Long-Term Backtests Can Be Misleading [Premium Articles]

Authors of popular books and blog articles often present long backtests of certain strategies that exhibit superb risk-adjusted performance. It is important to realize that it is highly possible that these superb results are due to market conditions that may never occur again. Thus, they can be quite misleading.

For access to premium content, you must be a subscriber. Please login if you are already a subscriber or subscribe to continue reading...

This entry was posted in Trader education and tagged , , , , . Bookmark the permalink.

3 Responses to Long-Term Backtests Can Be Misleading [Premium Articles]

  1. Gerald Morrison says:

    Technology changes everything it touches and the markets are no exception. Speed and sophistication have significantly changed and we now live in a world of E-Trade babies. Old data does not and cannot account for that. There is practically no chance that the statistical properties of pre-computer data are representative of the data we trade with today. But if someone wants to model and trade with the assumption that it is, well, good luck with that!
    Question: How important is the change in daily autocorrelation to a weekly or monthly trader? (Does that change show up in those things frames?).

    • Hello Gerald,

      Significant 40-week autocorrelation from the 1960s and 1970s disappeared afterwards, only to appear again temporarily after the 1987 crash and at the bottom of the bear market in 2003.The chart in the tweet link below shows that. This means that traders and investors using trend-following based on weekly data rely more on drift and those models are not significant.


      • Gerald Morrison says:

        Thank you – just what the doctor ordered. Is the market even investable using quantitative methods anymore? (rhetorical question). I recently signed up for your new book. It looks very informative. I'm looking forward to reading it over the holidays!

Comments are closed.