FanGraphs Baseball


RSS feed for comments on this post.

  1. Fantasy owners do, too. Last year’s MR relief hero is almost always a bad investment on the rebound. Don’t chase last year’s Benoit, find the new one.

    Comment by pianow — May 12, 2011 @ 1:23 pm

  2. I definitely agree teams are overpaying medicore relievers. Take a look at the exodus of the Twins bullpen last year. Rauch, Crain, Guerrier, Fuentes. Collectivley, they signed for $39 mil total dollars and average $4 mil a year. Plus a couple team options for more. None of these guys are front line closers, they are simply solid set-up guys. It is crazy that these guys got not only those dollars but 2 got 3 year deals and 1 got a 2 yr + team option. Insane.

    Comment by nick5253 — May 12, 2011 @ 1:42 pm

  3. Leverage, especially in a short series, is extremely important. While a 5th starter may be more valuable to a team over the course of the season than a shutdown reliever, he is quite literally worthless in the playoffs. Teams with serious October asperations must have a high quality closer, and dearly want a shutdown set up man, as well.

    The scarcity of true shut down relievers tends to drive up the price, and teams also face pressure to prove to the fans they are serious about competing in the post-season in order to sell regular season tickets. This may account for some of what’s going on here.

    I have learned through the years however that you ignore the market to your peril. It seems to me that the free market of baseball is telling us one thing, and statistical analysis is telling us another, despite the embrace of most teams to sabermetric principles in general. This tells me that it might be time to revisit the high quality reliever debate in the sabermetric community.

    Comment by Dan in Philly — May 12, 2011 @ 1:48 pm

  4. Having a strong bullpen is very important, but they are very tough to put together. A lot of it is just random for the most part. As you see teams go out and sign guys with solid track records to big contracts and a lot of them get injured or suddenly fall off a cliff. You have to patiently acquire a lot of arms who do something right and put them in a position to excel.

    Comment by Pat — May 12, 2011 @ 1:51 pm

  5. “I restricted my sample to players who signed for over $1 million,”

    Why? This will obviously increase the average salary, which will increase the dollars per WAR. I don’t see a benefit to this restricting (other than making your point easier to make, which I would assume is not the reason you did it)

    Comment by Matt — May 12, 2011 @ 1:54 pm

  6. Next time don’t make your plot so obviously made from Excel, and make the original version big enough so that the one posted in this isn’t enlarged and pixelated.

    Comment by Louis — May 12, 2011 @ 1:54 pm

  7. WAR clearly undervalues relievers. Saving a run in the 8th is obviously more valuable than saving a run in the 3rd.

    Comment by Ryan — May 12, 2011 @ 1:58 pm

  8. A shutdown bullpen is very important in the playoffs, much moreso than in the regular season. Given the schedule, the Yankees can run Rivera out there every game if they want too, and when he was younger it could be for multiple innings each game. For that reason, likely playoff teams should be investing in premium bullpen arms. Maybe not to the extent they have been though.

    Comment by Dave — May 12, 2011 @ 1:59 pm

  9. Not sure that there is a lot a team can do if they don’t want to overpay relievers. Considering the entire market is inflated, if you want decent relievers, you have to overpay. Imagine a competitive team going into the season with a replacement level bullpen. Even if the team could theoretically make up for the missing 2-5 wins elsewhere, the fans would be outraged. And the draft and farm system only gets you so far, because relievers are overpaid by the arbitration system as well.

    I think the thing for smart teams to do is just to limit their risk by not giving out big multiyear contracts to these guys. Overpaying a few million on a 1 or 2 year contract isn’t really a big deal.

    Comment by Yirmiyahu — May 12, 2011 @ 2:08 pm

  10. I don’t know if you’re being sarcastic or not. But WAR accounts for higher-leverage situations.

    Comment by Yirmiyahu — May 12, 2011 @ 2:10 pm

  11. Nope, it’s exactly the same.

    Comment by Mike P — May 12, 2011 @ 2:14 pm

  12. “Where offensive WAR used wRAA and UZR, pitching WAR uses FIP. Based on how many innings a pitcher threw, FIP is turned into runs form, converted to represent value above replacement level, and is then converted from runs to wins.”

    Comment by Ryan — May 12, 2011 @ 2:16 pm

  13. The misguided comments of Dan, Ryan and Dave very probably represent the attitude of baseball executives which leads to over-evaluation of relievers, especially “closers.” Note all the unsupported, and often patently false, statements in these posts, e.g. “Teams with serious October asperations must have a high quality closer,” “Saving a run in the 8th is obviously more valuable than saving a run in the 3rd,” “A shutdown bullpen is very important in the playoffs.”

