Random sports discussion, plus anything else I want to write about.

Saturday, January 06, 2007

Projecting the Mariners in 2007

Hello there. I have been trying to avoid the Mariners as a subject of discussion as of late, due to the saddening ineptitude of their front office this offseason, but with the recent struggles of the UW basketball team, and the absolutely atrocious KC-Indy game on right now, I have turned to my trusty companion, the Interwebs, in an attempt to find solace regarding the prospects of the upcoming baseball season for the M's. One of my favorite aspects of the offseason is making projections for the upcoming season, and a simple search of the webs reveals that there is clearly no shortage of widely available forecasting systems for MLB. Of course, the gold standard for forecasting systems is PECOTA, published by Baseball Prospectus, but alas, it is yet to be unveiled for this upcoming season, and due to the proprietary nature of its formuli, I would feel squeamish about reproducing too much of its data in this freely available forum. The good news, though, is that there are a number of freely available forecasting systems that are also quite good, among them ZIPS, from Baseball Think Factory, Tango Tiger's Marcel, and a new system I've discovered this year, Chone, produced by Blogger's own Chone Smith, who has really done some impressive work this offseason, despite being an Angels fan. Chone ran r-correlations for the various forecasting systems about a month ago, and came up with these results for offensive projections for these three systems: ZIPS, .684, CHONE, .677, and Marcel, .664. Earlier research by Tango Tiger revealed that the strongest r-correlation a projection system can have is about .73, and all three of these systems come reasonably close to that figure.

With the individual player forecasts available on these sites, it's fairly easy to make a reasonable projection of a team's preformance for next season, particularly an offense's performance, just by making some predictions for playing time. A team's cumulative OPS corresponds amazingly well to their runs scored; i.e., a team with an OPS of 750 will score about 750 runs, within 10% in either direction. Last season, the Mariners' OPS was 749, and they scored 756 runs. The Giants produced an OPS of 746 and scored 746 runs. 15 teams in MLB last season had OPS and runs scored figures that differed by less then 20. This measure works really well for middle-of-the-pack offenses, which is where I figure the Mariners to be. This measure does not work as well when predicting a team's pitching performance, but it can still be used to make a rough estimate of runs allowed, and with the runs scored and runs allowed figures, one can estimate a team's Pythagorean record (explained here for the uninitiated). As it currently stands, I would expect the Mariners line-up for next season to look something like this, with my predicted PA's for each player in parentheses:

CF Ichiro (700)
3B Adrian Beltre (650)
DH Jose Vidro (500) (goddammit)
LF Raul Ibanez (650)
1B Ritchie Sexson (600)
RF Jose Guillen (500)
C Kenji Johjima (500)
2B Jose Lopez (600)
SS Yuniesky Betancourt (600)

This seems about right, though I have no clue what they plan on doing with Ben Broussard. I hope they don't plan on playing Vidro at 2B because he's useless in the field at this point. There was some discussion about him playing first, which makes me think they'll move Broussard for nothing, as Bavasi seems to enjoy doing that. Here's how I figure the bench distribution of PA's will look, if they keep Broussard:

Broussard (300)
Rene Rivera (ugh) (150)
Willie Fuckin' Bloomquist (200)
Mike Morse (100)
Jeremy Reed (150)

These are complete guesses, as they will probably make another terrible move that will alter the bench, most likely involving Reed or Broussard, but I can't predict the future. These estimates give the M's 6200 total PA, about what they had last year. Now, here's the projected OPS figures for each player, by CHONE, ZIPS, and Marcel:

Ichiro: 770, 789, 794
Beltre: 811, 779, 833
Vidro: 780, 726, 761
Ibanez: 762, 800, 812
Sexson: 842, 824, 854
Guillen: 795, 776, 800
Johjima: 785, 771, 774
Lopez: 775, 737, 739
Betancourt: 695, 697, 739
Broussard: 792, 805, 811
Rivera: 643, 545(!), 726
WFB: 697, 607, 665
Morse: 696, 715, 800
Reed: 741, 728, 725

Each system basically agrees on Ichiro, Johjima, Broussard, Sexson, and somewhat surprisingly, Reed. More variance on Beltre, Ibanez, and the young guys not named Reed. So here's the team OPS for each projection system: CHONE: 771, ZIPS: 757, Marcel: 784. The three systems come up with pretty similar numbers. The Mariners can be expected to score between 700-850 runs next season. Out of the three projections, I think the ZIPS numbers seem too pessimistic, and the Marcel numbers a bit optimistic. Marcel is the least sophisticated of the three systems, as it is just a three-year weighted average regressed to the mean, and adjusted for player age, and I think it may be a bit bullish on Beltre, Sexson and Guillen.

