General Red Sox General Yankees

The 5th Starter

Interesting piece in Fangraphs recently, especially given our recent discussions on pitching depth. It links to a 2008 piece done on Lookout Landing about how good starters are by spot in the rotation:

  • #1 STARTERS: 130 tRA+
  • #2 STARTERS: 112 tRA+
  • #3 STARTERS: 100 tRA+
  • #4 STARTERS: 91 tRA+
  • #5 STARTERS: 76 tRA+

What is tRA+? It's a linear-weights pitching metric that assigns run values to every event under a pitcher's control, adjusting it for ballpark and defense, then adjusting it by league and placing it under the familiar 100=league average scale that we expect for stats with a + after them. Put another way, it's a pitching version of wOBA+, or more basically, it's billed as a more accurate ERA+.

So, in 2009, the Red Sox' principal starters looked like this:

  1. Jon Lester, 131 tRA+
  2. Josh Beckett, 116 tRA+
  3. Tim Wakefield, 103 tRA+
  4. Brad Penny, 98 tRA+
  5. John Smoltz, 94 tRA+
  6. Daisuke Matsuzaka, 90 tRA+
  7. Clay Buchholz, 89 tRA+

Smoltz's performance in Boston has always been a little mystifying because the defense-independent numbers always showed him as being far better than his results. *shrug* In any case, this shows that despite the injuries and/or ineffectiveness of everyone beyond their top two starters, the Red Sox still were above average at every slot in the rotation in 2009. And even their seventh-best starter, by this statistic, was better than the league-average fifth starter. Even if we say Smoltz was a lot worse than this stat's ability to describe (no objections from me), the Sox are still above average in the fourth/fifth/sixth spots in the rotation.

The Sox also got non-trivial contributions from Justin Masterson and Paul Byrd, who combined for 66 innings as starters with a 121 tRA+ and 101 tRA+, respectively.

Now, the definition of #5 Starter in the Lookout Landing post is the combined performance of the worst starters in the league, so we would actually combine the performances of Penny and Smoltz (or Matsuzaka) for our fourth starter and the performances of Buchholz and Smoltsuzaka for our fifth (and do something with Byrd and Masterson, maybe combine them with Wakefield since he was hurt, too). Either way, we'd come out above average because all of them are above average for a fifth starter and no worse than two points below average for a fourth.

Here are the Yankees in 2009:

  1. CC Sabathia, 123 tRA+
  2. Andy Pettitte, 103 tRA+
  3. A.J. Burnett, 100 tRA+
  4. Joba Chamberlain, 92 tRA+
  5. Sergio Mitre, 77 tRA+
  6. Chien-Ming Wang, 75 tRA+

I'm not sure any Yankee fans have been arguing that the back end of their rotation was all that terrific — or that even the front end was overwhelmingly dominant, so this probably isn't much of a surprise. The Sox' rotation last year looks deeper and got better results, and that is borne out by the Sox' 108 ERA+ last year, to use a more traditional stat, and the Yankees' 100. 

But the point of all this isn't to say, "Ooooh, the Sox' pitching was better than the Yanks' pitching last year." The point is to note that this data makes it clear that we are overrating significantly how good we should expect the fifth spot to be should one of our starters go down for an extended period of time. Sergio Mitre was a league-average fifth starter last season. Brad Penny was comfortably better than the average fifth starter. Neither team was particularly happy — at least, we as fans were not — with these men taking the mound, given how they performed for the most part, yet neither team really struggled to reach the playoffs. Now we know why.

The site where I found these tRA+ numbers and definition doesn't have career totals, for some reason. So for fun here are the last three years of tRA+ for the projected starters for our respective teams. (They do split out the stats for starting and relief for those who did both in one season, which helps in providing starting numbers for Joba and Hughes.) It's clear that tRA+ likes Vasquez a lot and wasn't such a fan of Buchholz's seeming resurgence last year. It also dings Beckett more than I would have thought while showing nearly as significant a decline for Sabathia for the past three years. Interesting.

