Wednesday, October 01, 2008
Postseason odds
Posted by Dave StudemanEach day of the postseason, we'll be posting the odds of each team winning its division series, league championship and the World Series. Those odds will be calculated by James Holzhauer, who just happens to run a website called BaseballPlayoffOdds.com. Here are Jamie's odds as of this morning:
| DS% | Lg% | WS% | ||
| Boston | .534 | .301 | .168 | |
| Tampa Bay | .621 | .316 | .167 | |
| Chicago (A) | .379 | .143 | .055 | |
| Los Angeles (A) | .466 | .240 | .114 | |
| Philadelphia | .543 | .272 | .139 | |
| Chicago (N) | .520 | .284 | .147 | |
| Milwaukee | .457 | .200 | .092 | |
| Los Angeles | .480 | .245 | .117 |
And here's an explanation of how he develops his odds, written by Jamie himself...
The concept of postseason odds is not new to the internet. Popular odds sites include Baseball Prospectus, CoolStandings, and Sports Club Stats. These are good resources, but all fall short in some departments: some don’t break ties, others use primitive methods of estimating each team’s true talent level, and none account for injuries or pitching matchups, which have a huge impact on a team’s chances.
In the playoffs, these errors are especially glaring. Carlos Quentin is a big reason the White Sox have made it this far, but he won’t do the Pale Hose any good on the sidelines. The Phillies have gotten terrible production from their number 5 starters this year, but that’s irrelevant now. My numbers are designed to only reflect the things that really matter.
How are the numbers developed?
The Framework
I work as a professional baseball handicapper; it’s my job to know what the real odds in every game should be. For each game left on the schedule, I calculate the chances that the game will be won by Team A or Team B, based on the projected lineups, defenses, starting pitchers, bullpens, and home-field advantage. These estimates incorporate a combination of preseason projections and current season stats.
During the regular season, I use these probabilities to perform a Monte Carlo simulation, which provides estimates of each team’s playoff chances. This is similar to the method used by the three sites above, but more accurate inputs lead to more accurate results.
The Playoffs
For each playoff series, I calculate the individual probabilities of a team winning Game 1, Game 2, etc. Once I’ve done this, it’s easy to calculate the chances of that team winning the series. This can happen in ten different ways for a 5-game series:
- 3-0: 1 permutation (WWW)
- 3-1: 3 permutations (WWLW, WLWW, LWWW)
- 3-2: 6 permutations (WWLLW, WLWLW, WLLWW, LWWLW, LWLWW, LLWWW)
For example, here’s how to calculate the probability of the team taking the division series by winning Games 1, 2, and 4:
P(WWLW) = P(Win G1) * P(Win G2) * (1-P(Win G3)) * P(Win G4)
The team’s probability of winning the series equals the sum of the probabilities of each of the ten permutations:
P(Win series) = P(WWW) + P(WWLW) + … + P(LLWWW)
The math for a 7-game series is similar but more complex since there are more winning permutations.
Forecasting Future Series
If we know the probabilities of the Cubs defeating the Dodgers in one NLDS and the Phillies defeating the Brewers in the other, we also know the probability of Chicago facing Philadelphia for the NL Pennant:
P(ChC vs. Phi matchup) = P(ChC win NLDS) * P(Phi win NLDS)
Using the same method as above, I calculate the odds for each game of this potential NLCS matchup, and use those to determine the chances of the Cubs beating Philly in the series. Then we multiply probabilities to determine how often the Cubs will defeat the Phillies to win the NL:
P(ChC over Phi) = P(ChC win NLDS) * P(Phi win NLDS) * P(ChC defeat Phi in NLCS)
This is one way the Cubs can win the NL. They can also defeat the Brewers in the NLCS:
P(ChC over Mil) = P(ChC win NLDS) * P(Mil win NLDS) * P(ChC defeat Mil in NLCS)
P(ChC win NL) = P(ChC over Mil) + P(ChC over Phi)
The same methods are used to estimate World Series probabilities if the Cubs get that far.
Shifting Odds
On my website, BaseballPlayoffOdds.com, you can watch the odds change each day in response to that day’s results. For example, if the Cubs win their first game against the Dodgers, their World Series chances improve from 15.0% to 20.0%. Conversely, if they lose, that number dips to 9.6%. Even if you don’t like the Dane Cook commercials, you can see how important every October game is.
Dave was called a "national treasure" by Rob Neyer. Seriously. Comments about this article can be sent to him through the miracle of e-mail.