    Comment by GiantHusker — May 12, 2011 @ 2:20 pm

  14. But it’s obviously true for good relievers that are only brought in during close games.

    Comment by Justin — May 12, 2011 @ 2:23 pm

  15. WAR does incorporate leverage. Although not to the degree of the reliever’s pure leverage number because of chaining (

    Comment by Reed MacPhail — May 12, 2011 @ 2:23 pm

  16. You sure about this? In the library I couldn’t find any references to using leverage in the fWAR calculations.

    Using WPA seems to support the notion that relievers can be more valuable than their WAR suggests. 5 of the top 7 pitchers in 2010 by WPA are relievers.

    Comment by Justin — May 12, 2011 @ 2:25 pm

  17. The most obvious explanation, to me, is that teams MUST overpay relievers to compensate them for not being starters. I.E., if starting pitchers received significantly higher compensation than relievers of comparable “skill” (however measured) in absolute terms then every pitcher would want to be a starter. There is not enough specialization of skills for a “good” (again, however measured) pitcher to get “stuck” in a relief role if the rewards for starting are significantly higher. This is different from, say, football where specialization of roles is so great that significant pay gaps can rationally exist – a long snapper can’t reasonably switch to QB just because the pay is better.

    Comment by mcbrown — May 12, 2011 @ 2:26 pm

  18. The psychological impact cannot be overstated enough. It also extends beyond fans to the media and to the players themselves. No matter how much you might intellectualize the situation, giving up the winning run in the 9th is far more devastating in the clubhouse than giving up a single run in the 1st.

    For me, it comes down to changing expectations … a notion that is notoriously difficult to implement. We can all talk about how a run is a run but that simply isn’t the case in the real world. Players are not computers. They have emotions. Very few things are as deflating to a team and their players than giving up a late inning lead and taking a loss.

    Comment by BJsWorld — May 12, 2011 @ 2:30 pm

  19. Ok nevermind Reed answered my questions with the link he provided. Looks like WPA overvalues high leverage relievers because of the different replacement level.

    Comment by Justin — May 12, 2011 @ 2:35 pm

  20. Not really. A 5th starter isn’t used often in the playoffs. During the season Joe Blanton is more important than Ryan Madson, but during the playoffs Madson will almost certainly be more important.

    Runs late in games can be in higher leverage situations than runs early in the game ever can be. A run in 9th during a tie game is more likely to win the game than a lead off home run.

    Comment by DavidCEisen — May 12, 2011 @ 2:44 pm

  21. Very well said.

    Comment by BillWallace — May 12, 2011 @ 2:48 pm

  22. I am slowly but surely switching from being a classic baseball fan (who overvalues things like “clutch” and “grit” and “hot/cold streaks”) into a sabermetrician, and as such I absolutely love this site.

    I want to agree with the premise that teams are overvaluing relievers, but I just wonder one thing:

    A win in the playoffs is worth substantially more than a win in the regular season, right? In order to make the playoffs (i.e. “win” the regular season), one needs, generally, 85 or so wins, which means they can lose 77 or so games. In the playoffs, in order to win the world series (i.e. “win” the playoffs), you need 11 wins, and the most games you can lose is 3 out of 5, or 4 out of 7.

    A top notch reliever, it is suggested in this column, is worth 2-3 more wins than your average reliever. Those 2-3 wins are worth a lot more in the playoffs, when you only need 11 wins, or when you need to avoid losing 3 out of 5, or 4 out of 7 games, than they are in the regular season, when you need to win at least 85 games, or avoid losing more than 77 games.

    Doesn’t that mean that the statement, “A team with playoff asperations…” ignoring the egregious spelling error, is at least kind of correct?

    Comment by Daniel — May 12, 2011 @ 2:50 pm

  23. I don’t agree with what you’re saying. I think most relief pitchers do want to be starters but their lack of skill in comparison to starting pitchers is why they are relief pitchers. Obviously, some relievers do have the skill-set to be starters, but I think those pitchers are the exception and not the rule.

    Comment by Mcneildon — May 12, 2011 @ 2:52 pm

  24. I think that it’s a few things messing with the average.

    First, relievers are notoriously volatile. So when you find one that is consistent, like Rivera, then that’s an extremely valuable commodity because it’s so rare. How many relievers can say they’ve been consistently successful over even the last three years? So teams are paying for that reliability even though those pitchers don’t produce much more WAR than other relievers simply because the innings pitched numbers are so low.

