This is no knock on Cabrera or the Triple Crown. Both are impressive. It’s just that whether Cabrera wins the Triple Crown or not has absolutely no bearing on the MVP. Look at it this way: Josh Hamilton hit his 43rd home run last night, breaking a tie for the home run lead with Cabrera. Does a home run by a Texas Ranger (or, to flip it, blurry vision by that same Ranger that kept him out of the lineup over the weekend) decide whether a Tiger or an Angel wins the MVP? Is Cabrera less worthy because Hamilton hit a home run?
The Triple Crown is cool in the way that typewriters, rotary phones or vinyl records are cool: There’s an undeniable acquired history there, and it’s been so long since most people have seen one that now there is great novelty to it. But the truth in all cases is that we know better now.
The Triple Crown is an interesting relic from baseball cards and Sunday morning newspapers, which used to be the way people found out batting averages of major league players. Actually, it wasn’t until the 1940s that the Triple Crown held much resonance in baseball at all. RBIs did not become an official statistic until 1920. The home run didn’t become a regular staple of the game until Babe Ruth popularized it in the late 1920s. In the days of Cobb, people thought batting average, hits and runs were the most important offensive categories.
...The bottom line is that the AL MVP should be . . . too close to call right now. Trout did enter the month with a huge gap in value, but Cabrera has closed it considerably. If you believe in impacting games in more ways more often than anybody else, Trout is your MVP. If you believe in the greater volume of hitting stats, particularly down the stretch, Cabrera is your man. And if you believe in the Triple Crown defining the MVP, sit down with a big hunk of blueberry pie and hope the idea passes.
Repoz
Posted: September 26, 2012 at 11:08 AM |
50 comment(s)
Login to Bookmark
Tags:
awards
Reader Comments and Retorts
Go to end of page
Statements posted here are those of our readers and do not represent the BaseballThinkFactory. Names are provided by the poster and are not verified. We ask that posters follow our submission policy. Please report any inappropriate comments.
1. Mom makes botox doctors furious Posted: September 26, 2012 at 12:35 PM (#4246247)it didn't really take that long, Tom
I do have to say it's surprising that RBIs are SO irrelevant nowadays that the Triple Crown isn't a bigger deal. I don't remember a serious Triple Crown chase in my period of baseball fandom (1998-present), so had expected it to be something everyone was hoping for with bated breath.
I think we've crossed over to the point where RBIs are too undervalued, at least around here.
Projecting player performance/establishing "true talent" is not the purpose of baseball. The goal is to win games.
RBIs are an important record of who helped score runs.
The first season I ever followed was 1967, and there had just been a Triple Crown in '66 as well, so I got the initial impression that it was sure to happen again soon. One of those things I'm still waiting for, like that Academy Award Peter O'Toole is fixing to win any year now.
Definitely, Runs are just as important.
I'm just reacting against the tendency to prefer theoretical value over actual value delivered.
Clutch hitting is important, even if not repeatable.
This. Actually, it's not even a purpose of baseball. It's at most a baseball-related activity or something that might be able to be accomplished from the data generated by baseball -- much as fantasy or rotisserie baseball are.
Clutch hitting is important, even if not repeatable.
This, too.
I'm open to the argument that the difference between predicted rbi and actual rbi has value (though you'd want to use a sophisticated rbi estimator -- Tom Ruane has a good one) but rbi in itself has essentially no value if you know a player's SLG. A player's rbi total is an extremely predictable product of opportunity (which a player has no control of) and power (most simply represented by SLG)
NOTE: This isn't being asked with any misunderstanding of the effort to tease out "true talent" and repeatability. It's being asked knowing all that, yet entirely on its merits.
Miguel Cabrera is 2nd in runs scored, so it isn't like he is some slouch there either.
