Projection Systems Part II – The Pitching


Last week, I looked at how the computers liked—or rather, didn’t like—the M’s offense. But how about those pitchers? Again, we look to Marcel to forecast the AL West division. This charts projects total runs per nine innings allowed.

Marcel2011 M’s2012 M’sAngelsRangersA’s
Starters (R/9)

It’s almost impossible to figure out who’s going to start for the A’s. They have the 39-year-old Bartolo Colon, and a couple guys coming back from season ending surgeries. Those concerned about a closer with the name Grant Balfour should be happy to know that Josh Outman is there to back him up. The Angels project the best, and that makes sense. Any manager would be happy with Ervin Santana as a fourth starter. And how are we supposed to project Yu Darvish? I went conservatively—in the spirit of Marcel—with 180 innings and a 3.80 ERA

The M’s staff is full of mystery and intrigue this season, so let’s look at how the projection systems average them out. Again, I use the averages of the six projection systems—Bill James, Marcel, ZiPS, Fans, Rotochamp and Steamer—and then look at how consistent the projections were for each player. (The “Dev Score” is the sum of coefficients of variation in each statistic over the six projection systems, for those interested).

“Easy”K/9BB/9ERAFIPDev Score
Felix Hernandez8.422.603.043.100.11
Jason Vargas5.792.664.214.200.12
“Meh”K/9BB/9ERAFIPDev Score
George Sherrill8.103.713.653.770.20
Brandon League7.052.393.333.460.21
Shawn Kelley8.033.163.874.000.22
Charlie Furbush7.273.194.564.660.25
Hector Noesi6.733.194.434.100.25
“Hard”K/9BB/9ERAFIPDev Score
Blake Beavan4.601.764.404.220.34
Chance Ruffin8.
Tom Wilhelmsen7.073.874.344.200.47
Steve Delabar7.915.324.554.570.64

The “easy-to-predict” guys, Felix and Vargas, are familiar faces in Seattle. We know what they’re capable of, the projection systems know what they’re capable of, and I think those are fair predictions. I might have penciled Vargas in to have a slightly lower ERA, being a lefty pitching in Safeco half the time, but in three seasons he has shown no ability to utilize Safeco and outperform his FIP. So 4.21 seems accurate.

On the other end of the spectrum, Wilhelmsen and Delabar were all over the map. ZiPS had Wilhelmsen at 5.3 K/9 while Rotochamp has him at 8.4! As for Delabar, ZiPS pins his walk rate at 7.8 walks per nine while Marcel says 3.1. For some context, I found ZERO players since 1990 (100+ innings requirement) that posted anything close to 7.8 BB/9. Coincidentally, the player that came closest was our very own Randy Johnson in 1991, who recorded 152 walks in 201 innings as a spry 27-year-old. Obviously there is some sampling bias in that most pitchers who are walking that many batters will never be allowed to reach 100 innings, but at the same time, I would never project any major league pitcher for that many walks.

To wrap up the projection system version of the Mariner’s outlook, here are the final AL West Standings, based on my interpretation of the Marcel Projections (both offense and pitching).

TeamRuns ScoredAgainstWLGB

This looks a little optimistic for the A’s and M’s, but remember that Marcel regresses all players and teams toward the league average, and I think we’re seeing a little of that here. Last season the AL West went 323 – 325, but the league also gained Albert Pujols, Jesus Montero and Yu Darvish, among others. However, this projection ups the West’s record to an unlikely 349 – 299. I think something closer to 75 wins for the M’s and A’s is more appropriate, and a division-wide record of 336 – 312.

In the final episode, I’m going to provide playoff odds for various projections. In the spirit of Bryant’s poll on expectations, you’ll get to choose what time of team you think the M’s are this season—like, say, a .400 team with high-variance versus .500 team with low variance—and I’ll give the playoff odds for each type of team.