July 1, 2006

Innings Per Start: Pedro/Schilling

In comments to my Pedro post, Jack said "the inability [of Pedro] to finish games has to be factored in" when discussing his all-time greatness. Part of that way of thinking is that the longer an ace starter lasts, the more he prevents a lesser arm from coming into the game. He mentioned Curt Schilling, Roger Clemens and Sandy Koufax as more durable pitchers.

Baseball is not played the same way in 2006 as it was in the 1960s and the lack of complete games should not be held against Pedro. Even if Martinez could go nine innings every time out, no manager or front office would allow it.

Here are the starts, innings pitched, average innings per start and ERA+ for Martinez and Schiling. Pedro's stats are his entire Red Sox career (1998-2004) and the 2005 season with the Mets. Curt's stats are with Philadelphia (1993-2000), Arizona (2000-2003) and Boston (2004).

Martinez Schilling

26 33 233.2 7.07 160 34 235.1 6.92 100
27 29 213.1 7.35 245 13 82.1 6.33 96
28 29 217.0 7.48 285 17 116.0 6.82 121
29 18 116.2 6.47 189 26 183.1 7.05 138
30 30 199.1 6.64 196 35 254.1 7.26 143
31 29 186.2 6.43 212 35 268.2 7.67 134
32 33 217.0 6.57 125 24 180.1 7.51 130
33 31 217.0 7.00 148 29 210.1 7.25 124
34 35 256.2 7.33 154
35 35 259.1 7.40 136
36 24 168 7.00 159
37 32 226.1 7.07 150
In their respective age 26, 27 and 28 seasons, Martinez averaged more innings per start than Schilling. Since then, Pedro has pitched fewer innings per start.

But it doesn't seem like that big a difference. At age 31 and 32, Schilling averaged almost 1.1 more innings per start. That sounds significant -- is it? -- but how many hits and runs did Schilling allow (if any) while recording those additional four outs. Relying only on innings pitched doesn't tell us what we want to know. And assuming effectiveness from those totals would be a mistake.

For example, what if Pedro pitched 6 shutout innings and Schilling allowed 4 runs in 7 innings? Schilling pitched deeper into in the game, but was not as effective. Judging by their respective ERA+s, something similar to this example must have happened more than a few times.

Schilling's best ERA+ seasons (2003 and 2001) are superior only than Pedro's final year in Boston (2004) when he turned in an un-Pedro 3.90 ERA. In every other one of his Boston seasons (save perhaps the first, 1998), his performance has dwarfed the career bests of Schilling.

That doesn't mean Schilling isn't a damn good pitcher (his outings in June, mostly against other top pitchers, were superb), but it makes me question whether not sticking around for those extra outs (in their age 33 seasons, the average is less than one out per start) should be considered a strike against Martinez.


Earl said...

On top of that, Schilling got to face pitchers, and Pedro did not, in every year you listed except the last. That in itself could account for much of the difference. (There's a jump in Pedro's IP/GS when he went to the NL, but that may not be statistically significant.)

allan said...

Oh, the innings are in thirds (.1 is one-third, .2 is two-thirds) and the averages are based on 100 (so 7.33 is 7.1 innings). But you probably figured that out.

(Some stat sites list innings pitched as .3 and .7., btw)

Jack Marshall said...

Redsock: that's good stuff.

I think the crucial result of staying in the game less time is the increased chance that a lesser pitcher will blow the game. That happened to Pedro last year, and is happening this season; it's the main reason he's on track to win only 15 games despite having a good team behind him. The fascinating question would be where that extra out or two generally occurs. If it means Pedro finishes the 6th but Schilling et al. go one out into the 7th and have to be replaced, you're right: can't see any advantage. But if the extra outs mean that Schilling is generally finishing innings while Pedro has to leave with men on base, that one out is very significant.

You know, of course, that I would never argue that Pedro at his best isn't an order of magnitude better than Schilling at his best; only that when you are asked to pay 12-13 mil for a pitcher, the difference between 6.3 and 8 is significant.

laura k said...

Good analysis - and explained well. Thanks.