    Second, it’s possible that the market for relievers is simply different. Teams have to pay a premium for relievers because they’re overvalued in the market with respect to their WAR. If you’re a GM that’s trying to win games, and you need a reliever, your choices are essentially: overpay or get a bad player. We saw this a lot in the last offseason; the early and expensive deals for setup men led to players and agents holding out for more money, which leads to a higher dollar/win ratio.

    Third, we don’t really see relievers “developed” in the way other players are; they’re usually failed starters who convert to relief in the upper levels of the minors or even the majors. But relieving and starting are fundamentally different roles, and this has two impacts: it makes player development people cautious about promoting a reliever from the minors into a major league role since they’re not as experienced as one might like, and it makes the pitcher need more development time that they probably didn’t get. Therefore the pitcher is inexperienced in their new role, since they “wasted” time as a starter, and they don’t get promoted on the same schedule as other players. Their development curves are therefore different, a lot of their potential is likely wasted, and their performance probably isn’t as good as it could have been. All of this leads to volatility and a lack of teams that are willing to improve their bullpens via the farm system.

    Fourth, older starters can convert to relief in an attempt to further their careers. But after a successful and lucrative career as a starter, they want good money to relieve, even though they’d be pitching a lot less. This also increases the supply in the market, which would lower prices if the psuedo-collusion of players holding out for more money wasn’t there. What it does do is reinforce the third effect, since developing relievers is now even less important since there are always more reliable pitchers to be found via free agency.

    Comment by quincy0191 — May 12, 2011 @ 3:03 pm

  25. GiantHusker is wrong:

    Comment by GVeers — May 12, 2011 @ 3:08 pm

  26. Thank you.

    Comment by Dan in Philly — May 12, 2011 @ 3:19 pm

  27. I have slowly switched from a classic baseball fan who overvalued things like “clutch” and “grit” and “hot/cold streaks” to a sabermatrician, and then just as slowly saw the severe limitations of such an approach and have been attempting to find some balance between the extreme empiricism of the saber crowd and the extreme rational approach of the “old school” writers. As such, I face ridicule from both sides who seem to think they know more than me (not naming names but I’m posing in this thread for a reason ;) ).

    And spell checking a dashed off internet post is a waste of intellectual resources… :)

    Comment by Dan in Philly — May 12, 2011 @ 3:26 pm

  28. Phillies are in the playoffs already?

    Comment by neuter_your_dogma — May 12, 2011 @ 3:53 pm

  29. It’s funny I’m a Tigers fan and they have 7 million a year invested in their closer, 16M for 3 years on a setup man, had to give up 2 first round picks on said players and have 2 former 1st round picks in it yet arguably their best reliever thus far is a guy that I think they signed for at or close to the league minimum. Yes small sample size but it’s just another example of how spending big on a bullpen can be a crapshoot.

    Comment by Matt C — May 12, 2011 @ 3:54 pm

  30. By construction this analysis has limited itself to a relatively elite set of relievers (those who both regularly pitch in the late innings, and who earn more than $1mm/year). It may be true that the average reliever can’t make it as a starter in the majors, but Derek Lowe (for example) would reject the notion that the same is true of top-tier relievers. Even if there are only a handful of top-tier relievers who could become starters given a reasonable transition time (and for the record I reject this notion, but that is a side point and one I cannot fully substantiate), that could be enough to ensure that no significant mean starter/reliever pay gap emerges, as average salaries on both sides will be heavily skewed by these players.

    (Now that I think of it, it might be more interesting to compare median starter/reliever salaries rather than mean. It could reveal a different picture.)

    Besides, if the “specialization” argument is taken to its logical end, then the average late inning reliever is less skilled than the worst major league starter, and every team could upgrade its bullpen by firing almost everyone and substituting retreads and minor leaguers, so long as every replacement they choose is a “starter”. Though I suppose once those starters get colored with the tainted brush of the “relief pitcher” they might lose all of their superior “starter” skills and need to be fired and replaced themselves… :)

    Comment by mcbrown — May 12, 2011 @ 4:04 pm

  31. I think this article and many comments are off-point: the best bullpens are built through obscure trades, pickups, and late-round draft picks, resulting in cheap, effective, and obscure bullpens. Look at the Padres, for example. KT dominated the reliever market for years, never overpaying, rotating guys in after trading some. Heath Bell is the best closer in baseball currently but wasn’t that lauded coming out of the minors, for example. Gregerson, Adams, Bell is the best 7 8 9 in baseball.

    So yes, teams in general may be overpaying but I think the smart ones are not, i.e. the ones incorporating saber analysis who know overpaying is not really necessary to have a good bullpen.