You are conflating different values. Team RBI record value to the team winning games, but an individual's RBI doesn't really record an individual's value as well as we can possibly do. If you just want to say that Cabrera was in the middle of a lot of run scoring that also involved good hitters like Jackson, Dirks and Fielder, then fine, but you haven't done much to isolate Cabrera's value.
15 - Saying Trout should be getting more credit for his counting stats should not be taken as a denigration of Cabrera in any way. This is a kind of, make that really annoying thing that has really taken off recently. Saying player A should get a lot of credit for something has nothing to do with player B.
If you're making some form of clutch argument then what you need to do is account for his opportunities (not merely his at bats with runners on -- though this works reasonably well). My (quick and dirty) rbi estimator predicts about 127 RBI for a player with a BA of .329 a SLG of .609 and 284 AB with runners on base.
The formula for estimated RBI is ABROB*(SLG*1.09-BA*.66) (wouldn't be surprised if it works less well these days -- it's approaching 2 decades old)
It also predicts 66 RBI for Trout. He's +12. While Cabrera has significantly more power, he's also got 69% more at bats with runners on and that's a big part of his RBI edge.
EDIT: A reasonable clutch credit is (Actual RBI - Predicted RBI) / 3 (But you'd really want to take a good look at the distribution of baserunners)
Now I know that the primary source of error in the estimate is the distribution of baserunners. Which is why if you want real precision you want to go with Tom's (formula available in his paper at Retrosheet)
RBI should track better to total bases than SLG. Bonds and Mantle had extraordinarily high SLG, but lower RBI totals because of their unusual BB rates. Brainless MSM criticism of especially those two players -- e.g. "he's not as good a run-producer as [player X]" -- is one of the origins of the backlash agains RBI among BTF denizens.
Look, this is pretty simple. RBI are good for players to get. More RBI are better than less RBI. Hence, value.
Telling us that it doesn't record "an individual's value as well as we can possibly do" is both missing the point entirely and expressing a stathead truism that we have all known for decades. And you could make the same comment about any number of counting stats.
My argument is far simpler and more fundamental than this. It's that RBIs measure valuable occurrences in a baseball game.(*) No more, no less.
Our quibble is with the quibbling with this seemingly straightforward observation.
(*) Miguel Cabrera comes up with a runner on 3rd and two outs. Instead of making an out, he gets a single and a run scores. The fact that his single caused a run to score -- as opposed to not causing a run to score -- should be noted and recorded for posterity. That's what the RBI does.
One example that I've used with some success in showing how strange RBI can be is to look at Hack Wilson and the 1930 Cubs.
Assume Rogers Hornsby was healthy and played as well as he had in 1929. He'd have batted 3rd and Cuyler would have moved to bat 6th (you might think they'd have gone English, Cuyler, Hornsby, Wilson. Not a chance. Cuyler was with the Cubs because of a bitter battle on the Pirates about batting him second. No chance McCarthy goes there -- as you can see from the 1929 batting orders) Now the team would obviously have scored a lot more runs. And yet Wilson would have driven in fewer. Yes, Hornsby had a higher OBP than Cuyler but it's almost perfectly explained by the difference in home run power.
So Hornsby is going to drive in more runs than Cuyler and be on base about the same number of times (Cuyler was the fastest guy in the league and the speed difference might also cost a few RBI). Wilson's no worse. Team scores more runs but most of the difference goes in the #3 and #6 spots in the batting order. (#7 and #8 also pick up some extras)
Most of the time, a guy who has lots of RBI will also have lots of TB, and a high SLG, and many other indicators that show he's a good hitter. The interesting cases are the ones that don't align. Jeff Francoeur drives in 103 runs in 2007 despite being a pretty weak hitter by any other measure. Hanley Ramirez drives in 67 runs in 2008 despite being an awesome hitter.
