Yesterday on Twitter, Buster Olney had some thoughts on WAR, and specifically, the way it values middle-of-the-diamond players compared to first baseman. A few selections from his comments:
#1: “Love advanced metrics,but anybody find something a little skewed to 2B/SS/CF? Ben Zobrist No. 6 overall, ahead of all first basemen, McCann?”
#2: “Zobrist, Victorino, Howie Kendrick and Yunel Escobar all ahead of Prince Fielder in WAR. You do wonder if positional adjustments too steep.”
#3: “If you asked 30 GMs who they would pay the most among these players–Victorino, Yunel, Kendrick, Fielder–off ’11 stats,30 would say Fielder.”
There were a few others sprinkled in there as well, but you get the general point. The common wisdom in baseball has been that run producers are the most valuable players in the game, and since WAR does not line up with that assessment, Buster is questioning whether WAR is wrong.
...I get where Buster’s confusion is coming from. For years, we’ve been told that power hitting first baseman are the creme de la creme of baseball players. WAR does not agree with that assessment, and it challenges long-held beliefs about how players have been valued. But that does not make it wrong by default. Challenging assumptions is something any good metric should do, and the fact that it’s shining a light on previously underrated stars like Zobrist is one of the reasons why WAR is so valuable to begin with.
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. Harveys WallbangersOffense and baserunning are not relative to position, only fielding is. Although the way it is presented on bb-ref has the positional adjustment tied in to the replacement runs category.
Batting compared to league average + baserunning compared to league average + positional adjustment + replacement level adjustment + fielding compared to positional average
The baserunning numbers cited (Zobrist +2.5, Fielder -4.2) are compared to overall averages, not to positional averages. The positional adjustment is factored in on its own. The only number which is presented as itself relative to position is UZR defense.
I still go back to the comparison between Cano and Pedroia, in which each, depending on the fielding metric, is either a fabulous defender, or about average, or terrible, costing his team wins with his glove.
It's because of this that I take my WAR with a huge grain of salt; it gives a quick idea of a player's performance, but to start parsing players to the tenth of a WAR point seems ludicrous to me, and to use solely WAR to make some specific judgment (which I have seen, comparing Player A to Player B, either for a season or for multiple seasons) is beyond ludicrous....
Indeed. Standard error for WAR among full time position players is not under 8 runs (regardless of which WAR we're talking about) for a full time player. And is probably closer to 10 or 11.
All players within 5 runs in a season are of clearly indistinguishable value. The limitations of any given metric don't permit more accurate evaluations than that. I understand why WAR is listed to tenths, but some place there ought to be a little disclaimer.
Also, given this the standard error when using WAR for the purposes of career comps will not be under 3 wins for careers of any substantial lengths (probably closer to 4)
Best sentence EVER.
WAR also regularly ranks dh's as significantly more valuable on defense than Gold Glove fielders. (Before you get all exited, yes, we are all aware that the Gold Glove voting is flawed, relax.)
Bill James Win Shares are a much more acurate tool for performance measurement because it does not give negative values for defense. Everything you do on the field has some positve value.
It's unclear why history (so far) has chosen WAR over Win Shares.
WAR is so flawed that its proponants really should show greater humility. WAR is valuable as a tool for a quick superficial look on comparing offensive production.
And it may not even be used 5 years from now.
I think it's useful but the tenths should not be used to make definitive arguments. If I see;
Pedroia - 6
Bautista - 7
It's relevant to me if the details are;
Pedroia - 5.5
Bautista - 7.4
or
Pedroia - 6.4
Bautista - 6.5
I think more detail is better as long as I know not to go overboard assigning meaning to it. The problem is that I find people using WAR a little too definitively recently. There are enough issues with the adjustments and the defensive numbers that I think WAR is a starting, not an ending, point. Instead too many arguments I read around the interwebs come down to "Player X has 6.7 WAR so he is superior to Player Y at 6.3 WAR."
This is a joke right? Not even going to bother going over the problems with Win Shares.
Not if you don't make the error of ignoring the positional component of defense, it doesn't.
I think a big factor is that a WAR unit is more intuitive than a Win Share. A win is a pretty basic unit of measurement. When someone says a guy is a 5 WAR player, we all know he means that he gives his team 5 wins. Win Shares as units are somwhat arbitrary in this sense (James multiplies by 3 for more or less aesthetic reasons doesn't he?)
Of course the biggest reason probably is WAR is freely accesible and you have to pay for Win Shares.
I used to use Win Shares for all my databases, but once B-Ref started including WAR it just wasn't worth the money or poor navigating of Bill James' site.
