Nate Silver of FiveThirtyEight.com fame has crossed over into football. A few days ago, Mr. Silver in conjunction with ESPN released the Soccer Power Index on ESPN.com. According to Silver, “the SPI rating is designed to provide the best possible objective representation of a team’s overall skill level. In particular, the SPI ratings are intended to be forward-looking: They measure a team’s relative likelihood of victory if a competitive match were to be held tomorrow.”
Translation: If you bet, you’ve got a new index to help you justify why you make your next incorrect bet.
Although Silver comes from a baseball background, it will be interesting to follow his foray into real sport.
Silver launched FiveThirtyEight.com in early 2008 using statistical models to predict political races around the country. He hit the big time when he basically predicted Super Tuesday results during the 2008 presidential primaries. He’s repeatedly beat the projections of “respected pollsters” (whatever those are) since then. And then just like that, he’s famous. Well, at least to people who follow U.S. politics.
The SPI, like most statistical modeling, is based on a series of assumptions. Assessing a model’s assumptions provides some insight into the potential weaknesses of the model. In short, the best statistical data available coupled with reliable and rational assumptions yields the best results. Where there is little reliable information to base assumptions, you get a model that may not predict very well. Enter the problem with football and indexes.
Silver acknowledges that football presents a unique challenge given that international football has little reliable information to go on. Then he proceeds to create this index. So take that into account when placing your bets. I’m not hating on the SPI. I know I couldn’t build the thing. But it just makes me think, why? Why are you building this given that you’ve concluded that there is little reliable information in football? I could make a list from the top of my head that would be very close to the SPI index. In fact, I would rather trust a person who follows the sport religiously to tell me what team is more likely to win than follow some index. And if you don’t know people like that, or don’t follow the sport yourself, you shouldn’t be betting on it in the first place.
Which brings me to another point. Why do we feel the need to have to quantify everything? I can only come up with two reasons: 1) to provide objective support for drunken argument, and 2) for placing your assets at risk, also known as betting. In certain areas, creating predictive models makes all the sense in the world. But in football, at least for me, these indexes make the game more mechanical, more prose than poetry. If you think about it, whether Brazil is ranked above England in any given index is irrelevant at kick off if your goal is to simply watch football.
To me, if Silver gets it right, it will simply be an interesting story. Another notch in Silver’s belt and additional evidence that if you build the right model using the right data, you can get solid results. Yay. But fundamentally, I have no interest in knowing what team is statistically more likely to perform tomorrow. I’m interested in the game’s narrative as it plays out on the field. Call me simple. You wouldn’t be the first.
What do you think of this SPI thing? It’s better than FIFA’s poor attempt at ranking, but it is really do we need this?
