How Do I Assess whether my Strategies are Curve Fit or Not?
This is perhaps the most important consideration to assess in testing the long term performance of your strategies. It is very easy these days to generate wicked equity curves that ascend in a linear way to the stars….however fall off the cliff as soon as you take them to the live trading environment.
What you need to realise is that it is the market condition that decides your fate…..not the cleverness of your system design.
For example, let’s assume you have developed a trend following system. It is illogical to conclude that your strategy will perform favorably at all times. Your fate is determined by the market. If markets are not trending then your system will not perform…..but this is where you need to deploy risk management methods to prevent your drawdowns building too fast.
So the way we can identify if our strategy is curve fit or not is to map the equity curve against the market condition. We do this by superimposing market data against the equity curve and mapping those conditions when the strategy performed well, and in those conditions it didn’t.
Have a look at the equity curve below.
Firstly have a look at the blue realised equity curve. Note that it is not linear. This is symptomatic of any trading strategy. Trend Following strategies perform well when market conditions are favorable, but build drawdowns when they don’t. For a trend follower you should be looking for stepped equity curves…..not linear ascending ones. If you are a trend follower with an equity curve that is linearly rising……then you are kidding yourself and your strategy is curve fit.
Next, have a look at the grey line which represents market data. Note where there are strong trends and where they are not. Your equity curve should rise during strongly trending conditions and drawdowns should build when they are not. Refer to the yellow and pink shaded regions. The yellow shaded regions are when the equity curve performed strongly. Ensure that during these times that markets were also trending. Note that despite the strength of this group of strategies in capturing most trends, it does not catch all of them. This shows there is room for improvement in finding additional strategies that can plug these gaps.
The pink regions are when drawdowns were building. Refer to market data to understand why? You will see that conditions were choppy during these periods.
Now note how few trending conditions were observed over the 20 year time horizon in general. In a nutshell there were only 6 periods which allowed your equity curve to soar using this particular suite of trend following systems. This is symptomatic of trend following. Most of the time you are managing drawdowns. Trending conditions are unpredictable in nature.
Key to a successful trend trading strategy is actually not in catching trends as any trend following system should be able to achieve this. The key is in how to prevent capital deterioration during unfavourable market conditions. Trade frequency is essential here. You want to keep it low per strategy as significant trending conditions are few and far between. Avoid the noise and frequent whipsaws of over-trading when significant trends do not exist. This is contrary to what most traders think (that prefer higher sample sizes) and why trend trading needs to stick to higher timeframes where noise is less prevalent. Diversification is your only way to keep trade samples low per strategy but increase trade frequency through more ‘significant’ opportunities.
Trade well and prosper