Also in '05, David Ortiz had a fielding + position value of -17 (small negative fielding value, presumably from first base in interleague season, but it rounds to -17 either way).
So... yeah, you're going to have to provide an example that does factor in position.
WAR is inherently flawed and will be discredited and in history's scrapbin within a few years.
Let the witch burning commence. Enjoy.
ummm
1: Because the way wins are allocated between offense, defense and pitching seems a little too arbitrary
2: Because there are no negative values for defense (I assume you were trolling when you cited this as a strength)
3: Because there are no negative winshares
4: Because there is no attempt at setting a replacement level (see 2 & 3)
OTOH I like the fact that winshares uses real rather than pythag/3rd order wins
I wouldn't say regularly, but you occasionally see high profile examples like this:
Manny Ramirez 2004-2006 (404 G LF, 25 G DH) - Defense and positional adjustment -63 runs
David Ortiz 2004-2006 (398 G DH, 54 G 1B) - 38 runs
Thus, Ortiz the DH, despite being an inferior hitter, was deemed more valuable than Ramirez the LF.
Ramirez - 11.6 WAR, 15.9 oWAR
Ortiz - 14.9 WAR, 14.6 oWAR
All the metrics are flawed in on away or another, to one extent or another.
yes, when something better comes along it will rejoin winshares which is already there.
Jan Hus begs to differ (or would if he wasn't burned alive)
Being a heretic is a great way to get a following.
Ortiz wasn't actually the inferior hitter over that span - he has more Rbat, 136 to 129. (If you add in the baserunning/ROE/DP components, Manny gains a few runs but doesn't catch up.) oWAR is position-adjusted, so it's not good for comparisions of straight offense.
Your overall point, that Manny is penalized for playing the field, is correct, and is something that may merit adjustment when evaluating Manny's career. However, Manny was nobody's idea of a "Gold Glove fielder," which was the claim made in the original post.
Is Prince a better player than those other guys? Yes, of course. However, the kinds of players you can get to replace Prince are much better than the guys walking around that play 2b/ss/cf.
I have no quarrel with WAR saying guys like Zobrist are worth more wins (and money) than Prince. It doesn't change the fact Prince is the better player. What you pay them relative to position, is an entirely different question.
No, Prince is better HITTER than those other guys, whether or not he is a better PLAYER is a different question
Yeah, I forgot about the positional adjustment being in there. Taking that out, Manny has 121 offensive runs (batting and running components), Ortiz 125.
Better hitter, not better player.
The problem with this is that the offensive core of win shares in at heart (park adjusted) runs created per out. And that's a mediocre method.
WAR triumphs in part because of the ease of access. But that's only part of the story. Sean (talking the version of WAR seen on BBRef now) has incorporated aspects (baserunning, reaching on error and better adjustments for grounding into double plays) that win shares is blind to. They're not generally all that important, but since WAR starts from a more accurate base on the offensive side, the offensive component is simply better in WAR than in Win Shares. Add in that WAR is modular which makes it convenient for hypothesis testing and WAR is miles ahead.
Then there's the baseline. WAR sets replacement level sensibly. Win Shares does not and thus considerably over-values mediocrity. Yes, you can deal with this with either loss shares or by using win shares above base.
On the defensive front, I like the win shares concept that all fielding value is positive and I wish I could find a way to incorporate that notion into WAR. Win Shares is probably at its best specifically when comparing DHs to bad fielding corner OF or 1B. And that's an important enough issue that I'll forgive a multitude of sins.
There is no justification for the way method errors are assigned to players in Win Shares either.
As for WAR going away, I doubt it. Combination of availability and design are powerful enough that it would take something very good to unseat it.
EDIT: The primary problem with James' defensive methods is that the spread between the best and worst fielders is almost certainly too small. But I'd expect a very high correlation between the way BBRef WAR defensive values and win share defensive values. They are both fundamentally range factor based with an attempt to adjust for context.
If you look at '04 on its own, Manny has more offensive runs (36-33) but a lower WAR. So again, your overall point stands, regardless of my nitpicking of the specific example used.
Jesus was a heretic.
Now quit self-dramatizing and make an argument, or shut the #### up.
Wait, are you congratulating yourself on being a heretic? This board sees multiple arguments every single day on the utility of WAR. I think the only real reason it has the currency it does is laziness - seriously, it's just easier to use a single number to compare players than it is to have to cobble together more than a few. Everyone (on this board at least) is 100% aware of its flaws and would be excited to greet a superior statistic.
This is incorrect. The other guys are worth more because of the relative scarcity of their position. Whereas Prince plays a position that is relatively loaded with good players. The "replacement" 1b has a much, much higher floor than the "replacement" 2b/ss/cf.
