- One Thing I Learned from the WOPR
By Brian Burke
The WOPR is my game simulation engine. I've had a ton of fun experimenting with different things, finding out when teams should make various tactical decisions that might be uncommon or hard to isolate empirically (directly from the data). But one of the more profound things I learned from the WOPR relates to game outcomes between completely even teams.
The default in the simulation is that the two opponents are exact twins. They're perfectly evenly matched. Without tinkering with any of the parameters, each team wins 50% of the time. I usually run the sim at least 10,000 times to generate reasonable confidence levels, but sometimes I run it just one game at a time and look over the play-by-play and the score. I do this to validate the underlying algorithms and the overall model.
Despite both teams being perfectly equal, sometimes the score is 35-3. Sometimes it's 3-35. Other times it's 34-31. Some scores are 7-6. Admittedly, scores like 27-24 are more common, but not nearly as common as I would have expected knowing the teams in every case are exactly the same.
The lesson is that real NFL teams aren't terribly far from evenly matched, at least most of them. We may observe lopsided outcomes, tight nail-biters, improbable comebacks, low-scoring slogs, high-scoring shootouts and every other kind of game, but we'd be foolish to read anything into a single game, or even just a few games.
Even when teams aren't so evenly matched, we shouldn't ever be too surprised by any game outcome.
published on 9/28/2014