The Optimizers (part I)

Professor Cockrum pointed to our laptops and said, “these things… they’re making your brains go soft.”


Tough love from a man worth countless millions, who donates his teaching salary back to the school. “All this optimization you’ve been taught, for what...?” he went on, “don’t get lost in the numbers!” That was his mantra - don't get lost in the numbers. This at a school that prides itself on “the numbers,” aka stressing math in the business curriculum. Not more than six months ago all this would have sounded provocative.


With the recent bubble burst now noone trusts the numbers. Consider that for the past decade the world's top banks hired the smartest math wizards, from MIT, Harvard, all PhDs no less, to create their "models." These models were used to allocate the bank's assets. To summarize Taleb, the strength of the model rests not on the numbers within it, but on the assumptions it rests upon.

(This is all pulled from Taleb's "The Black Swan," who explains it much deeper than I. I only offer a sampling here so one can get a flavor for where we're headed.)

Sorta like the foundation to a house. The model could be made of stucco, but if it rests on stilts overlooking the Gulf of Mexico, it loses to a hurricane. Take one assumption - that real estate values will always - always... always, go up. Not stay firm or hold its value. Or track inflation which it's historically done - providing an excellent "invisible" retirement piggy bank for middle-class Americans. No, not this at all. The mood during the bubble was that it would go UP - like Google stock...

And when it didn't, the models failed. And when the models by the smartest math wizards in the country failed, some parts of the bank failed. And depending on how much of the bank relied on the models, that part and then anything related failed. It was a snowball.

A few survived of course. This isn't to claim this is the only reason for what was, and is, a historic collapse which we may, or may not, ever recover from. But it's one reason. And part of it goes back to this idea of optimization.

The idea of optimization says that if you find a good deal, then you can use the combination of a high IQ and computer to make it perfect. In other words, it says you CAN have too much of a good thing.

For instance, let's take a simple allocation of stocks, stocks A and B. Let's assume the goal is risk versus return. This says you want to minimize your risk and maximize your return. Very sensible. This follows on our natural intuition to "have something for nothing."

Now lets' say you studied the market and trends, various rates of return, etc, and you discovered $10 of stock A plus $5 of stock B gave you a nice return for a fair amount of risk. Well, you could be satisfied with this result and try it out. Or you can begin plugging numbers in Excel, run a tool called Solver, and let Excel find in the very best "optimized" allocation - let's say it's $9.88889 of Stock A and $5.11111 of Stock B.

Note you didn't change anything from the first go. You didn't change your "approach" to allocating stocks. You just perfected it. This works when there is one correct answer, like an exam. Or in a competition. If the rules are the same, and I go with the first allocation and you go with the second - you win.

This is why they often call professional sports a battle of inches. If you gain 3 yards a carry, you punt. 3.4 - you carry on.

Of course this is all super-simplistic, but the point here is not think about allocation - it's to consider the idea of optimization. Imagine you now work at bank, you're a math whiz, the new nerdy girl from Stanford just kicked-ass from company C across the street. And people are wondering what's wrong with your model? Do you throw it out the window and start over? Or do you try to squeeze just a little bit more? After all, that's probably what she's doing...