I’ve had folks tell me lately that even if Cabrera wins baseball’s first Triple Crown since 1967, Trout remains the MVP pick. My apologies. I simply cannot comprehend that.
But fans of Sabermetrics, these new-age stats that attempt to give you an all-around worth of a player, actually will tell you — and with a straight face! — that batting average and RBIs are irrelevant today.
There might — might — be the case with batting average, because that doesn’t necessarily tell you how clutch or how much of a run producer a player is. There were some in Seattle who thought Ichiro Suzuki could have hit many more home runs, but didn’t because he wanted his batting average to stay sky high.
But RBIs? Really? Irrelevant? The critics like to say that’s only a measurement of how often guys get on ahead of you. Well, to that I have a two-word response: Delmon Young.
...Now, I ask you, is that an MVP? That’s for 28 of my peers to decide; I don’t have an MVP vote this year.
But if that is an MVP, then let me say this: Voters better be consistent and make Tigers ace Justin Verlander the Cy Young winner for a second straight year, because he leads the major leagues in WAR, too, not contenders Felix Hernandez or David Price.
By the way, here’s another thing FanGraphs’ WAR tells us, folks: Anibal Sanchez (3.1) and Rick Porcello (2.9) are worth more to their team than Angels ace Jered Weaver (2.7) is to his. Never mind that Weaver is 18-4 with a 2.79 ERA. What do those stats matter anymore?
Repoz
Posted: September 23, 2012 at 10:23 AM |
180 comment(s)
Login to Bookmark
Tags:
awards,
mvp
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.
I think this gets at the main point that many people are missing, when it comes to the criticism of fWAR for pitchers. Neither ERA nor FIP tells you everything you need. The problem is that many people don't seem to realize that bWAR vs. fWAR is not the same as ERA vs. FIP. bWAR is fairly well-thought out and designed specifically to avoid the pitfalls of both ERA and FIP; fWAR is lazily dependent on FIP as if it was the end-all stat for pitchers. So when I say that fWAR is bad for pitchers because it is based on FIP, that's not to say it should be based on ERA; it means that it should be better than both ERA and FIP, like bWAR.
Alright, the scaling is problematic (if we're saying FIP is a record). But it's also just a constant added to everyone's number so it can be ignored. Other than that, FIP is pretty similar to slugging, assigning weights to three different things that actually happened and then dividing by opportunities (IP or AB).
Not really. Slugging is just TB/AB. It is incredibly simple and straight forward, and in plain terms tells you exactly how many bases a player can be expected to accrue every time he gets an at bat. FIP is based on nerd-derived (I say this with love) weightings that have no obvious connection to reality.
Sure. His job isn't to make good pitches; it's to prevent hits and runs.
And as an account of what happened, we care about the result, not the pitch "quality."
I think this is oversimplistic; a pitcher has an effect on the batter, beyond just the pitch f/x of the specific pitch. Also, your example isn't relevant for a discussion of FIP, since a HR will negatively affect both FIP and ERA.
Regardless, I think it is obvious that you have to adjust for defense, and you have to adjust for the pitcher's level of competition. bWAR attempts this in a very thorough way. fWAR takes FIP and lazily assumes that merely using that one stat makes all these adjustments.
Exactly.
This still strikes me as a one-way street. If two batters get equal results from two very different pitches, one gets a hanging curve and rolls it between Arod and Jeter and the other takes a good outside pitch with movement and hits a sharp single into right, we don't consider the quality of the pitch when evaluting the hitters. They both just hit singles. It's only with pitching where we try to divvy up the responsibility for the outcome.
Like I said I would give some weight to bWAR and some weight to K-BB WAR. I wouldn't exclusively use one of the other.
I think we don't consider it because it's beyond all of us, even the best statisticians, to figure this out. I'll bet teams are considering it.
I'm not exactly sure what you mean; how is this something fWAR does better than bWAR?
I think fWAR, like FIP, is something useful to look at when making predictions for the future. It is not a good stat for retrospective valuations.
Again, we have stats like PrOPS specifically tailored to tell us what "should have happened." All the ball-in-play data that informs xFIP and the like is available just as much for batters as it is for pitchers. It's just that no one takes it seriously for hitters, for various reasons.
