Before I get started, Neifi Perez sucks. Let's get that out of the way.

In case you hadn't noticed, in Mark Mulder's last three outings he's thrown 22 innings, giving up only two runs. He also struck out only seven batters while also walking seven. At this point we're thinking about the post-season -- that's the only way I can maintain my composure when Hector Luna is batting 6th or when the Cardinals are getting beaten by Henry "176 / 202 / 282" Blanco -- so the question is whether the low strikeout/low run games will turn from gold into lead come October. On one side you have the party line that Mulder's a solid #3 starter and on the other side you have the Bill James ideas expressed well by L Boros, which says you gotta have some strikeouts. Like L Boros, I'm also not comfortable with the concept of Mark Mulder pitching a big game, so I ran some numbers to check on my stat-head prejudices.

Anyway, I first tried regressing post-season ERA on regular season ERA and K-rates. I did that by season and by career for individual pitchers, and in both cases all I got was noise. As it turns out, there have been 299 pitcher-seasons (e.g., Matt Morris-2002 and Matt Morris-2004) among post-season starters, so I decided to group post-season starters over the last ten years into three K-rate groups of 100, 99 and 100. Here's what I got:

Class    RegERA   RegK          PostERA  PostStarts PostIP
Low K     4.24   13.60% (5.3)    4.42       205      1181
Mid K     3.79   17.02% (6.4)    4.32       208      1234
HighK     3.36   23.41% (8.7)    3.51       247      1608 
TOTAL     3.77   18.26% (6.9)    4.03       660      4023

Note that these are overall post-season numbers for pitchers who started at least one post-season game. This means, for example, that I included Pedro's strange little relief stint against the Yankees in last year's ALCS in his 2004 post-season stats.

My K-rate groupings are based on strikeouts per batters faced ("RegK") during the regular season, as opposed to the traditional K per nine innings. However, I've included each group's regular season K-rates per-nine in parentheses for familiarity's sake. Mulder's K-rate stands at 13.3% in 2005, so he fits comfortably in that Low-K group. Each group experienced a decline in the post-season, which is what you'd expect since there aren't a lot of bad offenses in the post-season. High-K pitchers experience a bit less decline (note that Rick Ankiel-2000 is in that group) than the Low-K group, but not enough to get excited about. Actually it's the in-between guys who take the biggest hit.

Whether or not this is statistically significant, it does answer my question on Mulder. For now I'll bag my theory that low-K pitchers have wider bad team/good team splits than high-K starters.

Originally I had planned on just showing the Low and High K pitchers and leaving it that, but it was strange to me that the Mid Ks lost more value than anyone. So being the inquisitive guy that I am, I decided to see what happens when you sort by walks:

Class    RegERA   RegBB         PostERA  PostStarts PostIP 
Low BB    3.51    5.09% (1.9)    3.49       231      1458
Mid BB    3.91    7.62% (2.9)    4.42       229      1386
HighBB    3.91   10.07% (3.9)    4.23       200      1179  
TOTAL     3.77    7.48% (2.8)    4.03       660      4023

Well, that's interesting. Low BB starters don't lose lot a value in the post-season, or at least they haven't in the past ten years. Once again the middle group took a bigger hit than the "bad" group. One last table, this one by K/BB:

Class    RegERA   RegK/BB       PostERA  PostStarts PostIP
Low K/BB  4.22     1.59          4.46       206      1197
Mid K/BB  3.77     2.31          4.57       211      1230
HighK/BB  3.38     3.90          3.28       243      1596 
TOTAL     3.77     2.44          4.03       660      4023

Again the guys in the middle have been dropping the most, to the point that they've been worse than the low K/BBers in the post-season. There is probably some sort of selection bias at work here or maybe I need to control for the opponent. Whatever the case, there isn't anything in the aggregate data to suggest Mulder is particularly vulnerable to collapse during the post-season. My gut still feels otherwise, but the gut feelings of stat-heads aren't especially useful.

[UPDATE: Now included post-season IP. And Mulder looked terrible last night.]