March 04, 2008
Rotation Evaluation, New York Mets
The series looking at team pitching rotations using the Marcel the Monkey projections continues with the New York Mets. Their starters posted 4.40 ERA in 2007, fifth in the National League.
Note that in figuring ERAs, I'm using Marcel's mIP and mER columns. The ERA listed in the spreadsheet uses the average of mER and bsrER. I wanted to avoid posting 1/2 runs in the table. Pitcher order is taken from the CBSSportsline depth chart.
Johan Santana
Photo: Icon SMI
Marcel predictions for the New York Mets top five starters for 2008.
| Starter | Innings | ER | ERA |
| Johan Santana | 193 | 71 | 3.31 |
| Pedro Martinez | 87 | 38 | 3.93 |
| John Maine | 163 | 73 | 4.03 |
| Oliver Perez | 160 | 80 | 4.50 |
| Orlando Hernandez | 148 | 74 | 4.50 |
| Totals | 751 | 336 | 4.03 |
The addition of Johan Santana to the rotation gives the Mets a boost. The rotation threw 940 1/3 innings last year, and a 0.3 plus reduction in earned runs allowed should be good for three wins. That would have been enough to put the Mets in the playoffs last year.
Correction: Looks like I used Livan's projection instead of Orlando's.
Even with a low prediction for innings from Pedro, the Mets are in decent shape to average six innings per start. The Mets are going to need help to get to six innings per start. Martinez probably gives them more than 87 innings, but those may just trade off from the injured Orlando Hernandez's totals. Given their abilities to prevent runs, the New York will be happy with the tradeoff. Pelfry is predicted to post a 4.83 ERA, so trading Mike for Orlando doesn't make much of a difference. Still, he'll be a perfectly acceptable spot starter.
New York's starting staff, especially if Pedro is indeed healthy, should be among the best in the NL this season. Remember, Marcel doesn't take into account parks or leagues. Santana's predicted ERA is likely on the high end of his range.
Previous posts in this series:
Correction: I meant Orlando, not Livan Hernandez. Sorry about the mixup. My head doesn't seem to be on straight this morning.
The Baseball Musings pledge drive continues through March. Please consider making a donation.
I believe you mean El Duque instead of Livan, whom you mention twice in the analysis.
Why does Marcel never predict anyone to throw 200 innings? So far it only has Webb and Sabathia throwing 200 and only barely (202 and 200 respectively). Over the past 3 season there has been an average of 44 pitchers in the majors throwing 200 or more innings. It seems the system is ultra conservative on the number of innings a pitcher is going to throw.
I'm with you Tyler. Most of these projections seem extremely conservative but at the same time inconsistent. With el duque's injuries last year and the fact that he hasn't thrown 180 innings in eight seasons, I fail to see how he'll even sniff 120 let alone 184 (unless he grabbed livan's projection by accident). I just find it hard to believe that he'll throw more innings than three other starters ahead of him.
Ok, I see now that el duque's actually projected for 148 IP, which makes more sense but is still on the high side I think.
I wonder what the other prediction algorithms say about innings pitched (ZIPS, PECOTA, etc.).
The way I've always (which is to say, the last two years or so) looked at Marcel projections is like an over/under. We're talking about the simplest way to project players with any kind of reliability... about half of these guys will exceed those projections and about half will fall short.
So think about guys at the top end of the IP spectrum. You can't reasonably project them for 220-230 IP, because in this day and age, there's pretty much no room for them to exceed that projection. And guys who fall short of their projections probably won't do so because of a change in talent level, where suddenly they can't pitch well enough to stay in the rotation, or to avoid the early hook. They'll probably do it because of injury. Between the higher possibility of season-ending injury for pitchers, and the effect of injury on those who return, they're especially tough on pitchers' playing time.
So if you look at the top, say, 20 pitchers' projections as a group, yeah, no single projection may look observationally right. But taken as a group, with the likelihood that the underperformers drastically fall short, and the outperformers have a low ceiling, the lot of them are going to come out pretty close when all is said and done.
Predicting any one player is a crapshoot, no matter who's doing it, how they're publishing it, or what they're charging -- projections show their worth by correlating on the group level. That's why you're not going to see the Monkey with a dozen guys throwing 220.
If Hernandez really pitches ca. 150 innings -- call it 25-30 starts -- can the Mets finish third in the division?