Looks like Jackaroo Dave predicted your response, more or less. Runs scored (hitters) and runs allowed (pitchers) is an example. I would like to run something less simple, like what Dave suggested. Not sure if I have the data.
Looks like Jackaroo Dave predicted your response, more or less. Runs scored (hitters) and runs allowed (pitchers) is an example. I would like to run something less simple, like what Dave suggested. Not sure if I have the data.
Jackaroo Dave: I have been reading this thread with some interest, but my eyes widened when I read the above passage. In the pure context of hitter performance evaluation, what is so awful about RC/PA for hitters, as long as the following conditions are in place:
1. The hitter's RC/PA is related to the prevailing average for such RC during his playing years, perhaps also refined as to League[s] in which he played each season?
2. The intent [context] of the evaluation is made clear: against contemporaries; against contemporaries at specific position[s]; comapred across generations or MLB history or a portion of the whole?
There are also variations of RC determination, some quite well relayed to actual run scoring within a team or league, some admittedly more down and dirty shortcuts. However, if the presenter makes clear his degree of specificity, where's the problem with RC/PA?
Just seems strange that one so ready to come to the defense of WAR [at least in general principal] should find notable blanket fault with RC/PA.
I think there's some confusion about nomenclature. "Runs produced" is a stat that has been around since the 1960s, maybe 50s. See my post on Pie Traynor in Bluesky's Third Base thread in the hall of fame forum for more background.
It's equal to Runs + RBI - HR. We had a discussion about it in the stats thread a while ago, concerning the wisdom of throwing out homers. I don't think anyone now, even its fans, and they are out here, would claim that it's a good comprehensive stat. Something to look at, but not the One True Great Stat. Its critics think it's mostly a function of choice of teammates.
Runs created, as you know, is something very different, a multiplicative stat that Bill James thought up as TB*OBA, and which has gone through refinements in various directions, some which are pretty much indistinguishable from linear weights based offense.
I pretty much agree with all you said. The qualifications you mentioned are the ones I tend to follow. I like the original version when I'm in a hurry. RC/AB, instead of PA, you can pretty much do in your head by multiplying SLG * OBA.
Indeed the first step toward finding out is to acknowledge you do not satisfactorily know already; so that no blight can so surely arrest all intellectual growth as the blight of cocksureness.--CS Peirce
JD: Thanks - got it.
I was recently thinking about how for pitchers, sabermetrics still prefers to use basically ERA, or runs allowed, with modifications to adjust for defense, and inherited runners etc. This is because it becomes apparent that using a pitcher's linear weight's projected runs allowed, or some other method based on batting events allowed varies in its run predictive ability because pitchers vary significantly with runners on base versus bases empty. ERA+ basically gets pitchers right, with about +/- 10% variation due to the defense playing behind him.
So with batters, how much do runs and RBI vary based on the batters around them (independent of ballpark effects on team offense)? And most of the effects of the rest of the lineup on a batter in RBI are due to him facing a different proportion of different situations. These proportions of situations also account for most of RBI variations due to batting order position.
So using a simplistic example, lets say
one batter has 300 plate appearances with the bases empty and 200 with men on base, and drives in 18 in the first split and 80 in the other for 98 RBI.
another batter has 250 plate appearances with the bases empty and 250 with men on base and drives in 15 in the first split and 100 in the other split for a total of 115 RBI.
They have the same split neutral rate of RBI. If we basically adjust a players RBI rate for each MOB or even out/base state to a standard rate of occurances we should get a very good estimate of their true relative RBI ability independent of where they hit or how good an offense they play for. The only variation would be the speed of the men on base.
Unfortunately we don't have all the scoring splits for a player scoring runs based on what comes after them, and RBI can't be the whole story because players can contribute to team scoring directly without driving in any runs. A guy might sacrifice an RBI for a walk that turns into a run for example.
anyway as a real life example of what I am talking about, Rickey Henderson is a good example because he batted leadoff, and pinch hit with the bases empty, and as a result 65.5% of his plate appearances occurred with the bases empty. If we give him Mike Schmidt's proportions of 48% men on base and 52% empty he would have driven in 1421 runs with his rates, and probably better if we adjust for RISP, or each out base state.
Brett, are you familiar with this study by pizzacutter? I found a reference to it searching for RBI posts here.Sorry about those dingbats. They were in the original.But, if collecting players who collect RBIs is important . . . .who are the guys who are particularly good at taking advantage of the situations presented to them.� Does such an ability exist?� Answer: yes.�
I started by taking my 2003-2006 data base and calculated the average number of RBIs for each base-out state available (that is, runners at 1st and 2nd with 1 out).� It’s much easier to knock in a runner already on third, and it’s easier to knock in a runner when there are less than two outs.� A fly ball to the outfield with a runner on third and less than two outs… yeah, you know what happens next.� So,� if you had 200 PA with nobody on and no one out, 50 with a runner at 1st and 1 out, etc., it’s easy enough to figure out how many RBI you should have had if you were an average hitter.� (Note: Nate Silver from Baseball Prospectus proposed�a similar approach�a while ago, without factoring in�outs.� It’s also�entirely possible that someone has already used my exact approach.� If you have… sorry.)
So, now I can tell you how many RBI’s above or below an average player’s expected output, given what you had the chance to do.� If the average player would have had 50 and you had 60 RBI, you my friend are an RBI machine to the tune of ten RBI above average.� To make things fair, I divided each player’s RBI above expectation number by his number of PA.� And those numbers were pretty reliable over the course of four years.� Restrict the sample size to those with more than 100 PA, and the intraclass correlation comes in at .50.� At 250 and above, it’s .60.� Pretty reliable stat.� Last year’s stats are a pretty good predictor of the next year’s stats.
I ran the 2007 numbers, and the leaders in the per PA stat�(min 250 PA) were A-Rod, Ryan Braun, Magglio Ordonez, Carlos Pena, and Ryan Howard.� Do those names have something in common?� In general, they hit a lot of home runs — although Ordonez hit “only” 28 HR last year.� In fact, I’ve shown previously that the correlation between HR totals (usually pretty consistent year-to-year) and RBI totals is .88.� But, they’re also pretty good hitters overall.)
The table of hitters with expected and actual RBI is here:
https://spreadsheets.google.com/pub?...yYNqPlO8BqLCZQ
The article itself is archived here:
http://statspeakmvn.wordpress.com/20...-wrong-places/
Didn't Pizza Cutter use to post here?
Indeed the first step toward finding out is to acknowledge you do not satisfactorily know already; so that no blight can so surely arrest all intellectual growth as the blight of cocksureness.--CS Peirce
And a recent one
http://www.hardballtimes.com/main/ar...tunity-of-rbi/
Using data from 2012, let’s look at a couple of leader boards. The list below displays the top 10 batters who drove in runs at the highest rates:
Code:Batter eRBI aRBI aeRatio Edwin Encarnacion 60.4 107 1.77 Josh Hamilton 72.9 127 1.74 David Ortiz 35.8 60 1.68 Evan Longoria 32.6 53 1.63 Miguel Cabrera 86.3 138 1.60 Jose Bautista 40.0 63 1.58 Ryan Braun 70.7 111 1.57 Giancarlo Stanton 55.3 84 1.52 Alfonso Soriano 71.4 108 1.15 Garret Jones 57.3 85 1.48
Bookmarks