Algorithms are everywhere these days.  They control what we see online, even when we don’t want them to.  They are behind all sorts of marketing, even when we don’t understand that to be the case.  And they play a significant role in the investment world.

So, as part of a sideways look at investment styles,the research puzzle | This PDF will update with information about the series of postings on the topic. it makes sense to think a bit about the spread of the algorithmic culture.

If a simple definition of an algorithm is “a set of rules for solving a problem,” then we can see that they are at the very heart of investment process, whether it’s found in computer code or the brain of a sixty-year-old investor that’s been building that set of rules his whole career.  But increasingly it’s the former, taking a bigger and bigger share of the duties from the latter.  And all the major firms are happy to help you make the transition (witness, “Where to go when you need an algo”Securities Technology Monitor | Interesting review; too bad it is in an annoying, unnecessary, click-seeking slideshow format.).

There has been a lot of attention paid to high-frequency trading (HFT), its effects on markets, and whether the game has gotten out of hand.  We’ll leave that for another day and focus on algo penetration into more traditional parts of the business.

The first stop is in the trading realm, where a very high percentage of trades occur via some type of algorithm.  HFT accounts for much of that activity, but also included are orders for asset management firms, for example.  Instead of blocks worked by traders with their Wall Street counterparts, many orders are now parceled out by machine according to prescribed rules.

In portfolio management, there are purely quantitative approaches and purely qualitative approaches and everything in between.  Ascertaining where a manager is on that continuum is very important.  The patterns of returns and patterns of errors differ based upon the mix between qualitative and quantitative (and the processes that govern the integration of them).

Research can be automated as well, so knowing whether a firm is strictly quantitative or not can help you figure out how to use the research well — on its own or in concert with other research.  (A tip-off to quantitative research in the written form is the stilted wording; it’s difficult to get the verbiage to sound natural, although you’ve probably read some research that was composed by a computer and not realized it.)

All of this brings up critical questions for due diligence (assuming you don’t take the view that the proof is in the performance).  How do you figure out where the human issues are and the algorithmic issues are (which, of course, are human issues too, committed to code)?  It is harder to audit a machine, but as with auditing a qualitative investment process, it’s all about picking up the pieces of the mosaic and figuring out where they go.

There will always be unknowns, but by understanding as best you can the overall decision structure, you can get an idea of how a firm is different from others — and whether it has made the proper adjustments in light of changing market dynamics.

Among the outlier events to be worried about are so-called “runaway algos,” although other problems can unfurl slowly over a long period of time, as happened with AXA Rosenberg.SEC | This is a summary of the charges and settlement related to the firm’s computer error. And, you have to remember that the computers run in packs too.  So “live by the algo, die by the algo” might be an appropriate motto.  Whether through direct programming or via some artificial intelligence, the computers do what has worked, just like humans.  Therefore, a previously-winning strategy can come unwound quickly.  The quant fund bath of August 2007 is a good example of that.

We are all skipping our way to Algo City it seems.  When we get there, we need to get a good look at the wizard behind the curtain, with his levers and his dials, and determine whether he knows what he’s doing — and whether the machine is doing what he wants it to do.  That all requires taking a fresh look at the due diligence process, to make sure the the yellow brick road is all that it is promised to be.