Mathew’s recommendation when running a back test has always been to use a period such as October 2001 to October 2009 to get both bull and bear movements within a zero index change.
In recent years we’ve seen many changes to the way markets are traded. For example, the influx of new traders through the various mobile platforms, influencers on Reddit etc, AI trading and so on. Do you think the market now is so significantly different that we should view the results we get in the 2001-2009 with more caution?
Kim, Second part first… I specifically really can’t answer your question with any data that proves a structural change in the market but with new tools like AI but it would not surprise me that previous patterns that were profitable could be arbed away as they are discovered etc. but that should also mean some new patterns develop or emerge over time. As for using the Bear to Bull to Bear market cycle I personally prefer testing using rolling start periods because if you are using any market trend or regime filter (especially over that time frame) you know you are going to outperform in that period because of the lower exposure your system will have in the market so you get a point of entry advantage.