Why would you believe that? Pitcher miss their spots, hang breaking pitches, etc. all the time. Just like hitters get fooled, or take bad swings.
How does xFIP use batted ball data?
It's just FIP with the HR/FB ratio regressed to league avg.
Okay, but I'm saying one is patently better suited for that purpose, and I'm asking you to flesh out your explanation as to why you disagree (which you must, if you really think they both deserve weight). I don't follow when you say:
The problem is that it does everything at a team level and not at an individual at bat level.
Why, if Hellickson's preventing runs more effectively?
I think I missed the part where pitchers have full control/responsibility/whatever over HRs, Ks, and BBs. How are Ks and BBs not hitter-dependent? Obviously they aren't defense-dependent, but they still aren't entirely the "reponsibility" of the pitcher or entirely within his "control." Throw the same pitches to Mark Reynolds and prime Joe DiMaggio and you aren't getting the same numbers of Ks and BBs.
Does FB rate not count as batted ball data? If it makes my point more clearly, replace xFIP with tRA or SIERA.
He's having a better year in results, which is what we should care about looking backwards (i.e. for Awards).
Looking forward, we'd want to get a sense of the quality of contact the two are allowing, LD%, pop-up %, GB/FB ratio, etc. to determine who had better "true talent" this year.
He's having a better year in some results (runs and earned runs). Why limit which numbers you look at?
Why? What is the "gut feel" problem with saying Hellickson has had a better year thus far (i.e., up until now, he has executed better when it counted), but that based on the peripherals we'd expect Shields to be better going forward?
Because these are the results that actually have the most import, no? There is such a thing as Fielding-Dependent pitching, even if isn't as sustainable as its independent counterpart.
Because the ultimate purpose of pitching is to prevent runs (*), not to strike people out or walk fewer hitters. Striking people out and not walking them is a secondary effect that correlates with the important thing, but it isn't itelf the important thing.
Games are won and lost by teams based on runs scored vs. runs allowed in defined 9-inning increments.
(*) At higher leverage game points, but that's neither here nor there WRT this conversation.
Except it isn't. The ultimate purpose of pitching is to win games and no one here would use pitcher wins to evaluate.
Well, yeah, but everytime you say that you get shouted down. Sticking with runs is less controversial and the conversation can proceed just as effectively.
(But, yes, in fact the ultimate purpose of pitching is to win games, i.e., to allow fewer, or contribute to the allowance of fewer, runs than your team scores within the defined 9-inning increment.)
What I mean is that the outcome varies much more. A bad pitch from a pitcher is a shitty fastball down the middle 2-0. That's going to be hit out 15% or something. A good outcome for a batter is a home run. The pitcher cannot control what the batter does, he can only control the quality of the pitch. I'm asking if you're ok with the batter performance being mixed in with the pitcher performance. Sugarbear Blanks is apparently. That's fine, then use RA. I personally like to weigh RA and K-BB.
No, that's not true. The ultimate purpose of baseball teams is to win games. The ultimate purpose of pitching is to prevent runs.
One reason to limit which numbers you use is because you have to draw the line somewhere, and limiting it at the most basic and most important unit of measurement in baseball, runs, seems as good a place as any. Sure, you can go one (more theoretical) step beyond where you try to isolate pitching from defense, but then why not go beyond that to where you isolate pitching from the opposing hitter.
Well, since you can't score runs on defense in baseball, preventing runs is all the pitcher can do as a pitcher (excluding pitcher hitting in the NL) to win games.
Good, then look at RA+, IP, and K and BB rates. No need to involve FIP.
I think this a bit too cute, and you know it. The pitching/defense half of winning games is to prevent runs. Wins are a bad pitching measure because pitching is less than half of winning a game. Run prevention is clearly the end game for a pitcher and that is why people focus on it as a baseline of analysis.
Because some numbers are useless. Like anything related to FIP. Look, if you are using FIP in a discussion about league awards you have disqualified yourself from intelligent discourse. You have no business speaking on the subject because you don't know what you are talking about. That's it. Final. End of Story.
It's mostly a dig at SugarBear and Jack Morris.
Not true. A good pitcher will make a batter less likely to execute on bad pitches. This is why pitch sequencing and disrupting timing are important. This was Maddux's life blood.
