Sunday, April 29, 2018

Trend following, robustness and adapting to market conditions

There is a school of thought amongst some traders which says you need to adapt and adjust your chosen method, in response to changes in the market environment. These might relate to the markets you trade switching from a trending to a non-trending state, or from suffering varying levels of volatility. 

Against this, others will say having to go through this process on a regular basis indicates possible over-optimization and a lack of robustness in your method.

To me, continual re-optimization of a trading approach is always a reaction to what has already happened. In effect, you are always 'chasing your tail'. It is a glorified version of hindsight trading. 

The problem is, we will never know what the 'market environment' will be like going forward - as soon as you think you know the answer (possibly by backtesting and re-optimizing), the markets have a tendency to change the question. Far better then to try and to avoid this process.

Generally, trend followers attempt to adopt a robust approach to the markets. We know that markets can switch from a trending to a non-trending state and vice versa, at any time. Because of this, we are prepared to accept a win rate below 50%. Good risk management and adopting the Golden Rule do the rest.

In my own case, I used entry and exit parameters which are fixed regardless of changes in market state, however I use a multiple of Average True Range (which is a volatility-based measurement) to help calculate position size, and my own 'Volatility Factor' indicator as one of the criteria in my stock selection process. These measurements are unique to each individual stock or instrument and are constantly changing. 

This type of approach is used by lots of trend followers, including most famously by the Turtles, who referred to ATR multiples as 'N'.

Robustness is also affected by the inherent level of complexity used in your own method. Most trend following systems are very simple in their nature. I have found to my cost in the past that using additional criteria in my selection process and extra parameters in my core method have led to a loss of performance. As soon as I stripped out that complexity, performance instantly rebounded.

Based on my experience, I know what works for me, so I will keep going with it, and stick to a simple, robust approach while avoiding the siren call of optimization.

No comments:

Post a comment