As RonJ2 points out, most of the time there are obvious contexts that account for this; you could switch the two players' contexts even-up and the result would be a hilarious discrepancy, Hanley driving in 135 and Frenchy driving in 35, or something like that. And so the more interesting cases are those where the discrepancy between actual runs crossing the plate in front of somebody's hits being less a factor of context, and more of clutchness over a single season. Those are the most interesting cases of all, but the effects are relatively small. I do think clutchness over a season, if on a marked scale, should be taken into consideration in MVP voting, but the fact that I can't easily point to a case where it would be a big factor may be telling. Someone else might come up with a good one.
Basically, RBIs are very important, usually tell us something about a player's contribution that we know as well or better from other measures, and occasionally are very misleading. I imagine you'd accept that, Bear.
I'm curious to see the evidence on this. It sounds ... dubious.
They can be. Over at RLYW SG had a good little analysis of Robinson Cano that show's he's been about 18 runs worse than his context neutral production, b/c of poor performance in "clutch" contexts.
That's not predictive, but it's real lost runs to the Yankees.
Yes. The purpose of stats isn't to capture player value; it's to record what happened in a game. A record who drove in the runs is useful information to have.
Yes, that would be the kind of season I was imagining in #23 (but with negative implications for the hitter). Cano is 8th in the AL this year in Runs Created (to grab a handy less-context-laden stat just for argument's sake), but 22nd in RBI. Unlike Trout (2nd in RC, 24th in RBI), Cano hasn't been a leadoff hitter; he's been batting 3rd or 4th for a high-scoring team. Absent other explanatory factors, the discrepancy becomes interesting.
Chipper's MVP in 1999 perhaps. I think he deserved it anyway but it would have been a very close vote if he hadn't smacked around the Mets in key games.
As a reminder of how the world has changed, Chipper had a 1074 OPS that year -- good for only 3rd in the NL.
However the fancy stats suggest Chipper wasn't hugely clutch that year* -- just 10th in the league in WPA, etc. -- and that Bagwell was very clutch, so fancy stats might give it to Bagwell. Or Bonds of course.
Man, I don't think I'll ever understand WPA (don't bother explaining it). Bonds had only 3.6 WAR that year, 4 oWAR. But he has nearly 8 WPA. How can you add 8 wins but be only 4 wins above replacement? (No a replacement player doesn't get 4 WPA, Chipper was 10th in the league with 4 WPA).
*Which may be true. He got the clutch rep with some big hits in Sept; he might have been un-clutch up until then and nobody would have noticed.
However, the vast majority of guys with 100+ RBIs who have a .400+ RBI/TB ratio are guys having valuable seasons (of those in that category from 1961 to the present, 70% have an OPS+ of 120 or higher; only 4% of those guys--with good ol' Joe Carter being in that group twice--had OPS+ below 100).
As for your Cuyler/Hornsby-Wilson example, it's a little bit thick to be dissing the RBI with a guy who drove in 158 runs the year before he set the record! Wilson hit 17 more HRs in 1930 than in 1929; the guy batting #2 for the Cubs (Woody English) had a much better year in '30 than he did in '29, and the Cubs' #2 slot got on base a LOT more in '30 (.425 OBP as opposed to just .354 in '29). I just don't see this being much of an argument for RBI vulturing. Now, Maurice Van Robays (116 on just 230 TB and a 95 OPS+ for the Pirates in '40)--yes, absolutely. If Hornsby had stayed healthy in '30, the Cubs would have scored another 40-50 runs and some of those would have trickled to Hack--blending all of the effects, he might have only driven in.....180.
Bob, Cano's conundrum is very easily explained by his performance with RISP. In '11, 1.009 OPS, .636 SLG, 91 RBI; in '12, .747 OPS, .373 SLG, 41 RBI--in about the same number of PAs. He's also walking a lot more in these situations, which would drag down RBI as well. The Yanks #2 spot (Granderson/Swisher) has vultured RBIs this year.