These run figures aren't too bad when you consider the scoring environment of Safeco. I don't think the Mariners problem will be offense. It's too bad they'll feel compelled to throw a bunch of AB's at Vidro, and the bench is going to be ugly, but the M's should be able to score enough to win. The biggest problem I see is the lack of depth. An injury to one of the big guys would really kill the offense.

My methodology for the pitching portion of the projection will be different than the hitters' portion. Because OPS allowed doesn't correlate to runs allowed the same way as OPS does to runs scored, I'm going to take the projected ERA figures by the three forecasting systems, then create innings pitched projections for each pitcher, and finally, add on the average unearned run figure for American league teams from last season (60). I'm going to err on the side of caution on the innings pitched figures due to injury concerns, etc.

Starting Pitchers (ERA by CHONE, ZIPS, Marcel), and innings projected by me:

Felix Hernandez (3.34, 3.71, 3.95) 190
Jarrod Washburn (4.14, 4.45, 4.47) 180
Miguel Batista (4.58, 4.62, 4.47) 180
Horacio Ramirez (4.85, 5.13, 4.54) 140
Cha Seung Baek (4.58, 6.02, 4.38) 100
Jake Woods (4.54, 4.73, 4.61) 100

Wow, that's an incredibly mediocre rotation. If it wasn't for the presence of Felix, I'd be tempted to call it a contender for worst rotation in the league. Even with Felix, it won't be very good. The worst part is that they had to pay good money to acquire two of these guys, then had to trade Rafael Soriano to get another. Washburn and Batista are basically league average starters with durability, while Ramirez is a little worse than that and injury prone, and Baek and Woods are replacement-level. The bullpen should be pretty good once again, mainly due to the presence of JJ Putz, but the lack of Rafael Soriano and the uncertainty about Mark Lowe means the unit probably won't be as strong as last year (projected ERA by CHONE, ZiPS, Marcel), and innings projected by me:

JJ Putz (2.84, 2.88, 3.66) 70
George Sherill (3.24, 3.40, 4.31) 50
Chris Reitsma (4.35, can't find his ZiPS, 4.99) 60
Eric O'Flaherty (4.88, 4.35, 4.50) 50
Julio Mateo (4.11, 3.97, 4.35) 60
Sean Green (4.41, 4.79, 4.50) 50
Justin Lehr (4.46, 5.11, 4.88) 70
Sean White (no projection, 5.45, no projection) 50
Mark Lowe (4.61, 4.44, 4.10) 30

I used the ZiPS figure across the three systems for Sean White, and used a midpoint between CHONE and Marcel for Reitsma's ZiPS figure. I suspect these nine to get almost all the bullpen innings this season, barring injury. I guess I should throw in Jon Huber, too, but I'm going to be lazy. I don't really know what to expect with Lowe. Information on his injury has been incredibly hazy. I would guess he'll be able to pitch around the All-Star break. This only adds up to 1380 innings, and most teams pitch about 1440 innings, so I'm just going to tack on 60 innings of last years' league average ERA. This isn't exactly a scientific process, and there's a good chance that these innings could be at a much higher ERA, but they could also be lower, due to variance in performance in small samples. So here's the runs allowed projections, by system: CHONE: 740, ZiPS: 783, Marcel: 769.

Runs scored, runs allowed, by system:

CHONE: 771, 740
ZiPS: 757, 783
Marcel: 784, 769

With these numbers we can calculate expected Pythagorean wins:

CHONE: 84.3
ZiPS: 78.2
Marcel: 82.6

These numbers aren't nearly as bad as I thought they would be (that could be the Mariners 2007 slogan: "not nearly as bad as you thought"). Things working against the M's: Mike Hargrove, susceptibility to injury, lack of a bench, lack of minor league depth, inability to make in-season moves, inability to beat the A's, irrational Willie Bloomquist love. I think when all is said and done, the M's will be about .500, and be fairly boring in the process. I love this team!


Blogger Tangotiger said...

Each gain in 1 OPS point gives you 2.5 extra runs. So, if 750 OPS is 750 runs, then 800 OPS is 875 runs.

For example:

The Yanks' OPS was .824, and Tampa was .734. That 90 point difference I was make a quick guess at 225 run difference.

In reality it was 241 runs.


Phillies OPS was .794, Pirates .724, a difference of 70 OPS points, and a quick guess of 175 run difference.

In reality? 174.

6:36 AM

Blogger CSG said...

Oh, neat. I hadn't heard that before. Thanks for the input.

5:35 PM


Post a Comment

<< Home