The Official Blog of Retail Backtest
About     Entries
Entries for tag "theory"

Objectivity and the Trailing Performance Period

One of the most easily reached conclusions in all of logical thinking is that we should always do things the way that worked best in prior experience. Oh, that's not to say that we shouldn't or don't occasionally get inventive. But we surely believe that we should never do anything in a way that worked much worst in the past than something else that worked well.


The mistake has to do with choices. I usually frame the issue by talking about "parameters". By that I mean numbers, numbers that specify your portfolio management scheme's way of learning from the past. The mistake is to simply find out what choice of parameters would have produced the very best results in the entire available history, to thus fix those parameters, and then to use them going forward.

Continue reading »

Use Relative Strength Sparingly?

There is a theory of investing out there that generally goes by one or the other of two names— "momentum" or "relative strength".


But upon thinking relative strength through, we could worry a bit about the idea of focusing on owning only securities of particularly high momentum. What if the familiar commentaries about securities "getting ahead of themselves" or "overshooting" or "becoming overvalued" should fairly often prove to be applicable to our portfolio so that we would see our select few high-fliers reversing on us and returning to earth too often?

Continue reading »

Say What? You're Changing Your Hypothesis?

No rest for the wicked! Let's say that you're a doctor, a clinician, and you're reading a research article on a new treatment. How reliable is the conclusion in the article, which is that a certain treatment would work but only with female patients?


Let's say you've put together a dandy system for portfolio management based on a promising new principle. You try it out on stocks of many kinds. It hardly works overall. Boo! What to do? I know... let's try it on just the financials. Hurray! It works!

Continue reading »

Regime Change vs. Computerized Portfolio Management

In two of the articles on the Retail Backtest website, and especially in this one, you will find substantial discussions on the topic of how we can best go about using the recent performance history of some procedure as a guide toward better performance in the future.

This post is in particular about a two-phase procedure that is to some degree intrinsically adaptable to shifting behavior in the securities markets, something the like of which should quite possibly be a part of any good program for active portfolio management.

Continue reading »
© 2017 Michael C. O'Connor ∅ All Rights Reserved