Wednesday, June 13, 2012

Bayesianism in Baseball Oddsmaking

Though I personally do not have any interest in placing bets using actual money, I do like to look at the odds that online betting sites give for certain events. Since these odds move with the money that's placed on them, they feel to me like they're a version of an inTrade-style prediction market. Right now at Ladbrokes, the 2012 US Presidential election is at Obama 8/13, Romney 5/4, suggesting a genuine Obama lead. Tiger Woods is an 8/1 favorite to win the US Open. But this post is about the odds to win each division of Major League Baseball this season. The correlation between current standings and the odds given to each team is shockingly weak, in a way that I feel gives a bit too much respect for the pre-season consensus and insufficient respect for the 40% of the season that has already taken place. Thus the reference to Bayesianism in the title: the oddsmakers, it looks to me, are overweighting their prior probability estimates against the actual data.


In the National League Eastern Division, the Washington Nationals currently hold a 4-game lead over the Atlanta Braves, with the New York Mets 5 games back, the Miami Marlins 6 games back, and the Philadelphia Phillies 9.5 games back. According to Ladbrokes, the Nationals are 2/1 favorites to win the division, with Atlanta fractionally behind them at 21/10, the Phillies (!) at 3/1, the Marlins at 4/1, and the Mets at 22/1. That last figure is much shorter than the, say, 40/1 or 50/1 odds I've observed earlier in the year, but it's still pretty extreme.

In the National League Central Division, the Cincinatti Reds hold a 1-game lead over the Pittsburgh Pirates, with the St. Louis Cardinals 3 games back, the Milwaukee Brewers 5.5 games back, the Houston Astros 7.5 games back, and the Chicago Cubs 12.5 games back. Ladbrokes has the Reds as 4/7 favorites to win, followed by the Cardinals at 7/4, the Brewers at 12/1, the Pirates at 14/1, and the lowly Cubs and Astros at 350/1.

In the National League Western Division, the Los Angeles Dodgers hold a 4.5-game lead over the San Francisco Giants, with the Arizona Diamondbacks 9 games out, the Colorado Rockies 14.5 games out, and the poor San Diego Padres 18.5 games out. Ladbrokes has the Dodgers as 10/11 favorites, with the Giants at 6/5, the Diamondbacks at 13/2, the Rockies at 33/1, and the Padres at 250/1.

In the American League Eastern Division, the New York Yankees (ugh) have a 1-game lead over the Baltimore Orioles and the Tampa Bay Rays, with the Toronto Blue Jays 5.5 games back and the Boston Red Sox 6.5 games out. Ladbrokes has the Yankees as 5/4 favorites, with Tampa Bay at 15/8, the Red Sox (!) at 5/1, the Blue Jays at 8/1, and the Orioles at 12/1. Those would be the Orioles who've spent much of the year so far leading their division.

In the American League Central Division, the Chicago White Sox, under new manager and former Mets hero Robin Ventura, have a 1.5-game lead over the Cleveland Indians, with the Detroit Tigers 6 games out, the Kansas City Royals 8 games out, and the Minnesota Twins 8.5 games out. But Ladbrokes thinks the Tigers are favored at 5/6, followed by the Indians at 5/2, the White Sox at 13/5, the Royals at 33/1, and the Twins at 66/1.

Finally, in the American League Western Division, the Texas Rangers hold a 3.5-game lead over the Los Angeles Angels of Anaheim, with the Oakland Athletics 9 games out and the Seattle Mariners 9.5 games out. Ladbrokes says that the Rangers are 3/10 favorites to win, by far the shortest odds of any team, with the Angels at 21/10 against and the A's and Mariners at 200/1.

There's a clear pattern here. In the NL East, the betting market continues to love the Phillies despite that team's dreadful play, and to hate the Mets despite that team's decent-to-good play. In the NL Central, nobody's buying the good form of the Pirates, despite the signs of decency that team showed last year. In the AL East, similarly to the NL East, nobody's buying the early success of the lowly Orioles or the early struggles of the typically-mighty Red Sox. In the AL Central, people just can't believe that the Miguel Cabrera-and-Prince Fielder-and-Justin Verlander Tigers are really this bad, or that the supposedly-in-rebuilding-mode White Sox are really this good. In every case, the teams people expected to do well are doing much better in the odds than they are in the standings. Now, there's certainly a point in the season at which that's reasonable, but I have a feeling mid-June is a little bit past that point. We're through 60 games already, getting very close to the half-way point. This is not that small a sample size anymore.

(The AL West is a slightly different dynamic. Before the season, the Pujols/C.J. Wilson-acquiring Angels were heavily hyped, but they got off to an awful start while Texas was dominating. But for over a month now the Angels have been gaining on the Rangers, with Pujols finally waking up. Arguably this is a case where there has in fact already been a reversion to the pre-season expectations, but the markets haven't caught up with it yet. If, however, we assume that the dramatic early season divergence in this division was strong enough evidence to displace the pre-season assumptions, unlike in all the other divisions, then the pattern of being excessively slow to believe new evidence holds constant.)

I have a feeling that, if you put money on the Orioles, the White Sox, the Angels, the Mets, and the Pirates, you'd make money at the end of the season. Not, of course, that I'm going to.

No comments:

Post a Comment