I think I missed the part where pitchers have full control/responsibility/whatever over preventing runs.
Not true. A good pitcher will make a batter less likely to execute on bad pitches. This is why pitch sequencing and disrupting timing are important. This was Maddux's life blood.
Yup. A hanging breaking ball inside is much less likely to be crushed when the batter is looking for a FB outside.
Since this is so eloquently true, it doesn't need seconding, but I'll second it anyway.
Except that another word for "batter performance being mixed in with pitcher performance," is ... "baseball."
Why do you even want to bother with batters? Just have all the pitchers throw bullpens, measure the velocity and location of all their pitches and give the high scorer the Cy Young.
They have the large majority. The variance in pitching talent is much, much greater than the variance in team defense.
Also, no one on the RA/ERA side of this debate is against making an adjustment for defense, a la BRef WAR.
Again, what I'm proposing is RA + K-BB RA/2. Not some real radical ####. Just giving K and BB more weight than BABIP and HR rate.
I think I missed the part where full control/responsibility/whatever was a prerequisite to anything important. No one in a baseball game has full control/responsility/whatever over anything.
Right. The point is NOT to say that ERA is the true talent level, but to parse the FIP-ERA differential using something based in reality. fWAR merely takes FIP and then magically sublimates that whole disparity to "defense" (but without actually crediting anyone for these defensive achievements) based on lazy logic.
The problem is you're calling something a "RA" that isn't; it's just scaled to look like one.
If you want to use Ks and BBs, go ahead, most people look beyond ERA+ to those things. But don't average them into a stat that isn't what it says it is.
I think there are very very few people here who would use fWAR for anything. Even the staunchest critics of ERA probably think bWAR is a better stat.
If kwERA thinks a guy should have had a 3 ERA and his actualy ERA is 2.5, I'm going to give him a 2.75 deserved ERA for the purposes of figuring out who should be the Cy Young (note, I might not do this with Dickey because I have no idea what knuckleballers do to BABIP research).
I'm not sure exactly. Maybe because it seems like he's pitching worse but getting better results through better luck. I'm not sure why exactly I think he's pitching worse -- whether it's that I've been brainwashed by DIPS arguments or whatever -- but I do think that. I realize we don't hold hitters to the same standard -- e.g. we don't normalize their BABIP to some number -- but it seems different for pitchers somehow. I'll have to think about it.
To switch sides, fWAR completely ignores some things that should be attributed to the pitcher. DP rate, GB rate in general, ability to hold runners, ability to pitch from the stretch (somewhat included). It should not be used on its own.
You're the one who determined that Hellickson was better at preventing runs.
Then what you're saying is that you're only looking at strikeouts and walks.(*) I'm not sure what the purpose is of converting Ks and BBs to something "like" an ERA and comparing it to actual ERA, other than to lend an undeserved patina of legitimacy and rigor to it.
(*) And then, far worse, saying that Cy Young votes should be based on it. Or see 135.
He has been this year, which is all I care about and all that matters. He's not "solely responsible" for it, but no one is for anything in baseball -- including strikeouts, walks, and homeruns.
Because runs allowed is what matters for valuing pitchers. We don't particularly care that one pitcher has allowed fewer runs than expected.
EDIT: And yes, it is worthwhile to adjust for the quality of defense.
Edit > Also, I give a bonus to RA Dickey, because he's awesome.
bWAR does a much better job with defense, so the case for fWAR being at all useful for Cy Young must be in "batter variation." But again, I'm not following what you mean by this in terms of what fWAR does. Could you explain?
You are history's greatest monster.
There is also numerous research showing K and BB rates are very predictive and HR rate and BABIP are not http://www.baseballprospectus.com/article.php?articleid=14293. This implies that pitchers have more control over K and BB rates. Therefore, I would like to give more weight to them than other things involved in ERA.
http://www.insidethebook.com/ee/index.php/site/comments/verlanders_skill_with_balls_in_play/
often it actually makes it worse. (read the comments)
I read the comments; I didn't see any discussion of bWAR's defensive adjustments. Could you be more specific as to what you are referencing?
http://www.insidethebook.com/ee/index.php/site/comments/how_much_is_each_pitchers_babip_affected_by_his_fielders_that_day/
You actually don't need to read the comments.
