Uncategorized Comments Off
Back-testing is one of the most important tools for a technical analyst but it’s also important for serious investors. It is immensely useful to be able to understand how a strategy will perform over time. It lets us instantly test an idea like “what if, just before closing each day, I bought the stock that had the highest percentage gain today but didn’t have abnormally high volume?” You can run this strategy over years of data and get an immediate feel for whether it’s a completely bad idea, or something workable.
If you think the idea is workable you can tweak the triggers and see if it can improve. For example, does performance improve if restricted to trading just within small, mid or large caps?
Back-testing isn’t meant as a definitive record of how well a strategy performs, though. No trading simulation is perfect and each has their own biases. However, back-test performance statistics are very informative when comparing strategies. If strategy A performs twice as well as strategy B, strategy A mostly likely will perform much better in reality since all biases were applied uniformly in all tests.
On Market Filters we’ve always let users back-test their customized buy-signals, then scan our intra-day data for stocks meeting that criteria right now. We realized it’s inefficient for users to each have to all run the same, or similar, back-tests (but we didn’t remove the ability to do so). Also, when new to technical analysis it’s not easy to figure out where to start when looking for buy-signal ideas.
To fix both of those problems we just added a new feature that periodically back-tests a large number of popular strategies. Users can then view all the results and sort based on criteria such as Total Gains, Win Rate and Average Gain per Win. This is a great way to find a good strategy or find a starting point for one to adjust for yourself. The new page is: Back-test Results.
This data also has some powerful side-effects. The site used to show ratings for individual stocks based on an arbitrary selection of indicators. Now the site finds the top-performing strategies from the latest back-test run and averages the scores from the best ones. What could be better than ratings from strategies with the best proven performance?