The Official Blog of Retail Backtest
About     Recent Entries


Log Train

Retail Backtest is presently still simply my project. I'm Mike O'Connor. Read about me and what I'm up to at www.mocpa.com. If you want the career details, run your cursor over the double-breadcrumb-trail at the very bottom of that page.

This is a development that has been years in the making. I had pursued the quantitative analysis of investment alternatives, especially with regard to equities, for well over a decade. But I had another career, and computerized mathematics for portfolio management was an after-hours pursuit.

Things changed when I quit the other career and adopted a programming style and platform that permitted me to carry through with all of the steps that are necessary... to determine the likely extent to which any given proposed approach to portfolio management that worked well in the past would work well in the future.

But the circumstances are that it isn't a cut-and-dried business at all. It's quite difficult, and inside and outside of academia there are stark differences of opinion among mathematics-wielding financial professionals as to how to proceed. And so yes, we'll talk about those things in the pages of this blog.

To date, all of the analyses that I have done have relied simply on price histories, or occasionally on volume histories. Sometimes that's referred to as "technical analysis", to distinguish it from "fundamental analysis". The supposition is not that others can't produce good results based on fundamentals. To the contrary, it's good if they do succeed as that assures us that prices are reasonable and we might then have an easy time staying long in calm, favorable markets. Rather, price histories are attractive for analysis because the data sets are dense: we have daily prices or even intraday prices to study. And with statistical testing, the number of data points matters quite a bit. Quarterly data, which are actually even a bit stale by the time that they arrive, may not do. So the first question to ask is "does price history matter?"

A short answer is emphatically yes! I was quite skeptical of claims by a few academicians to the effect that a simple scheme involving trailing price ratios as a measure of "momentum" seemed to work. And my skepticism was supported by certain other academic research. However the schemes of those academicians who went so far as to attempt to debunk any such technical analysis of that kind were successfully challenged within academia as unreliable, in the same way that someone who throws the baby out with the bathwater is unreliable. More to the point, my own "walkthrough" simulation approach, with a certain follow-up step that I call "suboptimization", proved the value of momentum.

Another impetus behind the recent progress of the Retail Backtest project was the discovery that RB's New Program, as I have dubbed it, is potentially much better than momentum.

The whitepapers that are accessible via the Articles menu of the Retail Backtest site are long and parts of them are complicated. So these blog entries are designed to be more readable. Write to me directly with questions or comments, and corrections are also appreciated.

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