Positional Differences in the Price of WAR

This week, I’ve talked about the retrospective price of WAR on an aggregate level. What I haven’t studied is the retrospective price of WAR by position. I thought this was particularly important in light of my finding that positional adjustments didn’t matter much for arbitration salaries. Players who played tougher defensive positions were underpaid in arbitration, relative to those who played easier defensive positions. As it turns out, the price of WAR has been much more expensive for some positions.

Regard:

Position N 2007-11 $/WAR 2007 $/WAR 2008 $/WAR 2009 $/WAR 2010 $/WAR 2011 $/WAR
C 116 $4.2 $7.9 $6.4 $4.4 $2.2 $3.3
1B 132 $6.0 $6.1 $6.2 $5.1 $5.6 $7.8
2B 80 $3.0 $2.8 $2.5 $1.5 $4.1 $4.1
SS 87 $3.5 $3.0 $4.3 $4.6 $3.2 $2.7
3B 109 $4.2 $3.6 $3.6 $4.7 $3.7 $6.1
LF 106 $7.1 $6.6 $6.7 $6.8 $5.3 $17.3
CF 51 $5.9 $4.9 $6.0 $6.9 $5.1 $8.2
RF 77 $5.7 $4.7 $7.1 $7.6 $4.0 $5.6
DH 39 $7.4 $4.2 $8.4 $16.3 $5.1 $11.0
SP 317 $4.7 $4.6 $4.6 $4.3 $5.5 $5.0
RP 321 $16.4 $9.7 $16.2 $26.0 $16.0 $19.3
C-2B-SS-3B 392 $3.8 $3.7 $4.0 $3.8 $3.3 $4.1
1B-OF-DH 405 $6.3 $5.4 $6.7 $6.9 $5.2 $8.5
Hitters 797 $5.0 $4.6 $5.2 $5.3 $4.3 $5.9
Pitchers 638 $6.0 $5.3 $5.7 $5.9 $6.8 $6.3
All 1435 $5.4 $4.9 $5.4 $5.6 $5.1 $6.1

The elephant in the room, of course, is the astonishing $16.4 million per WAR paid to relief pitchers. Let’s ignore that for a moment and simply look at hitters. It’s clear that as you move to the easier end of the defensive spectrum, the more likely you are to be overpaid. Of course, that could be a mirage, because players can be moved to first base, outfield or even to designated hitter if they’re aging poorly. Because of that, I looked only at the price paid in the first year of deals. The same result held: hitters who play tougher positions are underpaid on a per-WAR basis.

Position—First Year Only N 2007-11 $/WAR
C 80 $3.1
1B 77 $6.1
2B 55 $2.5
SS 46 $3.0
3B 61 $3.9
LF 62 $4.9
CF 25 $6.3
RF 39 $4.4
DH 24 $4.6
SP 184 $4.0
RP 221 $10.7
C-2B-3B-SS 242 $3.2
1B-OF-DH 227 $5.2
Hitters 469 $4.1
Pitchers 405 $5.0
All 874 $4.5

Not only are first base, outfield and designated hitter overpaid either way, but so are relievers. In fact, relievers are paid nearly twice as much for the same WAR as other players. Does that mean that relievers are overpaid? Maybe not. That all hinges on WAR being correctly specified.

There are two different WAR statistics commonly available on the web — and they agree more often than they disagree. The other WAR,  which Sean Smith (a.k.a. “Rally”) invented, is called rWAR and is available at Baseball Reference. Smith’s rWAR has a much higher replacement level for starting pitchers and much lower replacement level for relievers. As a result, using rWAR suggests relievers are paid 30% more than comparable producers elsewhere on the roster, while fWAR puts that premium at 200%.

Does that mean that rWAR has a more accurate view of relievers than our WAR at FanGraphs? Not really. It just shows how complicated reliever evaluation can be. If we run the same analysis on Baseball Prospectus WARP, we would get answers even more extreme than the examples here. Relievers accumulated 89.4 fWAR this past season, but only about 50 WARP. Relievers might be overpaid by an even larger factor if we used Baseball Prospectus’ numbers, which means FanGraphs might represent a sort of a centrist’s view on reliever value.

