After identifying the need to diversify, my portfolio grew very much in number of strategies and, at a certain point, it was no longer possible to keep it under control efficiently! There were simply too many things to analyze and to consider: what was working and what not was hidden in the monster I created…
I understood I had to rationalize the approach, first to find a way to decide which strategies should work and which not and then to position size them properly.
Working on a portfolio I concentrated on the shape of the equity line of each strategy, I simply considered its most recent performance. After deciding how often I had to look into the portfolio, I fed my System with daily profit/loss figures of all the strategies and, on the decided period I looked into those numbers. I do that on a monthly basis but it can be done also every week (even though I don’t think this would lead to particular additional benefits).
During the check, I look into a longer-term performance (something around 12 to 18 periods, months in my case) and a shorter term one (3 to 6 periods), both shall be positive and the shorter term must also show an average monthly profit higher than the longer term. This means we want performing strategies and strategies that show momentum in their profits in the most recent period.
All the strategies passing this criterion are valid to trade during the coming month, all those which fail will stay in incubation.
To position size each strategy, again daily profit/loss figures are used: the worst day of each strategy is taken as Worst Case Scenario and used to calculate the number of contracts after deciding the percentage of risk to take.
The numbers we get can be further adjusted (reduced) considering the max drawdown of each strategy and correlations among systems.
The math related to this is not conceptually difficult but it requires some time and, once it has to be applied to a whole portfolio, it leads to an enormous calculating effort, with standard tools it could even take several hours to get some results.
We have also to consider that our basket should be widely backtested to identify the best set of parameters to use (risk%, long period, short period, etc.)… To do this titanic task we decided to developed a dedicated software: TITAN!
TITAN get as inputs the daily profit/loss of all the strategies, we can be set different testing parameters, and as an output, we get the list of strategies to trade during next period and the number of contracts for each of them, as simple as that! But incredibly effective!
A portfolio is important in terms of content (good strategies) and maintenance (all the above operations), this is where I am today, now it’s your turn!