In lieu of reading a 49-page pdf, might I offer the following graph of Roger Clemens’ career performance. The blue points are Clemen’s seasonal runs above replacement totals, adjusted to a league-average 4.50 ERA environment. I used 5.75 as replacement-level ERA for his years in the AL and 5.25 for his NL seasons because of the talent difference. One reason to use runs above replacement instead of simply ERA is that it’s harder to maintain a low ERA over more innings.
The red points represent a smoothing of the original data, weighting the current season 40%, the neighboring seasons 20% each, and the seasons two years away at 10% each. The black curve is the best fit second-degree polynomial.
The black curve of best fit shows a reasonable career path, peaking at age 30. Clemens posted many excellent ERAs in his late thirties and early forties, but his lower innings pitched totals made those seasons less valuable. Clemens only had one truly excellent season out of nine after his back-to-back Cy Young seasons with Toronto. And his second Cy Young season is remarkably similar to his last season with Boston. Yes, his ERA was lower, but the league ERA much lower as well. Sure, there haven’t been many pitchers to be so successful for so long, but there have been some and there haven’t been many pitchers to be so successful for their first ten years in the majors, either.
My point isn’t to defend Clemens — maybe he used PEDs and they had zero effect, or maybe they slowed his descent from star to has-been — but his career path isn’t crazy considering how great he came out of the gate. And this post saves you from reading the whole Clemens Report, which says exactly the same thing.
Popularity: 7% [?]
Share This
Sky is a baseball fan and racket sport afficianado living in upstate NY. His favorite color is orange and is just about ready to give up on his life-long dream to become the next Magnus ver Magnuson (World's Strongest Man). His favorite baseball teams are the Yankees and Red Sox, proving that there's hope in the Middle East.
February 4th, 2008 at 1:40 pm
I dunno - the curve isn’t as outrageous as Barry’s equivalent, sure. But it’s pretty darn flat, which ironically sticks out to me. What does Glavine’s or Maddux’ look like?
If Clemens burned out in 1997, I think you’d end up with a much more typical curve. 1997 and 1998 are just so utterly out of place given 1993, 1994, 1995, and 1996.
February 4th, 2008 at 6:47 pm
Yes, you’re right Mike — we definitely need to compare Clemens’ graph to those of other people. The Report made an attempt, but only used two other pitchers: Schilling and Randy Johnson. I’ll try to get around to that.
February 4th, 2008 at 7:43 pm
Schilling and Randy are odd cases, too, because neither of them were really all-stars at the beginning of their careers, and neither of them really got good until they were 28 or so. I mention Glavine and Maddux because I think, like Clemens, they had been very good pitchers from just about the very beginning and have pitched to relatively advanced ages. The down side of them is that they aren’t nearly the body type of Clemens, so it’s hard to assume they’d age the same. I guess actually maybe Schilling might be the best comp to Clemens, given his size and in addition the development of the splitter at some midpoint in each of their careers.
By the way, one more thing - when you say “second-degree polynomial”, I’m guessing you mean an equation where there’s something to the 2nd power? (forgive me if I’m way off, it’s been at least 8 years since I’ve taken a math class). If I remember correctly, wouldn’t an equation to the 4th power give a curve with two humps? If so, that would better approximate Clemens’ career, though I guess you’re sort of using knowledge of the stats to decide to use that kind of a line, so maybe it’s biased in that respect.
February 5th, 2008 at 7:22 pm
I saw something a few years ago about Sammy Sosa. Using career/age trajectories and league-wide homer rates, they showed his performance wasn’t too far from what you’d expect based on those factors. Take that for what it’s worth.
February 5th, 2008 at 8:01 pm
Mike, yes, a second-degree polynomial has one hump (or valley). A third-degree could have one of each, and a fourth-degree could two humps and one valley. There’s no reason to force a second-degree model just because it’s the traditional career path.
February 18th, 2008 at 1:40 am
I’m not really sure you can come to the conclusion that 96 was better then 98. Roger’s 96 season pretty much relied on his last ten starts - The stat difference is astonishing actually.
February 18th, 2008 at 10:18 am
Dave — ‘98 is better than ‘96 according to the graph, even accounting for the fact that the league ERA in ‘98 was half a run lower.