Sports News Without Fear, Favor or Compromise
Sports News Without Fear, Favor or Compromise

# Bud Selig's New, Random Postseason Knows The Soul Of Baseball Better Than You Think

By

Out of the 22 playoff games in the first two rounds of the MLB playoffs, 12 have featured at least one team on the brink of elimination, and six of those will have been sudden death for both teams. Bud Selig's new postseason format puts a lot more weight on single games, which has led many players, writers, and fans to criticize it as excessively random. But that's just the point: Unlike in the NFL, where best-of-one elimination games are meant to determine the better team, the one-game first round and short second round are designed to embrace baseball's inherent randomness, a property that we can see with the use of some simple statistics.

To illustrate the difference between the two sports, I'll use the concept of odds ratios. Say NFL Team A is playing NFL Team B tonight. They come into the matchup having faced similar schedule strengths, but Team A is 10-6 while Team B is 7-9. The odds that Team A will beat Team B are 10/7 times 9/6: that is, the ratio of Team A's wins to Team B's wins times the ratio of Team B's losses to Team A's losses. The result is about 2.14, meaning Team A is about 2.14 times more likely to win than it would be if the two teams were dead even. Converting odds into probability is easy: 2.14/3.14, or generally x/(x+1). We'd expect Team A to win about 68 percent of the time.

I went through the math because odds ratios give a great back-of-the-envelope estimate for a single matchup. If you know the teams' records, or even just their winning percentages, plugging the above into Google will give you a good idea of what might happen in the game. The estimate changes if you include things like strength of schedule and the relative advantages one specific team has over another, but odds ratios provide the framework for including those too if you're so inclined. Win-loss records are just one measure that works.

Let's use odds ratios to compare the Cardinals-Braves wild-card game to those in other sports. The Cardinals came in at 88-74, while the Braves were 94-68. Using those numbers, the Braves' odds were 1.16 to 1. Another neat thing about odds ratios is that you can make adjustments for other circumstances like home-field advantage (about 54 percent), which usually gives the home team an odds advantage of about 1.17 to 1. Multiplying these together gave the Braves odds of about 1.35 to 1, or a 57 percent chance of winning the one-game playoff.

Compare that with the most recent Super Bowl, another upset. The Giants came in at 12-7, while the Pats were 15-3. The odds ratio from those two records gave the Giants just a 25 percent chance; I would have estimated that the Pats were three times more likely to win.

Obviously, odds ratios aren't the best choice for predicting NFL results. With such a short schedule, it's silly to pin teams down to their records. So what about the 82-game NBA? Had we run the numbers before last season's NBA playoffs, we'd have figured that the first seed in the Western Conference, the Spurs, would beat the eighth-seeded Jazz 72 percent of the time. But if I pit the Nationals, the best team in all of baseball, against the Astros, the very, very worst, I'd get 75 percent, a meager increase when you consider that I spanned the entire league, not just the playoff qualifiers. (I think this helps to explain why the Bobcats were so fucking terrible.) Put more simply: No. 1 seed vs. No. 8 seed in the NBA is as about as certain an outcome as Very Best vs. Very Worst in MLB. What's more, the NBA insists on best-of-seven series for all of its playoff rounds. Baseball, meanwhile, has one best-of-one, one best-of-five, and two best-of-seven. The NBA craves certainty; baseball encourages randomness.

It's not that the sport of baseball is any less predictable than the others. In fact, it's much easier to predict over the course of the season. That's why statistics and baseball get along so well: The sport consists of discrete, random events that paint a very reliable picture over time. But where a single NFL or NBA game yields a great deal of information about the relative quality of the two teams, one MLB game doesn't say too much about whether one team is better than the other.