This undoubtedly requires more research, but I’ll propose one early possibility about the price of relievers: For a market failure this large to exist, there would need to be incentive problems at some level. The first thing I thought to check was whether general managers are building better bullpens to save their jobs. After all, GMs don’t give out contracts to maximize their chance of winning a division or a World Series. They give out contracts to maximize their own value, which is inextricably tied to keeping their jobs.

Between 1994 and 2011, there were 40 GMs fired among 532 team seasons. Below, I have a breakdown of hitter, starter and reliever WAR, looking separately at teams whose GM was fired and teams whose GM was not fired. In that table, we can see that the average team whose GM was fired had half as much WAR from relievers as the average team whose GM who were not fired. Overall, the average ranking of bullpens that belonged to GMs who later were fired was 21.6.

WAR Group Avg. if GM Not Fired Avg. if GM Fired Ratio Rank if GM Fired
WAR Hitters 21.9 18.2 83% 19.3
WAR Starters 12.5 9.8 78% 20.9
WAR Relievers 3.0 1.5 50% 21.6

What this suggests is that general managers have watched one another be fired for having bad bullpens — but they’ve seen peers be forgiven for having bad lineups or poor starting pitching. It seems that when a team falters because of poor hitting or bad pitching, the farm system, the scouts, the manager or the players get blamed. When a bullpen implodes, it’s the general manager’s fault.

I tried to run a couple of regressions to figure out if there was anything else we could learn that would prove or disprove this hypothesis. First, I ran a regression of a team’s wins on its WAR from hitters, starters and relievers.

Wins = 42.2 + .95*WAR_hitters + 1.03*WAR_SP + 1.36*WAR_RP

It does look like relievers might be undervalued, accoridng to WAR. Still, the 95% confidence interval for the coefficient on reliever WAR is (1.12, 1.60), so this could be a small difference. In reality, the fact that high-strikeout relievers have particularly good BABIPs could mean that their FIPs slightly underrate their skill levels.

But then I ran a logit regression to test the probability of a GM being fired as a function of his team’s hitters, starters or relievers. The logit regression isn’t linear — which makes it  difficult to read too much into the exact numbers — but the coefficient on reliever WAR is extraordinarily large when compared to the coefficients on hitter WAR and starting pitcher WAR.

GM_fired = f ( -.133 – .044*WAR_hitters -.093*WAR_SP – .203*WAR_RP)

Using the ratios of the coefficients in these two equations isn’t a perfect mathematical approach, but it does tell a compelling story:
Hitters: .047
SP: .090
RP: .149

The coefficients of WAR on winning are comparable, but the coefficients of WAR on being fired are very different. Reliever performance factors into a GM’s ability to keep his job far more than the batters he signs.

On the other hand, when I run the same two regressions for rWAR, we see entirely different results.

Wins = 42.2 + .94*rWAR_hitters + .89*rWAR_SP + .97*rWAR_RP

GM_fired = f ( -.636 – .074*rWAR_hitters -.094*rWAR_SP – .043*rWAR_RP)

The same tables are quite different, as well.

rWAR Group Avg. if GM Not Fired Avg. if GM Fired Ratio Rank if GM Fired
rWAR Hitters 17.5 12.2 70% 20.7
rWAR Starters 8.6 5.5 64% 20.6
rWAR Relievers 3.3 2.2 67% 19.4

Interpreting rWAR numbers is difficult, because I’m making an adjustment for draft picks that I do for fWAR. Using fWAR to develop a way to evaluate draft pick costs, the replacement level yields a pretty natural 10% discount rate. Teams behave as though they value wins a year from now 10% less than this year. So even though they somewhat value the WAR from future draft picks, they value them significantly less than WAR in the coming year. Using rWAR, though, I actually got a negative discount rate. That seems unlikely since teams probably don’t value future wins more than present wins — but that’s what the model suggested. Not only that, but the price of rWAR would probably not be linear using rWAR. It would be slightly lower from that of superstar players. In essence, the model would need to be reworked to accommodate rWAR, and that might suggest that the fWAR results are more likely to be accurate.

Like the previous two articles, it’s OK if you come away with more questions than answers. We’re dealing with a rough approximation of an increasingly complicated labor market, and definitive conclusions are hard to reach. What these tables do suggest is that there’s either room for improvement in the way managers put together their teams or in sabermetrics, as a whole. As we’re discovering, the answer is probably somewhere in between.




Print This Post



Matt writes for FanGraphs and The Hardball Times, and models arbitration salaries for MLB Trade Rumors. Follow him on Twitter @Matt_Swa.

76 Responses to “Positional Differences in the Price of WAR”

You can follow any responses to this entry through the RSS 2.0 feed.
  1. Dekker says:

    How about adding more leverage into reliever WAR depending on the inning they pitch? WARP calculations notwithstanding, it’s pretty clear that fWAR undervalues relievers that are constantly put into high leverage spots in 8th and 9th innings.

    For example, if Dr. Genius runs the bullpen and constantly pitches a guy named O’Flaherty in the 7th, Venters in the 8th, and Kimbrel 9th, then their value should be tied to the inning leverage they’re used in.

    Vote -1 Vote +1

  2. db says:

    Interesting idea. I will note that your article is yet another example why fWAR for pitchers is not a good measurement. We don’t use batted ball statistics for hitters in fWAR and it is inconsistent and frankly not very helpful that fWAR does so for pitchers. WAR is, or in my view should be, a restrospective measurement. FIP and xFIP I find are more helpful for predicting the future than measuring the past. (In other words, we can measure runs allowed by pitchers directly in rWAR, the only metric that means anything in terms of result, but fWAR seems to think that measuring components that lead to runs is more accurate. Why measure indirectly what can be measured directly? fWAR has no answer.)

    Vote -1 Vote +1

    • Matt H says:

      fWAR measures results, just not runs, because runs depend on defense. fWAR measures the things that a pitcher does that don’t depend on defense.

      Vote -1 Vote +1

      • db says:

        When a hitter doesn’t strikeout, walk or hit a home run, his result also depends on defense. Yet we measure hitter WAR by results, not batted ball types. It is inconsistent.

        Vote -1 Vote +1

      • TK says:

        Db, I think I agree with you, but it should be noted that true talent BABIP can vary a lot for hitters, but not for pitchers. I think a happy middle ground would be factoring in GB% in some way, because on average a GB is less damaging than a FB.

        Vote -1 Vote +1

      • MangoLiger says:

        db,

        The defense a hitter faces over a season is pretty close to league average. A pitcher’s stats have a persistent bias on account of always having the same defense behind them.

        Vote -1 Vote +1

      • Dan says:

        Why not measure using runs allowed as a starting point and then try to back out the effect of defense?

        Vote -1 Vote +1

      • Ed says:

        Dan – that’s exactly what Baseball Reference’s WAR does. They start with Runs Allowed (not just Earned Runs) and adjust it based on the team’s defensive rating.

        Vote -1 Vote +1

    • Guy says:

      ” Why measure indirectly what can be measured directly?”

      We’re trying to isolate the pitcher’s role in run prevention. Measuring the 3 true outcomes is in fact a way of doing this directly.

      Vote -1 Vote +1

      • Jason says:

        The problem is that most of the data is thrown away. The pitcher does play a role in the other outcomes too. Further, the pitcher doesn’t have complete control over walks, stikeouts and homeruns either (anyone who watched Sabathia repeatedly strike out Detroit hitters only to have the umpire award walks recognizes this). Hell, not even all homeruns are equal. A good proportion of doubles are hit harder than homeruns. There is no good justification for ignoring most of the data.

        Vote -1 Vote +1

      • NS says:

        ^DIPS. FIP.

        “We don’t yet know how to accurately quantify it” is actually a pretty good justification.

        Vote -1 Vote +1

      • Jason says:

        Actually, its a poor justification. You don’t know how to accurately quantify FIP either. When you know there is error, you are best to include as much data as possible.

        Vote -1 Vote +1

      • NS says:

        “You don’t know how to accurately quantify FIP either”

        Show your work.

        Vote -1 Vote +1

  3. Darren says:

    I looked at the differences between Starter and Reliever WAR, compared to Starter and Reliever WPA and the ratio is much lower for WPA. In fact I think WAR for RP is 30% of Starter WAR, but WPA is 70% of Starter WPA. Thus, GMs are paying for the Leverage and the “Closer Mentality” , rather than assuming they can get the same performance out of an existing bullpen guy who has pitched in lower leverage situations.

    Vote -1 Vote +1

    • Baltar says:

      WPA may have some value in describing what happened in a game, but it is has no value in describing how well a player played in a game.
      For example, if a player hits a bases empty HR to break a tie in a late inning and his team wins by one run, his WPA is huge; if he does the same in an early inning and his team wins by one run, his WPA is small. Yet, the one-run difference is affects the outcome exactly the same.
      Therefore, the later an event happens in a game, the greater the WPA (blowouts excepted). Since relievers play later in the game, their WPA will be higher for the same events almost by definition regardless of their actual effect on the game.
      Pinch-hitters also tend to have higher WPA’s relative to their actual production than starters. That doesn’t mean they deserve to be paid almost like starters.

      Vote -1 Vote +1

  4. Bryce says:

    This regression seems really critical. Why is WAR not calibrated so that these coefficients are all 1?

    Vote -1 Vote +1

    • Bryce says:

      Ok, so I suck at HTML. I was trying to quote:

      “Wins = 42.2 + .95*WAR_hitters + 1.03*WAR_SP + 1.36*WAR_RP”

      and ask why that regression isn’t the basis for WAR.

      Vote -1 Vote +1

  5. Matt H says:

    Great article, Matt. That is truly fascinating about GMs getting blamed for bullpens. You would think that managers would get blamed more than GMs, since managers have such direct control over bullpen usage. But maybe there’s a pretty strong correlation between managers getting fired and GMs getting fired.

    I also wonder if playoffs may play a factor. If relievers are more valuable in the playoffs, teams without good bullpens will do worse than fWAR suggests, and if a team with high expectations fails in the playoffs, the GM is likely to be blamed. Though, I’m not sure if this happens outside of the Yankees and Red Sox.

    Vote -1 Vote +1

  6. Wait A Minute says:

    This is a backwards way of looking at the market. Maybe fWAR positional values overcompensate, so it makes it seem that C is undervalued while 1B is overvalued.

    Frankly, I don’t like how some articles are written on the foundation of “fWAR is right, so let’s see how dumb GMs are!”

    Vote -1 Vote +1

    • Baltar says:

      If you have a better way of evaluating players than fWAR, please share it with us, along with some proof that it’s better.

      Vote -1 Vote +1

      • ben w says:

        i’ll take wOBA for my money, especially when evaluating hitters.

        1b/corner OF/DH show up disprotionately on wOBA leaderboards but because of their position(s), their value is docked. if you’re concerned with assigning value relative to position/location/whatever, go be a real estate agent. if you’re concerned with scoring runs, buy the wOBA guys.

        Vote -1 Vote +1

      • Oscar says:

        ben w: Do you even understand the concept of replacement value? That would be a good place to start when criticizing something called “Wins Above Replacement”. It is also why your snarky comment is wrong.

        Vote -1 Vote +1

      • ben w says:

        whoa whoa whoa. i’m not criticizing WAR or even WAR/$ per se. i’m saying that there are many ways to create value, and for me elite wOBA=elite $s because elite wOBA drives run production. i could care less how the elite wOBA producers who play 1B/DH/corner OF compare within an already elite subset of players. just give me the raw production.

        Vote -1 Vote +1

      • fjmanuel says:

        i love this argument.

        “if you don’t have a better model, you can’t criticize this one! lalalla shut up!!!”

        so fucking stupid.

        Vote -1 Vote +1

      • NS says:

        That isn’t the argument.

        Vote -1 Vote +1

      • Ronin says:

        I am with you fjmanual! Its like the people who say if you dont vote you cant complain but maybe I dont vote because I am showing my disdain for both canidates.

        Vote -1 Vote +1

      • Baltar says:

        fj and Ronin, I say what I mean and mean what I say. If you have to put your own words in my mouth to criticize me, you’re pretty pathetic.

        Vote -1 Vote +1

      • Baltar says:

        ben, wOBA is a large compoent of fWAR, as it should be.
        If you don’t understand why a good-hitting shortstop is more valuable than an equally or slightly better-hitting 1B and you don’t think defense and base-running have any value at all, your understanding of the game is pretty poor.

        Vote -1 Vote +1

      • ben w says:

        baltar, please don’t hilariously mischaracterize my argument.

        Vote -1 Vote +1

    • fjmanuel says:

      fuck and yes, dude. fuck and yes.

      Vote -1 Vote +1

    • Matt Swartz says:

      To “Wait a Minute”: you should reread the last paragraph– I said it could be informative both ways. I never said WAR was right absolutely.

      Vote -1 Vote +1

  7. Colin Wyers says:

    Matt, what are the average ages in each of your position pools?

    Vote -1 Vote +1

    • Ronin says:

      Good question, I wondered if perhaps this had something to do with the idea that the tougher positions tend to be filled by younger players that arent as likely to be in the peak of the $$$ curve yet, and then by the time they do attrition has forced some of them into the less difficult positions.

      Vote -1 Vote +1

  8. Ben says:

    I’m not sure what the mystery is here. It’s just the myth of the closer being stated in terms of $/WAR. Saves are neither particularly difficult to earn, nor all that deterministic in the outcome of the game … but guys who accumulate lots of them get paid handsomely. Thus lots of $ and relatively few WAR.

    I suppose that you can put that on the GM if you want, but the myth of the closer is much more deeply ingrained in modern baseball than just the big desk in the front office.

    Vote -1 Vote +1

    • NickH says:

      Here’s the thing, though – a GM may well realize that Prover Closers are overpaid relative to their WAR contribution, but then still overpay for that closer. How? That GM knows that baseball fans overvalue bullpen contribution, and consequently the GM’s job security is disproportionately controlled by the success or failure of the bullpen, so they choose to pay for a pricey closer – effectively serving as expensive insurance for the GM’s job security (at the cost of slightly reducing the team’s overall chances of success).

      Vote -1 Vote +1

  9. Jason says:

    I wonder why fWAR overvalues shortstops, second base and catchers so much…

    Vote -1 Vote +1

    • Jono411 says:

      or why teams undervalue shortstops, second basemen and catchers so much…

      Vote -1 Vote +1

      • Jason says:

        So teams should voluntarily pay their shortstops more?

        Vote -1 Vote +1

      • Jono411 says:

        No. But given the amount they’re willing to pay free agent 1B/DH/OF, they should be willing to pay free agent SS/2B more than they do.

        Vote -1 Vote +1

      • Bhaakon says:

        Because offense is easier to evaluate and ages better, so it is (or at least appears to be) a safer investment than defense. Which is why defense-first players–who disproportionately play up the middle–get undervalued.

        Vote -1 Vote +1

    • Matt Swartz says:

      Jason, here is a regression on wins for WAR by position:

      Wins = 42.62 + 1.46*NL + 1.06*WAR_SP + 1.44*WAR_RP + 1.37*WAR_C + 0.75*WAR_1B + 0.82*WAR_2B + 0.92*WAR_3B + 0.78*WAR_SS + 0.68*WAR_LF + 0.77*WAR_CF + 0.54*WAR_RF + 0.40*WAR_DH.

      So if you look at it that way, WAR would suggest that C, 2B, 3B, SS are at least as important to winning than 1B, OF, DH.

      Vote -1 Vote +1

      • Jason says:

        Matt, I understand that. However, WAR do not equal wins. WAR are a correlate of wins. Market value is also a correlate of wins. Your assumption is that your correlate (WAR) is better at measuring real world wins than the market. Using WAR to support your argument that WAR is the better correlate is tautological. Perhaps GMs do better than WAR at placing a value on players.

        WAR simplifies the world by making equivalent a lot of things that really aren’t equal. For example, wOBA treats 1.58 doubles as exactly equal to one homerun, yet they aren’t the same. The homerun always produces runs, while the doubles often times don’t (the variance in the outcomes are different). If GMs maintain more of the real-world complexity in their evaluations, they very well may do a better job than WAR. If this is the case, you ought to be asking what is wrong with WAR, not what is wrong with the GMs.

        Vote -1 Vote +1

      • Matt Swartz says:

        Jason: I’m saying that there is something to be learned on both sides. Reread that last paragraph. It’s very crucial to what I’m saying.

        Secondly, all I’m showing in that regression is that 1 WAR from each position leads to 1 win on average. The dependent variable is actual wins, so I’m not just saying “look how many WAR they have, and WAR is good.” I’m actually saying “WAR and wins move strongly together.”

        Now, what you seem to be arguing is that GMs are better at valuing 1Bs than WAR, you still need to explain why they are getting paid more cumulatively than other positions. Regardless of how you apportion credit for wins, there needs to be an explanation for the fact that 1Bs are paid more. Is replacement level for 1B too high? How would we find that out? I’m skeptical.

        Vote -1 Vote +1

  10. 300ZXNA says:

    Maybe reliever value is due to perception of the game due to selection bias of what is remembered? I remember back in the late 90′s when the Mariners had a good rotation, phenomenal offense but atrociously poor bullpen. With Ayala and Charlton it felt like they gave away 20 games a year, when the reality is that they only lost a handful more games. Losing a 5 run lead late in a game tends to be selected as an indelible memory more than being shut down for 9 innings by an ace, even though both lead to the same outcome. it is a far more maddening way to lose. So perhaps having a bad bullpen is an easy way to turn the executive/public perception against the GM, thus GM’s have inordinately high demand that makes RP seem scarce.

    Vote -1 Vote +1

  11. James M. says:

    WAR ignores leverage. Bullpens (well, setup men and closers anyway) can only be judged in the context of leverage. Therefore WPA would be a better way to evaluate the impact of relievers on game outcomes.

    Allow me to illustrate with my team, the D’backs. In 2010 they had the worst bullpen in the history of baseball. (And yes, the GM got fired.) Their WAR broke down this way:

    Hitters 28.1
    Starters 9.6
    Bullpen (2.1)
    Total 35.6

    If you apply your regression coefficients to these results you get an expectation of 75.9 wins. They actually won only 65 games.

    How did they do so poorly? WPA tells us that, although the hitters did pretty well overall, they didn’t hit in the clutch and cost the team 6.9 wins. The starters did OK, resulting in negative 0.7 wins. But the relievers racked up an amazing 8.4 losses, more than the hitters and starters combined. That’s the effect of leverage.

    Vote -1 Vote +1

  12. Darren says:

    Good analysis. However in WAR we are trying to peel away the luck associated with pitching in a high leverage situation to come up with talent. What we know is that typically relievers are less talented than starters, but teams are pitching less talented pitchers in high leverage situations that are more important in determining a win. I would rather know true talent than have that clouded by the context for which they pitch.

    Vote -1 Vote +1

    • Paul says:

      We know that relievers are less talented than starters? Sure about that, and how? What I see when I look at leaderboards are run values for pitches heavily favoring relievers, velocities, offspeed movement, etc. And when I watch actual games, I’m reminded over and over about how guys like Joakim Soria were converted to the bullpen because they couldn’t stay healthy. Pretty sure Madson was too. And Mariano. And Paps. Etc.

      Relievers aren’t as valuable by WAR in large part because they don’t throw very many innings, not because they are less talented. You’re confusing two important concepts.

      Vote -1 Vote +1

  13. sprot says:

    Everything must be regressed.

    Vote -1 Vote +1

    • JDNE says:

      SPRRRRRRRRRRROOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOT

      Vote -1 Vote +1

  14. sprot says:

    Someone should send this to GMs so that they stop paying so much for Reliever WARS!! This could be the new market inefficiencies.

    Vote -1 Vote +1

  15. Phils_Goodman says:

    Great article. My first thought is that players don’t really sign contracts on a $/WAR market (that’s just how FanGraphs likes to analyze them). Rather than just assuming that Tom Tango’s positional value adjustments are the Truth, let’s think about why these gigantic discrepancies exist, and what are the reasons (both good and bad). I think a lot of it can be explained in systematic WAR vs WPA gaps for certain roles. This should probably get us closer to Actual Wins.

    For instance, 1B and DH take gigantic positional deduction because they have the lowest replacement level. But as a group, 1B and DH are generally the very best hitters in the game, and hit with the most power. That’s not so easy to replace. If you go to any year, the annual leaders in WAR will be a diverse array of 2B, RF, CF, SS, 3B, 1B, etc. But if you sort by WPA, REW or even WPA/LI, you’ll find that the annual leaders are dominated by power-hitting middle-of-the-order 1B, DH, and corner outfield types.

    It’s very similar to the starting pitcher vs. reliever WAR/WPA chasm. The question is how much structural leverage value do you attribute to circumstance, and how much do you attribute to the player. It seems like MLB teams attribute a LOT more of that to the player than FanGraphs does. And there is some good reason for that. MLB late-relievers are the most effective pitchers on the planet (by rate). MLB 1B, corner outfielders and DHs are the best hitters on the planet. And those are the players who use a disproportionate amount of important leverage pitches and plate appearances.

    Vote -1 Vote +1

    • Matt Swartz says:

      I never said Tom Tango’s positional adjustment were the truth. But if you look at my numbers in the above comment, a regression of wins on WAR by position shows that positional adjustments are certainly likely to be fair or even understated. Runs are scarce– runs by any other source or not. If you go large on middle infielders and catchers, you actually are more likely to have a balanced offense. If you get a 4-WAR 1B and a 0-WAR C, they likely have a .370 wOBA and a .290 wOBA, while a 0-WAR 1B and a 4-WAR C probably are at around .320 and .340 or something like that.

      Vote -1 Vote +1

    • Ronin says:

      I tend to agree with this, I think the positional adjustments are out of whack because it assumes that elite hitters that play 1b/DH/LRF are as easy to replace as weak hitting infielders who flash good gloves.

      Vote -1 Vote +1

  16. Phils_Goodman says:

    Also, your findings make the Jose Reyes signing look like a little bit less of a bargain than I thought it was, if you compare his salary to SS Wins on the market.

    Vote -1 Vote +1

  17. Teams, managers, fans, etc all focus more on games lost near the end because these are viewed as “shoulda won” games.

    Managers use their BP’s in a similar fashion so the perception is that the difference is in talent.

    Given the sample sizes of relievers (far less IP), I prefer looking at shutdowns/meltdowns. Most don’t care how the results are achieved, just in the number of positive results.

    Vote -1 Vote +1

  18. Carlos Marmol and Valverde are good examples as both are famous for making the last inning “interesting”. When they pull it out, all negative is forgotten. When they don’t, they they are “cuss words”, fair or not.

    Vote -1 Vote +1

  19. CircleChange11 says:

    Is defense undervalued because it’s easier to find a good defender than it is an good hitter?

    Or because defense is harder to measure/notice than offense?

    Or because fans prefer offense?

    Vote -1 Vote +1

  20. Kevin says:

    “In reality, the fact that high-strikeout relievers have particularly good BABIPs could mean that their FIPs slightly underrate their skill levels.”

    I agree. Does anyone know if their is a significant correlation here?

    Vote -1 Vote +1

  21. Kevin says:

    Also, for the record, I disagree with the notion that WPA makes relievers inherently more valuable. Runs are runs are runs no matter what inning they take place in. A first inning bases-loaded jam is just “high-leverage” as a ninth inning one. Starters are more valuable because they face many more of these situations over the course of the season, and their higher WAR reflects that. WPA is time dependent, so relievers’ values are always going to be larger.

    Vote -1 Vote +1

    • Bill but not Ted says:

      Leverage is the power of influence over a desired outcome.

      It is driven by the time component. There is less opportunity to have another event or situation impact the outcome late in ball games.

      Compare it to trying pick someone up in a given night. The closer you get to the end of the night (9th inning) the more impact a significant event such as getting piss drunk/bad dance moves/saying something stupid (giving up runs) will have on a desired outcome of… holding hands (winning the ball game).

      If you get rejected early in the night (give up runs early in the game), you have an opportunity to get rejected again (score runs in the remaining innings, or not).

      Vote -1 Vote +1

  22. Matt Swartz says:

    BTW average age for my sample:

    C: 34.4
    1B: 33.8
    2B: 33.8
    3B: 33.5
    SS: 32.9
    LF: 33.8
    CF: 32.8
    RF: 33.4
    DH: 35.8
    SP: 33.1
    RP: 34.7

    Vote -1 Vote +1

  23. mttlg says:

    What I’m wondering about is the pool of available players. What’s the distribution of WAR for each position? If the pool of available talent at hypothetical position XB is a bunch of 1 WAR players, you’re looking at guys getting 1-2 year deals at $1-5 million per year, with the option of going with a scrub or AAA call-up at XB and putting the money into position YB, which has a bunch of players in the 3-5 WAR range making an average of $3 million per WAR more than XB. I would also be interested in seeing how actual replacement players compare to “replacement level.” If position ZB has a gap in the talent pool between -1 WAR and 1 WAR, there’s going to be an overpay to stay positive because there are no real replacements. A team has to put nine guys on the field, so sometimes they’ll hand out big bucks to make the most of a bad set of options. Maybe none of this comes into play, but I’m guessing that the difference between idealized statistical models and the limited data points of reality could explain some of this. And I get a bit twitchy when I see results without the raw data.

    Vote -1 Vote +1

  24. mcbrown says:

    Just catching up on my fangraphs reading as Spring Training arrives…

    I would note two things about your results:

    1. Despite the overall trend for “easier” defensive positions to be overpaid per WAR, I find it interesting that the opposite is true if you look at the outfield by itself – CF is paid more per WAR on average than RF and LF (and more than 1B even, but this is outside the scope of my comment). I’m not sure what that means exactly, other than that this might be a really complicated issue fraught with SSS problems.

    2. I have said before, and will continue to say, that I think that the “relievers are overpaid” argument generally ignores the obvious point that every RP would work to become an SP, even at the expense of becoming unemployed if they fail, if the upside for doing so were sufficient. Therefore relievers simply can’t be paid much less than starters in nominal terms, regardless of $/WAR, if MLB teams want to retain some number of “quality” relievers. The counterargument is generally that relievers can’t easily become starters, and yet still a few times every year, year after year, some try (e.g. Daniel Bard this season) and some even succeed. Imagine how many people would do so if the potential upside were an even more dramatic increase in salary than is possible today!

    Vote -1 Vote +1

  25. jdm says:

    Just saw this – great article – my thoughts are that since the role of closer is a much more binary event than any other position GM’s inherently overpay so that a Verlander 8IP 2H 15K game isn’t ruined by incompetency in the 9th. I would be interested in running some sort of regression in terms of team WAR, wins, and blown saves and see how much blown saves cripple an average team as this may provide more indication into GM’s valuation of relief pitchers. The other thing that may be confounding the reliever result is that I expect GMs of teams that are in the hunt to compete for playoff spots year in and year out are willing to overpay for relievers since they need depth and something like this could be additive to WAR without occupying a premium position. Another thing I’d be interested in knowing but would be fairly difficult but the incidence of extreme splits and $/WAR. For example a great lefty specialist might be paid more per WAR than an all around reliever so that managers can leverage that advantage.

    Vote -1 Vote +1

  26. I know I’m coming late into this, but I have some thoughts.

    First, I think leverage is an important consideration for the dichotomy of RP $/WAR. While a run is a run, no matter when it happens in a game, there is a huge difference between 1-0 after one inning and 1-0 after 8 innings. In the former, the other team still has 8 more innings to score one run. In the latter, only one more inning. Huge difference, for while the reliever is producing minimal WAR in the 8th and 9th (since only 1 IP), he’s actually contributing to a real Win, in the Win column, if he performs that in his role.

    Now, whether this explains the whole difference, I have no idea, but I do believe that it explains some, if not most, of the difference.

    Second, to see if there is any truth to this, based on the data available, it would have been useful if Matt could separate the RP data into two sets: closers and non-closers. Of course, each year, pitchers move from one category to the other (and back again!), so I think the classification has to be based on what the pitcher’s role was expected to be when signed, not what happens afterward.

    Vote -1 Vote +1

    • Matt Swartz says:

      WAR includes leverage, BTW. Check out Dave Cameron’s primers on WAR for the details, but it has to do with chaining basically.

      I’d like to check closers vs. non-closers but it’s pretty tough to gather any of this data since I’m pretty much marking things down by hand. I may try though. Thanks.

      Vote -1 Vote +1

Leave a Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>

Current day month ye@r *