Prince is the better player. The other guys are worth more. WAR doesn't exist to determine which players are necessarily better than others, it calculates value relative to replacement at a given position.
I could construct a hypothetical scenario where the replacement at a given position was so terrible, a guy like Mark Kotsay or even Aaron Miles had a greater WAR than Prince. Still doesn't change the raw fact Prince is the better player, even if his WAR would be lower than Miles or Kotsay under this scenario.
Take THAT Mr. Fancy WAR carrying but carrying a name of a Jewish attorney from the 1940's based in DC Zobrist!
Prince can't play 2B/SS/CF. Zobrist can play 1B. There is more to being "better" than hitting the ball, and Prince is highly limited in how good he is beyond his ability to hit.
We're splitting hairs, though, I think.
Have Prince play that hypthetical position in the field (or, heck, 2B) and then tell me if he's better.
But that said, B-Ref WAR has Zobrist at 7.1 in his big year in 2009, 3.1 last year, 4.5 so far this year when he's obviously hitting a lot better than last year. These hardly seem like they're weird random figures; in fact they sound about right for a guy who is an excellent hitter, a good defender, and had an off-year. Fangraphs has him at 8.6, 3.7, and 5.6. Ballpark figures, but they're clearly in the ballpark.
Dustin Pedroia is faster than Adrian Gonzalez, quicker than Adrian Gonzalez, has better footwork and better body control than Adrian Gonzalez, and throws harder and more accurately than Adrian Gonzalez. These very real differences between Pedroia and Gonzalez, as baseball players, are greatly measured in Pedroia's +10 run advantage on Gonzalez in positional adjustment.
This is absurd, I would like to see Prince TRY to play 2B, or CF or 3B, just go ahead and try.
Anyway, Zobrist is putting up a 146 OPS+ (and yes I know he put up a 97 last year, and 149 the year before)
Prince is at 166. So you are saying that the fact that Zobrist can play pretty much any position without embarrassing himself, whereas Prince Can't has zero bearing on who is the better player? That's just wrong.
There are situations where a guy (let's use Youkilis) CAN in fact play a harder defensive position- but isn't because the team has someone else there- so yeah Youk was likely a better player than some other team's guy who was getting a position adjustment that Youk wasn't. That's not Prince, there is no team in the MLB that would play him at a harder position than 1B, there are a dozen who'd move him to DH in a heartbeat.
I'm not sure about that first one.... Is Gonzo really that slow :-)
Kind of a mismatch though. At this point Alcides Escobar is better than Dunn - without even considering fielding.
This is just dumb. Who is it that has been telling us this? Yes, some first basemen have gotten sketchy MVPs, like Ryan Howard and Mo Vaughn, but so have Jimmy Rollins and Miguel Tejada.
Prince Fielder, who is the focus of the discussion, has finished behind at least one shortstop in the MVP voting every single year of his career.
Now I'm curious. There's a fielding metric that says Pedroia is a terrible defender?
Yep. I'd bet on Ortiz if the two were to race.
With Zobrist, though, at least this year, that's pretty much an academic concern: he's exclusively played second base and right field. If there's a degree of "versatility" that deserves a bonus in advanced stats (and I'm not convinced that it does), his 2011 doesn't rise to that level.
Maybe not terrible, but FRAA has him at -3.3 runs in 2009 and -7.2 in 2010, and a combined -1.2 from 2006-2011, while UZR has him at a combined +32.5 over the same period.
The difference with Cano is even more stark: UZR has him -39.4 for his career, while FRAA has him at +31.2.
What??? You lost me here, Harveys.
I know you were just being funny/clever, but you still lost me.
Where do you get that? How is the scarcity of position manifested in WAR? Rpos? That doesn't vary by more than 1 or 2 runs from season to season. In 1977, the average NL SS had an OPS+ of 77. The Rpos for a whole season was 10. In 2000, when the average AL SS had an OPS+ of 95 (AROD, Jeter, Nomar, Tejada, Valentine), the Rpos for a full season was 7. Those are two pretty extremes, and yet the scarcity bonus is only 3 runs. That same year (AL 2000), second base (OPS+ of 84) was 3. So even though there were many more good hitters playing SS that year, SS still got a much bigger position bonus than second base.
It is manifested in the replacement level for offense. If second basemen suddenly started hitting for a collective .880 OPS, Zobrist wouldn't be nearly as scarce a commodity.
No doubt in this case. However I did want to make the point WAR is great, not because it tells us who is better, (it doesn't), it tells us who is more valuable in terms of marginal value over replacement (which has a meaningful positional adjustment).
