Thursday, October 16, 2008

A guy who wonders both who will be the next president and what the heck is up with Miguel Tejada and Alex Gonzalez?

The man pictured at left is named Nate Silver. (You can tell he's smart because the newspaper or whatever that published this photo decided to have Silver sit near a computer.) Until a few months ago he was best known for his work at Baseball Prospectus, where he wrote articles with titles such as "Is David Ortiz a Clutch Hitter" and "Binomial Distribution (or What the Heck is Up with Miguel Tejada and Alex Gonzalez?" Also, he invented PECOTA, which the BP website defines thus:
Stands for Player Empirical Comparison and Optimization Test Algorithm. PECOTA is BP's proprietary system that projects player performance based on comparison with thousands of historical player-seasons. Analyzes similarities with past player-seasons based not only on rate statistics, but also height, weight, age, and many other factors.
This is pretty smart: there's just about nothing in baseball that hasn't already happened in it's 100+ year history, so let's let that mound of data help inform our conclusions. And it's pretty clear that PECOTA is just about the best thing out there for predicting what will happen in baseball.

Mr. Silver decided to bring his predictive genius to the world of politics, starting a blog called (named after the 538 electoral votes in the presidential election), with the confident byline of Electoral Projections Done Right. He wisely declines to divulge his exact methodology, but Silver does explain some of what goes into his projections:
Firstly, we assign each poll a weighting based on that pollster's historical track record, the poll's sample size, and the recentness of the poll. More reliable polls are weighted more heavily in our averages.

Secondly, we include a regression estimate based on the demographics in each state among our 'polls', which helps to account for outlier polls and to keep the polling in its proper context.

Thirdly, we use an inferential process to compute a rolling trendline that allows us to adjust results in states that have not been polled recently and make them ‘current’.

Fourthly, we simulate the election 10,000 times for each site update in order to provide a probabilistic assessment of electoral outcomes based on a historical analysis of polling data since 1952. The simulation further accounts for the fact that similar states are likely to move together, e.g. future polling movement in states like Michigan and Ohio, or North and South Carolina, is likely to be in the same direction.
Silver is becoming a big deal. Appearances like this don't hurt.

Silver's latest projection: a win percentage of 94.7% in favor of Obama, meaning that if this election was run 100 times, Obama would win nearly 95 of those times.

I'm happy that this dude who has produced amazing work in a somewhat limited field is getting the exposure he deserves. He's smarter than maybe all the people who get paid to sit on TV and talk about politics; it's time his talent is recognized. But I'll be satisfied as long he keeps writing about baseball.

No comments: