# Projecting the World Series

Predictions are fun, and if the proliferation of articles preceding every postseason series is any indication, I’m not alone in thinking so. I am, apparently, alone in thinking that postseason series previews don’t have nearly enough numbers. Let’s change that.

In my NLCS preview, I looked at a variety of methods one could use to tweak what we assumed about the Mets and Cardinals. I made a variety of tenuous claims and relied heavily on pythagorean records and the log5 method to spit out a bunch of series probabilities. The World Series has a representative from at least one good league, so it deserves better.

Here’s a simple way to estimate the outcome of any game: rather than taking the won-loss record of each team, consider the number of runs they are likely to score and allow in that particular game. For instance, the Cardinals are more likely to win behind Chris Carpenter (pitching like his normal self, anyway) than behind Anthony Reyes. They also hit quite a bit better against righties than lefties, so it’s reasonable to assume that they’re more likely to score runs against Jeremy Bonderman than against Kenny Rogers.

Given a projection of runs scored and runs allowed for each team, those numbers can generate a pythagorean winning percentage which can then be plugged into log5. For instance, if you think the Tigers will score 4 runs and allow 5 in Saturday’s Game 1, that puts their likelihood of beating an average opponent just under 40%. If you predict the Cardinals will score 5 and allowing 4, they have a 60% chance of beating an average opponent. Once you put those numbers together, the Cardinals have close to a 70% probability of winning the game.

Do that for every game in a series, add nutmeg, and you’ll have yourself a series prediction.

Once you go down this road, it’s easy to ignore the gas gauge and keep aiming for one more exit. For this exercise, I tried to find a happy medium between including a lot of interesting data and not obsessing over every detail. I probably failed on at least one count.

For each game, I considered four variables per team: the starting pitcher’s RA (that’s ERA, but including unearned runs, my token effort to include team defense), the starter’s average innings per outing, the bullpen’s RA, and the offense’s likely production for that game. To calculate runs allowed, I figured that the starter would go as many innings as in his average outing and give up runs at his usual rate, and that the bullpen (go figure) would pitch the rest of the game.

For starter RA’s, I used regular season data plus 2006 playoff outings. (For Jeff Weaver, I counted only his time with the Cardinals, because I felt like it.) For bullpen RA’s I did the same, but subtracted half a run, as the weaker parts of the bullpens shouldn’t play much of a role in the series. For offensive production, I pulled numbers out of a hat. Actually, I gave the Tigers 5.07 runs per game (their regular-season average) and the Cardinals 4.7 r/g against lefty starters, 4.9 against righty starters. (Their regular-season average was 4.85.)

I didn’t adjust for league difficulty (I’ll do that later), but I did give the home team an advantage; for more on that, read Cyril Morong’s informative article on the subject.

Now, let’s do some projecting.

###### Game 1 in Detroit: Reyes vs. Justin Verlander

It should come as no surprise that the home team is heavily favored. Verlander’s RA is 3.89; Reyes’s is 5.04. In a perverse way, the Cardinals get a boost here because Reyes is only projected to go 5 innings. After that, the bullpen takes over with its 3.72 RA.

###### Game 2 in Detroit: Weaver vs. Rogers

Weaver’s numbers (4.77 RA) are a little better and Rogers’s (3.99 RA) are a bit worse than the starters who will precede them, but the Cardinals offense isn’t as strong against lefties. Those two differences cancel each other out, giving Detroit a strong chance to take the first two games of the series.

###### Game 3 in St. Louis: Carpenter vs. Nate Robertson

Cy Carpenter to the rescue. Despite St. Louis’s lesser production against lefties, the combination of a bona fide ace (3.33 RA) and home-field advantage give the Cardinals an excellent chance to chalk one up in the win column.

###### Game 4 in St. Louis: Jeff Suppan vs. Bonderman

Numerically speaking, this is the closest matchup the series has to offer so far. Suppan’s NLCS starts suggest he’ll beat his 4.47 RA. Who knows what Bonderman will bring to the table after his long layoff?

###### Game 5 in St. Louis: Reyes vs. Verlander

Home-field advantage has the potential to turn the tables: the Cards have a much better chance of beating Verlander the second time around.

###### Game 7 in Detroit: Carpenter vs. Robertson

If Oliver Perez‘s six strong innings are any indication, every postseason game should be viewed as a tossup, especially those of the winner-take-all variety. Even if that weren’t the case, the potential Game 7 couldn’t be much more evenly matched; the Cards with their ace, the Tigers in their home park.

I don’t want to wallow too deep in the muck of league adjustments here, but it’s well enough established that the AL was the superior circuit in 2006 that we ought to at least consider it. Using Mitchel Litchman’s 2005 numbers, the AL has an advantage of .42 runs per game. Adjust the formula accordingly, and Detroit’s advantage goes from heavy to commanding. Games 1, 2, and 6 are better than 2/3 propositions for the Tigers, and only Carpenter’s Game 3 start favors the Cards.

###### Series Projections

Without the league adjustments, these numbers give the Tigers a 63.4% chance of winning the World Series. Add the league adjustment and that number shoots up to 70.6%. Here is the likelihood of each series outcome using the league-adjusted and non-league-adjusted figures: