Continuing the thoughts from the November 1st post, we concluded with the mandates
- Your strategy should behave in all types of markets
- The markets are always changing, and the configuration must change too.
Indeed, I know it is not possible to make a perfect strategy that works all the time (10th July post). At this time I thought I needed to run three strategies to take advantage of the market (Uptrend, Downtrend, Sideways).
However, as discussed on October 11th, I could easily see periods where a profitable trending strategy did not take advantage of an obvious trend. To do this, I need a different set of configuration and that is when I started thinking about running multiple trend strategies.
Yes that’s right, multiple trend strategies. It makes sense, if we categorize trends by steepness, the same configuration for a sharp short trend will not work for a slow long trend. Likewise, I’m sure we will eventually need multiple ranging strategies depending on the volatility and scale of the range. I can even see a combination of both (a trending but ranging market).
The first step is clear though, we need to identify the different markets types. i.e., categorize different trends. I have actually done some similar work already and I can continue this (see Gradient Calculation for Moving Average – June 30th)
So let’s go with that:
- I’ll do some research into categorizing different trends.
- Once done, I can determine the best set of configuration for each type of trend.
- I can ensure that each configuration only trades when it is best for that configuration.
- Then I can run multiple configurations at the same time and I expect only one configuration or none to be active at any one time.
Maybe I’ll even look into combining them all into a single strategy. This may make backtesting easier but I suspect it will be slow. It really depends on what data and algorithms I need to categorize the different markets.
However, my current aim will be to run multiple strategies simultaneously. Often I have seen the term suite used, for example, a suite of algorithms etc. That’s fine, and indeed I have a suite of different algorithms that I use. However, I believe the term a fleet of strategies is more fitting. This is more appropriate as the suite of strategies are now operating under a single command.