Why is this absurd? By your logic Wes Helms is a better player than Prince because he can play 3b. Ditto Carlos Gomez in CF or Alexi Casilla at 2b. I know this isn't your position on this subject. I would love to see any of these guys hit 50 HR in their early 20s or ever in fact....I'd even be happy if they slugged .650, for a month.
His name is Ben Zobrist. Everything between Mr. and Zobrist was meant in a light-hearted fashion.
WAR might be the most important metric when it comes to creating your roster intelligently using the limited available money. If you are a GM with a unique ballpark that seeks to have a roster designed for run prevention as well as fast base runners and position flexibility, then Zobrist, without question is the more valuable player and perhaps the better overall player. If you are in Yankee stadium and you NEED a slugging lefty to rape the shallow RF wall, then Prince is the better player.
But that would effect every player in the league, not just second basemen. Rrep is leaguewide replacement, not position adjusted. See Prince Fielder. He has the most PAs on the Brewers, and the highest Rrep. SS Betancourt has 83% of the PAs, and 83% of the Rrep. There is no position scarcity figured into that stat.
The comment you responded to is poorly worded, but I understood what he meant, and I think you are being intentionally obtuse here. Surely you don't believe the best players are the best hitters period full stop? Alex Rodriguez (162 OPS+) was a better player in 2000 than Jason Giambi (187 OPS+), and it's not controversial. Or do you disagree?
Fielder is a better player than Helms because the sum total of his offensive and defensive* contributions far outweigh Helms's. Zobrist is about equal to Fielder because his offensive and defensive contributions are about the same, if in a wildly different shape.
* including some positional adjustment, however you want to do it.
If hitting is measured against league average, the effect on league average would be small or non-existent, so 2B hitting WAR, in isolation, would not change. The positional adjustment would change, and comparing Zobrist's defense to position average would make him a more "valuable" defender. If I'm right about these premises, it would translate to a higher WAR (for hypo Zobrilla) if the change in positional adjustment was smaller than the attrition in overall position defense, and a lower WAR if the change in positional adjustment was larger than the attrition in overall position defense, right?
rPos is the one you want. There, the difference between Fielder and Betancourt is something like 16 runs per 500 PA.
No, what is absurd is your contention that best hitter = best player with seemingly no consideration given to defensive value.
Obviously Prince is a better player than Carlos Gomez - even if Gomez is better at every aspect of baseball except hitting- since hitting is hugely important and since the hitting gap between them is a vast unbridgeable chasm.
What is unclear is whether Zobrist or Prince is a better PLAYER than the other, the only way to conclude that Prince is a better PLAYER period full stop in the way that you have done, is to assign no weight whatsoever to fielding ability, baserunning etc.
Pos XR Fld TotC - 8.0 12.5 4.5
1B 12.6 -12.5 0.1
2B - 4.7 2.5 - 2.2
3B - 8.0 2.5 - 5.4
SS - 7.4 7.5 0.1
LF 0.0 - 7.5 - 7.5
CF 3.7 2.5 6.2
RF 10.3 - 7.5 2.8
DH 3.5 -17.5 -14.0
This'd be more complete with baserunning split by position. Does anyone *do* that?
Unsurprisingly, LF and DH suck, and 1B is not special.
Oy, I feel like I'm talking in circles.
This is true for all stats, not just WAR. People seem to get this for WAR, because there are different constructions of them out there. But the same lesson applies for any metric.
Yep. But you also forget that Rroe, Rbaser, and Rdp are all adjusted to a major league average rather than a positional average. I'm sure a difference between a 1B and a SS would also be there.
This is a point I've never really considered. Until this thread, I was perfectly fine with thinking about base running as relative to average player, but there is a point to consider when you are looking at other positions. Jason Kendall was a great baserunner for a catcher, but for his career he's basically an average baserunner. He still gets bonus points over other catchers, but it could be argued that it's not enough since his running is not being compared to other catchers but all other position players.
I don't see why it matters. If an average catcher is -5 vs the average player, and Kendall is +2, then he's +7 vs the average catcher, whether it's done in 1 step or 2.
edit: or to use real world numbers, Kendall is +1 for his career, Bengie Molina is -30. Does it really matter if they are compared to each other directly, or indirectly, through their ability vs the league average?
And the reason why big power guys get the money is that it SEEMS that big power is a skill that is subject to the least amount of variance.
I agree. I don't remember seeing him pull his hands in like that very often before. It was striking.
One thing Fielder has over nearly everyone in terms of 162 game reliability is incredible durability. He is always in the lineup.
You must be Registered and Logged In to post comments.
<< Back to main