IIRC through his first three or four years in the majors Ichiro was hitting something like .380 with RISP. (Too lazy to do the math)
As RonJ2 points out, most of the time there are obvious contexts that account for this; you could switch the two players' contexts even-up and the result would be a hilarious discrepancy, Hanley driving in 135 and Frenchy driving in 35, or something like that.
Player Runners On RBI HR RBI Pct. Jeff Francoeur-2007 487 105 19 17.66 Hanley Ramirez-2008 318 67 33 10.69With 169 fewer runners on, at a 17.66 RBI pct, that's 30 fewer RBI for Frenchy, or 75 total.
With another 169 runners on, at a 10.69 RBI pct, that's only another 18 RBI for Hanley, or 85 total.
source: http://www.baseballmusings.com/cgi-bin/RBIPCT.py?
Player Runners On RBI HR RBI Pct. Robinson Cano-2011 442 118 28 20.36 Robinson Cano-2012 YTD 447 80 30 11.19Chipper had both a low RBI pct and a relatively low number of runners on ahead of him, compared to MLB RBI leaders.
1999 MLB RBI LEADERS (MIN 110 RBI), SORTED BY RBI PCT.
Player R-O RBI HR RBI Pct. Manny Ramirez 507 165 44 23.87 Larry Walker 344 115 37 22.67 Matt Williams 494 142 35 21.66 Rafael Palmeiro 482 148 47 20.95 Vladimir Guerrero 430 131 42 20.7 Dante Bichette 491 133 34 20.16 Mark McGwire 413 147 65 19.85 Brian Giles 385 115 39 19.74 Ken Griffey Jr. 440 134 48 19.55 Alex Rodriguez 353 111 42 19.55 Jay Bell 380 112 38 19.47 Brian Jordan 479 115 23 19.21 Todd Helton 406 113 35 19.21 Roberto Alomar 503 120 24 19.09 Luis Gonzalez 447 111 26 19.02 Richie Sexson 449 116 31 18.93 Jason Giambi 476 123 33 18.91 Juan Gonzalez 471 128 39 18.9 Jermaine Dye 487 119 27 18.89 Carlos Delgado 498 134 44 18.07 Robin Ventura 489 120 32 18 Bernie Williams 505 115 25 17.82 Ivan Rodriguez 439 113 35 17.77 Magglio Ordonez 492 117 30 17.68 Mike Piazza 480 124 40 17.5 Greg Vaughn 420 118 45 17.38 Paul O'Neill 525 110 19 17.33 Jeff Bagwell 488 126 42 17.21 John Jaha 443 111 35 17.16 Eric Karros 457 112 34 17.07 Shawn Green 475 123 42 17.05 Sammy Sosa 462 141 63 16.88 Chipper Jones 395 110 45 16.46 Albert Belle 519 117 37 15.41The simple answer is that the small gain in information is outweighed by the massive loss of simplicity.
RISP stats typically include PAs with runners on:
* 2B only;
* 3B only;
* 1B and 2B;
* 1B and 3B;
* 2B and 3B; and,
* bases loaded.
Half of the RISP states include a runner on first base. So, by using RISP, one does not remove the effect of runners on first or the concerns about different mixes of baserunners for different hitters. I would not be surprised if runners on first base comprise nearly 50% of runners on in RISP situations.
Using RISP is going to improve the absolute RBI Pct for hitters. Will it improve the relative ranking of a low overall RBI pct hitter? Probably not.
SLG are correlated with raw RBI because each HR is also an RBI. One has to remove HR from a player's raw RBI total to see what pct of baserunners he drives in.
I would bet that if you look at (RBI-HR), its correlations with BA, OBP and SLG would show a far narrower spread among them than the correlations of raw RBI with BA, OBP, and SLG.
Given the spread in RBI Pct (net of HR) across hitters, a player's RBI total is more than just baserunners on and SLG.