The Sox:

  • Jon Lester: 74, 120, 131
  • Josh Beckett: 141, 127, 116
  • John Lackey: 124, 94, 118
  • Clay Buchholz: 129, 93, 89
  • Daisuke Matsuzaka: 109, 108, 90
  • Tim Wakefield: 96, 100, 103

And the Yanks:

  • CC Sabathia: 141, 139, 123
  • Javier Vasquez: 129, 128, 131
  • A.J. Burnett: 115, 124, 100
  • Andy Pettitte: 107, 108, 103
  • Joba Chamberlain: 131, 92
  • Phil Hughes: 94, 78, 80

6 replies on “The 5th Starter”

This is a very good argument, Paul. While I don’t consider tRA+ to be the end-all, be-all of pitching statistics (and I see that you don’t, either), it’s good enough to prove your point. Some things that were coloring my argument:
1) The Sox pitchers performed worse than their peripheral stats would suggest. I imagine this was mostly due to their below-average team defense, although I think that a pitcher like Paul Byrd performed exactly how he should have. Defense-independent pitching statistics still give guys who don’t walk anyone but give up line drive after line drive far too much credit. Smoltz is another example of this, I think we could all see he was simply grooving meatballs in the AL East.
2) A couple other teams were regularly trotting out pitchers even in the number 4 slot of their rotations that were even worse than Byrd/Tazawa/Smoltz. I underestimated how bad those marginal pitchers on marginal teams were.
3) The proclamations of ‘historic’ pitching depth raised even my skeptical expectations of the Sox’s performance. I didn’t think they’d perform like, say, the 2008 Jays (who quietly had one of the most fantastic pitching staffs of the decade, especially for a team that only got more than 30 starts out of its top two guys), but I thought they could reasonably get under 700 RA. When they shot past that, I went overboard in describing their depth as ‘non-existent’. It existed, clearly, but it wasn’t anywhere even close to the level everyone was touting it as.
The Sox had a good pitching staff in 2009. It had depth in terms of the rest of the league. I’m pretty sure it always does (non-2006 division). But in terms of our expectations, and in terms of past pitching staffs, it was not as good, or as deep, as what we’ve gotten from the Sox over the past couple of years.
So…how many more days until the 2010 season starts?

Someone in the Lookout Landing comments described tRA+ as a “skill statistic” as opposed to a “result statistic” like ERA+. That helps explain some of these anomalies (and probably also explains why it fails to adequately capture the suckage of Smoltz and Wang last season).
Daisuke’s results in 2008 radically outperformed his peripherals, as we all know. His 2007 numbers were much more in line with what you’d expect. tRA+ just smooths that out. You go from Daisuke having a solid if inconsistent rookie season to basically matching that (though in a different, more consistently maddening way) whereas ERA+ captures that he was bizarrely effective in spite of his sky-high walk numbers.
Like Andrew notes, stats like these that focus on peripherals and essentially try to adjust for luck by looking at line drive percentages, BABIP, etc., will overrate pitchers who are so bad they just give up line drives all the time. This is what Smoltz and Wang were doing, and tRA+ paints them as better than most of us are willing to say they were (even Wang at 25 percent worse than league average is probably given too much credit by tRA+).
Anyway, it’s data like this that have me excited about 2010. The Sox’ staff is entering the season in a much stronger state than it was in 2009, the defense is inarguably much improved, and the offense does not project to take much, if any, of a step down. And that’s after a 95-win season! It’s very exciting. April 4 can’t get here fast enough!

Paul, it also reinforces the perception asserted earlier this week that the Sox have pitchers at the bottom of the rotation who could, quite legitimately, be those #3 guys on other teams, assuming a few things happen – Buck improves, Dice-K returns to form, Wake pitches a good portion of the season somewhat healthily (this last one is the longest shot, as far as I am concerned). All of those guys could (not WILL) tip that 100 on tRA+. It wasn’t that outlandish a statement, at least not in terms of this statistic.

Leave a Reply