Okay, but this is an sustainability argument for refining the roots of FIP as a predictive stat. It doesn't add anything to question of what is a better reflective stat. bWAR adjusts to the strength of the opposing hitters based on their real world performance, rather than trying to interpolate that performance from K/BB rates. I fail to see how fWAR is better here, at least in a reflective capacity.
And with that I'm done. You guys can believe whatever you want.
That's an interesting article, but it's not particularly conclusive, with both Tango and MGL conceding that Verlander could be causing this effect at least partially himself. The main take away there is an excellent reminder that you need pretty large sample sizes before you can say anything with confidence about any stats with defensive components.
Similarly, a pitcher may give up fewer runs than you'd expect because of great numbers with runners on (and/or by giving up a lot of solo HR). This may or may not be an ability (there are good studies of hitting with RISP [or any other definition of clutch] but not much work done on the pitching side. There is one old study by James which indicates that veteran pitchers do tend to make better adjustments. ISO tends to go up, OBP down with bases empty for the veterans. But as I recall it, it was far from definitive) but it counts.
And yes, it counts on the offensive side too. The standard error for a full-time player is in the range of 5 runs and most of that is in the clutch/timing/luck area. If somebody has done something not considered by the metric and you believe it has value, by all means bring it up.
On the subject of counter stats it's also worthwhile to note that there's also a pretty big batting order affect. One of the earliest stathead studies is Steve Mann's on how batting order affects counter stats.
Estimated Effect of batting order position on total runs and rbi (unfortunately he only looked at the two combined. Obviously batting first will tend to increase your runs scored but will decrease your RBI even more. Not just because of the one AB per game when you are guaranteed to bat with nobody on, but because you are primarily trying to drive in the #8 and #9 hitters and as a group they tend to be poor at getting on base)
1 -5%
2 0%
3 +10%
4 +15%
5 +10%
6 0%
7 -5%
8 -15%
9 -10%
Slots 7,8,9,1 and 2 have different effects in the AL, but to my knowledge Mann never published this (wasn't of any interest to him. He was working for the Phillies at the time)
Because these are the results that actually have the most import, no? There is such a thing as Fielding-Dependent pitching, even if isn't as sustainable as its independent counterpart.
Shields is 15-9, Hellickson is 9-10. Shields is doing a better job at winning games, and that's really what matters more so than earned runs allowed or runs allowed.
(Understanding that you might have tongue firmly in cheek)
Hellickson's record is almost spot on what you'd expect given how much he's pitched, his runs allowed and his run support. Shields' record is in fact a tad better (You expect about 13.4 wins -- and since there's no reliable way for a starter to get .4 wins ...) and I'm fine with giving him some credit for this. (Though I'm thinking a lights out bullpen might have something to do with this too)
That's what you guys get for relying on Fangraphs and bWAR instead of counting stats and batting average.
Any stat that attempts to ignore both runs scored, or most of the components allowed is useless. If you are going to use an era estimator, it better include k/bb/hr/singles/doubles/triples/gidp in the equation, and probably also cs/po/sb for it to be any use as a backwards evaluating tool for an award.
The awards aren't trying to make a prediction of future performance. Just because the numbers show a pitcher should have only allowed 3 homeruns over a course of a certain number of innings, and he allows 8, doesn't mean he should have those homeruns removed from the equation. (or the opposite)
You don't look at a batter's 20 for 30 week, and say "well he's really only a .300 hitter so we need to remove 11 of those hits."
The fundamental question to me is whether the pitcher did something bad to allow those HR, or did he just happen to face batters that did something awesome. If it's the latter he really shouldn't be penalized for it in my view.
The only way to answer that definitively is to independently judge the "quality" of the pitch. Which is doable, but as SoSH noted, you'd have to do the same for hitters, crediting them more for hits on better pitches and penalizing them for making outs on BP-caliber fastballs down the middle.
This makes a lot more sense to me than rWAR, which makes a defense adjustment based on team that doesn't necessarily apply to individual pitchers (like the Verlander example above).
You must be Registered and Logged In to post comments.
<< Back to main