Share
That's a neat way of showing the distribution of outcomes vs time periods - hadn't thought of plotting it that way - I normally just plot the equity curve for each time period and compare them visually (plus compare performance data I compile in a table of results) - reason being I don't just want to see the relative return, equally as important as the total return is how the model gets there (consistency, robustness to trending and ranging periods in a market, depth & length of drawdown, etc).
As an example, below is a comparison of equity curves for 4 time periods after entry for a particular model, with 2 model variants> Its quite easy to see from those graphs the following:
1. The total return for each time period - lowest 1 period & highest for 4 periods
2. The degradation in consistency as the time period gets longer (by the depth of that drawdown in the middle for periods 3 & 4) - best consistency for period 1, worst for period 4
3. The comparison in return and consistency of the 3 options for each time period - clearly the original model provides the best return but the worst consistency
4. Knowing when that drawdown occurred I can review the market charts to look for differences between the good and bad sections of the curves - i.e. was the market ranging during that time, hence causing losses for longer periods but not shorter, was the market trending up and were most of those trades shorts during that period, etc.
From the above, plus data in the performance table, I'll probably choose period 2 as the optimal time period as the best compromise between consistency and return (50% better return than period 1 for a modest drawdown potential that I can tolerate) - and then lock the selected timeframe into my trading plan.
Cheers, Sharks.
