Q. It’s August, and we’ve had a couple of requests for this term, but we’re not talking vacation destinations here…
A. Nope… we’re talking about a statistical modeling technique that first was used to help build the atomic bomb, and has become the preferred way to predict outcome of complex events and financial instruments. These days, every serious financial firm makes extensive use of so-called “Monte Carlo” modeling.
Q. So, why in the world is it named after a haven for casinos?
A. Because in casinos you play the same game millions of times, and therefore know with great accuracy what the odds are of any given result. And that’s the big idea of Monte Carlo modeling: you use computers to run tens of thousands of “iterations” of the event you’re trying to predict, with many different values for the possible inputs. The output is a distribution curve showing just how probable any possible result is.
Q. So how is it better than other kinds of prediction methods?
A. The classic “deterministic” models use average inputs and generate things like “best case” “worst case” and “average case” results. But those don’t tell you how likely any of the three are, or anything about all the other possible results that you might get.
Q. And we all know how misleading averages really can be…
A. Right, easy to drown in a river that’s an average of 3 feet deep. You’d much rather get a complete image of the river at all its various depths, and that’s one good way to think about what Mone Carlo methods can provide to forecasters.