Saturday, June 21, 2014

The evolution of a trading approach

"There will come a day when easily discovered and lightly conceived trend following systems no longer work." - Richard Dennis, from his Market Wizards interview

Back in 2012, I made a change to my own approach. This was a direct result of the market conditions and resulting performance achieved in 2011. While the overall approach was not changed, one small element was refined, relating to the cutting of losing trades. To me, the change was simple, logical and easy to understand. Did it have the desired effect? 

The average loss, which prior to the change equated to -0.75R on a losing trade, has been cut and is now around -0.4R. While this change meant a small number of trades that eventually generated profits were eliminated, this more aggressive cutting of losses, and the effect on the compounding of returns and overall expectancy, meant the answer was unequivocally yes.

Since then, the approach has been unchanged, and has achieved decent results. However, after the implementation and success of the new loss rules, it has always been my intention to look again at the the entry rules. Was there a way to improve the quality of the signals taken, which may improve the performance of the winning trades?

Having looked again in this area, I believe it is possible, however it will take time for the markets and my performance to determine whether this has been achieved. By definition, a trend follower cuts losses short and lets profits run. While I do not use time stops, if a trend starts to develop then it is natural for a winning trade to be held longer than a losing one - hence any affect on overall performance will take some time to filter through.

To be clear, this is a small refinement simply to the stock selection or filtering process. Once I am in a position, then the trade management is the same as it ever was - a very regimented, rules based approach.

The aim is to eliminate some of the more sub-optimal set ups (hopefully removing some of the losing trades), and to try and increase the average R return on the winning trades.

Remember that I like to try and enter a trade as soon as a potential new trend is starting, and that is determined by the system parameters used. My own entry and exit parameters are at the shorter-term end of the scale.

A further evolution in my thinking has been to focus more on the risk and rewards element - this has been influenced by the writings of Van Tharp. It is very tempting to look at a long term chart where, on paper, a significant price increase or decrease has developed. But that is only one element to consider. What about your original risk? If you had the choice of obtaining a 10R profit on a trade that was open for say 6 weeks, against a trade that generated a 10R profit over 6 months, which trade would you prefer to take? Again, there are further considerations that each trader has to think about - the frequency of signals being generated, the susceptibility to noise or whipsaws when using shorter-term parameters or timeframes, the trailing stop placement and your ability to deal with the process of giving back open profits, etc.

As an example, think back to Spring 2009. As the markets started giving long signals, entering at that time generated plenty of decent trends and profits. If you were hanging around waiting for say new 52 week or all-time highs, you would have missed out on a whole bunch of decent price moves, and if you were still holding longer-term short positions from 2008, then there could have been a significant giving back of open profits before exit signals were triggered.

On a more general point, and to refer back to the Richard Dennis comments above, does all this mean that trend following systems or methodologies are moving away from a pure systematic approach, one that can easily be fully automated and programmed? In my own case then you would have to say yes, although I am sure that some of these purely visual or discretionary elements can be programmed into a code.

However, if you read my comments about Ed Seykota and his approach here, then you could argue there is always an element of discretion in using a trend following approach. This may be more relevant where someone choose the equities market as his arena, rather than a small basket of commodities or futures, which is where historically most trend followers have operated. As mentioned in that blog post, Van Tharp categorises these as 'rule based discretionary traders'.

No comments:

Post a Comment