FIRE JOE MORGAN: I Love These People


Where Bad Sports Journalism Came To Die

FJM has gone dark for the foreseeable future. Sorry folks. We may post once in a while, but it's pretty much over. You can still e-mail dak, Ken Tremendous, Junior, Matthew Murbles, or Coach.

Main / Archives / Merch / Glossary / Goodbye

Wednesday, October 17, 2007


I Love These People

You people, that is. It's only been a couple of hours, but of course we've been deluged with e-mails from good citizens who've bothered to do some work instead of just making fun of our friend Timothy like we typically do around here.

Here's John:
Using retrosheet data from 2000-2006 and a couple of quickie programs (written in ten minutes) to parse the data, we see the following result (best viewed in fixed font). This is for all of MLB combined; I didn't bother to separate out AL and NL:

1139 Leadoff HR => 294 2+ run innings, 25.81%
2705 Leadoff BB => 622 2+ run innings, 22.99%

1026 Leadoff HR => 247 2+ run innings, 24.07%
2238 Leadoff BB => 473 2+ run innings, 21.13%

1015 Leadoff HR => 229 2+ run innings, 22.56%
2296 Leadoff BB => 508 2+ run innings, 22.12%

1051 Leadoff HR => 252 2+ run innings, 23.97%
2261 Leadoff BB => 479 2+ run innings, 21.18%

1071 Leadoff HR => 246 2+ run innings, 22.96%
2321 Leadoff BB => 443 2+ run innings, 19.08%

979 Leadoff HR => 210 2+ run innings, 21.45%
2135 Leadoff BB => 462 2+ run innings, 21.63%

1069 Leadoff HR => 260 2+ run innings, 24.32%
2154 Leadoff BB => 490 2+ run innings, 22.75%

(Obviously, "2+ run innings" means innings in which 2 or more runs were scored).

Assuming that the retrosheet data is correct and that I don't suck at programming (iffy), this would seem to support what your correspondents are writing. Also note that Top Analyst Timmy (tm) would have been right in 2005 by the slimmest of margins.

And this from Joel:
Took a stab at the question with the data that I have in my database. It's just Cincinnati Reds games from 2001 to 2006, but it's still 973 games for reference and I checked both the Reds and their opponents in those games.

There were 1179 innings that had a lead-off walk. In those 1179 innings, at least one run score 515 times. At least 2 runs scored 328 times (27.8%). And there was a total of 1132 runs scored in those innings.

There were 589 inning with a lead-off home run. Obviously every inning had a run scored. There were 188 multi-run innings (31.9%) and a total of 956 runs scored in those innings.

I'm pretty sure my queries were right, but I only spot checked them. I'm sure you'll get a plethora of data dumps from nerds like myself. Hopefully we can draw some sort of a conclusion from it.

More analysis welcome, of course. I like to make jokes about dicks and poop and poop-covered dicks and dick-covered poop, but it's also nice to actually be right about something now and then. I think we're nearing a weird through the looking glass point where maybe Timmy wasn't so crazy to be surprised about the study he cited. Nah, still crazy.


posted by Junior  # 10:01 PM
I should have posted on this a long time ago, but then I left the country and just moved and blah blah blah. The point is, I got a few emails from people after my super snarky post about McCarver calling Stats. Inc. that showed he wasn't as crazy as I thought he was, and I never posted them. Pure neglect on my part.

Junior -- did we ever solve the mathematical accuracy of the Expected Runs Matrix for Bases Empty, No One Out when the leadoff guy homers? Is it (1+) .whatever? Or just .whatever?
Someone actually wrote in about this. I think it's 1+.whatever, because Tangotiger's system has to do with ... well, here:

I just thought I'd just chime in on the question about expected runs matrix when the lead off guy homers. So, if you assume that the inning is a Markov Chain
( that is the state of an inning is
completely described by the number of outs and what bases are occupied
(not really true) then the answer to your question of what is the expected number of runs given that the leadoff guy homers is simply
1 + expected number of runs with no outs and no one on.

This is because you can separate what happens in the past (the lead off guy homers) with what happens in the future, due to the Markov property.

Of course, this is not exactly true, since other things change if the pitcher gives up a homer, such as the likelihood that the pitcher will get pulled. That make this model just an estimate.

Love your blog,

P.S. Simple proof that it must be 1+whatever and not whatever: (Lets call "whatever" the variable A)
Assume that the answer is A, then what if the first B batters (with B > A) all hit homeruns. Then we still have no one on and no outs, but we have B runs scored, this would mean that the expected number of runs (A) is less than the
number of runs actually scored, which is a contradiction.

Markov Chain Tremendous
Post a Comment

<< Home


04.05   05.05   06.05   07.05   08.05   09.05   10.05   11.05   12.05   01.06   02.06   03.06   04.06   05.06   06.06   07.06   08.06   09.06   10.06   11.06   12.06   01.07   02.07   03.07   04.07   05.07   06.07   07.07   08.07   09.07   10.07   11.07   12.07   01.08   02.08   03.08   04.08   05.08   06.08   07.08   08.08   09.08   10.08   11.08  

This page is powered by Blogger. Isn't yours?