I obviously agree. Most counting stats except RBI already have an associated rate stat that puts the opportunity into context, usually using AB or PA as the denominator:
H: BA, OBP
BB: OBP, BB/PA, TTO%
SO: SO/PA, TTO%
2B, 3B: SLG
HR: SLG, HR/AB, HR/PA, TTO%
SB: SB/(SB+CS)
(You could even look at runs per times on base, but a baserunner's contribution to that ratio is limited when compared to the batters'.)
historical: http://www.baseballprospectus.com/sortable/index.php?cid=1044579
2012: http://www.baseballprospectus.com/sortable/index.php?cid=1253205
Hello! Don't forget me!
Rk Player RBI Rbat OPS+ TB HR BA Year Age Tm1 Moose Solters 134 -11 95 293 17 .291 1936 30 SLB
2 Dante Bichette 133 3 102 321 34 .298 1999 35 COL
3 Vinny Castilla 131 5 109 312 35 .271 2004 36 COL
4 Pinky Whitney 124 -2 99 280 13 .298 1932 27 PHI
5 Garret Anderson 123 3 104 321 28 .289 2001 29 ANA
6 Jeff Kent 121 3 105 274 29 .250 1997 29 SFG
7 Joe Carter 121 3 112 295 33 .254 1993 33 TOR
8 Glenn Wright 121 0 102 295 18 .308 1925 24 PIT
9 Sammy Sosa 119 -6 99 308 36 .251 1997 28 CHC
10 Del Ennis 119 1 97 247 25 .261 1954 29 PHI
Fewest RBI with 15+ HR and WAR Batting Runs of 50 or more, since 1920:
Rk Player RBI Rbat OPS+ HR TB BA Year Age Tm1 Rickey Henderson 61 62 189 28 282 .325 1990 31 OAK
2 Chipper Jones 75 50 176 22 252 .364 2008 36 ATL
3 Ted Williams 82 50 172 24 242 .345 1956 37 BOS
4 Ted Williams 83 52 209 28 225 .356 1955 36 BOS
5 Rogers Hornsby 83 54 187 17 266 .384 1923 27 STL
6 Pedro Guerrero 87 53 182 33 281 .320 1985 29 LAD
7 Jim Wynn 87 55 166 33 251 .269 1969 27 HOU
8 Ted Williams 87 83 233 38 307 .388 1957 38 BOS
9 Wade Boggs 89 58 174 24 324 .363 1987 29 BOS
10 Mickey Mantle 89 58 195 30 228 .321 1962 30 NYY
11 Ted Williams 89 61 201 29 245 .345 1954 35 BOS
#10 The problem is that raw runs are heavily influenced by factors that have zip to do with ability (specifically batting order position and the SLG of the 3 players after him).
According to posts in an earlier related thread, "triple crown" as an MLB item dates back to the 1930s, which for huge RBI totals was THE decade. That's probably why the RBI was featured in the TC, where runs scored had been considered the more telling stat during the dead ball era (helped by the fact that RBIs weren't even an official stat then.) IMO, they each have the same value as a record of what actually happened. And that appears to be the sense of most posts in this thread...
Well I checked that too when I looked at RBI. RISP is indeed significant in explaining RBI (as in cuts the standard error by two runs), but it's not a primary facot in the rbi process.
First of all, while it's a lot harder to derive in somebody from first, generally speaking you get way more chances to drive somebody in from first.
Second (and probably more important) people tend to assume runner on second, single = rbi. Not close to being true. Unless Alfredo Griffin is the baserunner, you don't generally score a guy from second on an infield singe. Not only that, a single to right (or left) only scores the runner ~60% of the time (don't recall the numbers on a single to center but it's not that different. Now it's been a fair number of years since I've checked this, but I'm genuinely doubtful that it's changed a lot)
I have looked at this. ISO (SLG-BA) is the single most important factor (under the player's control). Given equal opportunities and a more or less typical distribution of baserunners, if two players have the same SLG the guy with the lower BA will tend to drive in more runs.
You must be Registered and Logged In to post comments.
<< Back to main