In working on my testing models, I just realized an error I have made in my approach – specifically in measuring the trend. Thought this was worth to quickly share – a type of “tip of the day”.
When developing a back-testing engine model, you might/should include a variable that assesses the trending conditions of the instrument you are testing. There are various ways to do this – some are more difficult than others (particularly from a coding/argument-articulation perspective) – examples might be alignment of multiple moving averages, market structure assessment using swing points (higher highs, higher lows) or indicators of some sort.
How often do markets trend? There isn’t a black & white answer to this. However, doing a quick Google search will probably get an answer in the region of 25-30% of the time – that’s a good enough estimate for the purpose of this blog post.
With that information in mind, what proportion of the time does your testing engine label market conditions as ‘trending’?
The hint of the day is – ideally around 25-30% of the time. If the figure is significantly different (say 50-60%), then your method for assessing trending conditions is most likely quite poor. To illustrate, my model was saying that the market was trending (either up or down) about 90% of the time – hence my mini eureka moment!