THE CRANE POOL FORUM thecranepool.net (.com)


Forum Home

Master Index of Archived Threads


SABR

Johnny Dickshot
Aug 03 2005 10:50 PM

I'm leaving tommorrow for SABR in Toronto (and a Jays-MFYs game at Skydome). Will report back all the poop or perhaps see you there.

metirish
Aug 03 2005 11:05 PM

Have fun, what goes on at a SABR convention, I mean it's not like a regular baseball convention, is it?,would a Bill James or a Billy Beane get treated like William Shatner or Leonard Nimoy at a Star Trek convention?

smg58
Aug 04 2005 12:04 AM

Have fun. You catching Halliday?

Rockin' Doc
Aug 04 2005 09:40 AM
No offense to Johnny D, but...

Even though he's in the midst of a slump, I think the Blue Jays would still prefer that Greg Zahn catch Halliday.

MFS62
Aug 04 2005 09:52 AM

If you haven't left already, check your PMs Johnny.

Later

Johnny Dickshot
Aug 08 2005 08:19 AM

This was fun. Got to meet some good people, hear some good ideas, get inspired to do stuff, and saw a lousy baseball game (Friday, when the MFYs beat a listless Jays team at Rogers Centre).

Some random stuff I learned:
* The 62 Mets were among the "unluckiest" teams since 1960 (as measured by a variety of stuff -- career years, pythag, etc) and when such luck factors are stripped away, the 1965 Mets tie the '77 Blue Jays as the worst team in the last 45 years (they "should have" won 54 games, not 65, the speaker said)

* Quality Start is a quality stat -- just one-third of all games started meet the specificications (6 IP, 3 ER or less), and the ERA of 4.5 that produces drops to below 4 in just one more inning. Teams whose starters get a QS win 67% of the time. This came up as Pete Palmer and Gary Gillette introduced a stat called "Quality Relief" which is set to be published in an article on ESPN this week I think. This stat uses park-adjusted run expectation figures (based on bases/outs situations) and credits and/or debits relievers for the situation they leave for the next guy.

This stat ideally would replace the worthless "hold" in measuring middle relief and expose guys like Scott Eyre last year who "save" runs based on his arrival in games -- but lose them by leaving a bad situation for the next guy to pick up.

They said the average reliever would receive a QR credit around 69% of the time, and that the very best relievers get a QR 90% of the time.

* Amazin' -- Leo Mazzone has a statistically significant effect on improving pitchers -- .62 off the ERA or "the Coors Field effect in reverse" -- he works with, based on an examination of some 60 or 70 pitchers with significant sample sizes before, during and after their work with him. Of the group, only Jason Schmidt, Odalis Perez, Jason Marquis and Dennis Martinez underperformed in ERA, K/BB, K/IP, BB/IP under Mazzone.

* Pitchers who throw fastballs age better in this era than they did 20 years ago. Also, hitters are peaking later than they used to -- suggesting (to me) that similarity scores across eras might be better examined by service time than by age. The current peak standard age for hitters is approaching 30 today rather than the 27 Bill James first revealed years ago. James himself sat near me for this presenatation.

* The controversial DIPS stuff that came out a few years ago suggesting pitchers have little effect on the result of balls in play was broken down and argued that because only MLB pitchers were used in the study, that it was biased to show no effect in that the ability to prevent hits on BiP is indeed pre-selected for (iow, pitchers who cannot prevent hits don't become MLB pitchers). Further, groups of pitchers (for instance, those who get numerous Ks), do perform better on BABIP than other groups.

* Skydome is a nice looking, clean park but for the turf, the fact that the bullpens are mostly impossible to see into, and there's no ball-strike and OoT scoreboard visible from the baselines without looking to the outfield. We were seated in the 11th row between 3rd base and LF -- good seats -- but watching the game and the scoreboard required moving my neck where I'd prefer just to move my eyes.

* Other speakers included Blue Jays prez. Paul Godfrey, former MLB reliever Paul Spoljaric (a Canadian), the former Expos play-by-play guy whose name I can't remember.

smg58
Aug 08 2005 10:08 AM

Could you please explain the DIPS stuff in a bit more detail? I'm still trying to figure out how strikeouts could be irrelevant for hitters but vital for pitchers, and how what happens when a ball is put in play depends heavily on the hitter but not at all on the pitcher.

Frayed Knot
Aug 08 2005 10:22 AM

]Quality Start is a quality stat -- just one-third of all games started meet the specificications (6 IP, 3 ER or less), and the ERA of 4.5 that produces drops to below 4 in just one more inning. Teams whose starters get a QS win 67% of the time.


I never understood why the 'Quality Start' takes the beating it does among fans & media. Sure it's not a perfect stat (like there's a stat out there which is!) but it's clear that (like stealing bases or getting the leadoff man on) a larger pct is better than a smaller one and that even reaching the minimum requirements represents a decent, even if hardly spectacular, outcome.
And even the 4.50 ERA figure - something which gets quoted extensively whenever the stat is being put down - represents the statistical maximum that could possibly happen if and only if the pitcher just reaches the min-IPs/max-RAs level each and every time; which is sort of like pointing out that a 70 game hitting streak could be accomplished by a .250 hitter.

metsmarathon
Aug 08 2005 10:38 AM

or that you could convert all your saves with an 18.00 ERA (giving up two runs, maintaining a three run lead)

Johnny Dickshot
Aug 08 2005 11:05 AM

The initial DiPS findings in 2001 suggested that since balls in play (everything but HRs, Ks and BBs) that turn into hits or outs are not consistent or predictable from year to year that pitchers had little control over what happens once the bat meets the ball.

This theory, which rocked everyone’s world initially, has been backed off on gradually, even by McCracken who first proposed it. Still, the feeling is that McCracken had made a valid point in examining the question of “what things do a pitcher, a defense, or a hitter have control of?”

The speaker this year, Mike Emeigh, mentioned that one flaw of the 2001 study was that by only using MLB pitchers with significant sample sizes it was biased to show that result since pitchers who do not have the ability to prevent hits don’t accrue significant sample sizes. IOW, the ability to prevent hits on balls in play is a skill that’s selected for; and one that most all MLB pitchers have.

The thrust of Emeigh’s talk was examining batting everage on balls in play among various groups of extreme pitchers – extreme groundball guys; extreme flyball guys; high-strikeout pitchers, etc. He wanted to start to see whether a “type” of pitcher was better at preventing hits than another. (I don’t want to go much further or I’ll butcher what he said because I can’t do the math and didn’t take too many notes; He said he would publish his findings on the net at some point and I’ll link when he does)

This was not in the talk but generally it’s believed that K’s “help” pitchers more than hurt batters because pitchers who have a number of K’s generally excel in other areas (High strikeouts correlate with other “good” stuff like low ERAs, better BABiPs, etc). So it’s not the strikeout itself; it’s the “strikeout pitcher.”

Johnny Dickshot
Aug 08 2005 11:44 AM

I forgot to mention this but there was a complicated but interesting poster presentation that used space-age mathematics to examine the degree to which teams may effect their players. The idea was: If you’re studying a topic like education, you don’t just look at student records but also at the effect of the school; the same would apply to study of other organizations, businesses or, teams.

I’ll never understand how he did it, but what the guy found was that players on the same team tend to be more similar to one another than to players on other teams in certain performance criteria (I forget which category it was, but the highest rank was 27% more similar). Other categories showed little correlation. Like a lot of the stuff presented, findings suggest further examination is needed.