting the $ expectancy with a % of initiating price then translating that to dollars.
For example, let’s say my long-term expectancy is 10%. I enter the ES tonight at 1377. My expectancy for the trade would 1377 x .10 = 137 points x 50 x number of contracts. If I were trading gold and my entry is 890, my expectancy would be 890 x .10 = 89 x 100 x number of contracts.
I find using the % of initiating trade price a more accurate way of assessing expectancy than dollar values.
So then, the Expectancy Formula I use is:
((AvgWin%IPrice x WinRate) – (AvgLoss%IPrice x Loss Rate)) = % Expectancy