    Comment by Boo Cocky — May 12, 2011 @ 4:11 pm

  32. I’ll bet you $100 that they do.

    Comment by DavidCEisen — May 12, 2011 @ 4:21 pm

  33. But are most teams signing these relievers because of the value they will provide in the playoffs? That’s would be silly in and of itself, because many of these teams have no chance of making the playoffs. They should be signing guys that will help them _get_ to the playoffs, and there WAR is more relevant. Spend the money elsewhere.

    Which is to say that it would have to be a team-specific analysis going on, and I think the reasons given in the article are more plausible.

    Comment by Matt — May 12, 2011 @ 4:24 pm

  34. I think the problem with the “collusion” angle is that it assumes there is something special about the starter/reliever label that doesn’t apply to other positions. Imagine that instead of starters and relievers we were talking about second basement and shortstops. If I proposed that starting shortstops should band together and demand to be paid twice the going rate for starting second basemen, you would (rightly) point out that this couldn’t work, because there is enough overlap in skills between the positions that at some point in the development system (maybe even at the major league level for a number of players) teams could turn second basement into “good enough” shortstops and tell the shortstops to stuff themselves.

    I maintain that the same thing holds for pitchers. This puts pressure on reliever salaries both from above (teams can prevent salaries from getting out of hand by turning starters into relievers) and below (skilled relievers can refuse to accept less than comparable starters through the implicit “threat” of switching roles).

    Comment by mcbrown — May 12, 2011 @ 4:29 pm

  35. As if we need more fuel to this fire…

    The only RP the Indians bought this year was Chad Durbin…and he is also their most expensive. And need I bring up the Kerry Wood debacle?

    Comment by TribeFanInNC — May 12, 2011 @ 4:30 pm

  36. WAR doesn’t really apply to relievers — especially closers — for one simple reason: WAR hypothesizes theoretical wins added, while closers actually give you wins or losses.

    Let’s say you’ve got a one run lead in the sixth inning of a road game, and you bring in your middle reliever, who gives up two runs. Your probability of winning goes from, let’s say, 60% to 40%. Then, let’s say we have the exact same situation, but it’s the ninth inning. Your probability of winning has now gone from 60% to 0%. You have completely lost the game, with no opportunity to make it up on the back end.

    The problem with closers is that, unlike basically any other player on the team, their actions guarantee wins and losses, while other players’ actions merely impact the probability of wins and losses. No matter what your risk aversion level is, the work that’s done to get you to a 100% or 0% win probability is always the most important, even if it’s only a five percent swing.

    Comment by JimNYC — May 12, 2011 @ 4:41 pm

  37. The assumption implicit in your comment and the article is that baseball GMs are stupid. They are unwittingly throwing away a competitive advantage by using money paid to relivers rather than something else, which would give a better chance to win, or keeping it. Given that there are literally milions, possibly hundreds of millions of dollars at stake, it seems likely that GMs who make such mistakes are soon former GMs, as owners like money and try to make as much of it as they can.

    As far you your assertion that it’s silly for owners to build teams to win in the playoffs when they have no chance of making it, I would argue that most teams actually have a chance to make the playoffs, if you believe in odds. Take this:

    According to the stats guys, only a few teams have a less than 5% chance right now of making the playoffs. Let’s take the White Sox as a random example. Their current odds are about 19%.

    If the teams conclude that winning the playoffs odd’s are increased by having a high value reliever, they’ll calculate how much that series win means in terms of $$$. They will then multiply that amount by 19% and that’s the current year’s worth of that reliever.

    Also remember that when a team does not have a high value reliever, fans feel the ownership is not trying, and may turn on them (I’ve seen this myself), so there is money which might be lost by not signing one even if you are 100% convinced you are not going to make the playoffs.

    Comment by Dan in Philly — May 12, 2011 @ 4:50 pm

  38. I think the volitility of most relievers is the main issue. If you have a Rivera, someone who has a track record of good pitching, then yea spend away. But we just see so many of them go from night to day over the course of a season or two- in both directions.

    It seems to me that, for the most part, you find a live arm and spin the wheel.

    Comment by Chris — May 12, 2011 @ 4:57 pm

  39. Unless his name is Betancourt, or he pitches for SD :)

    Comment by descender — May 12, 2011 @ 5:02 pm

  40. I think–and please correct me if I’m wrong–that both the mean and median salaries are higher for starters than for relievers.

    Most veteran relief pitchers are failed starters. Some have failed in the Minors, while others have failed in MLB.

    Let me approach this another way: how many current relief pitchers on Major League rosters have demonstrated that they can succeed as a starter? There aren’t many. You used Derek Lowe as an example, but he is one of the exceptions I mentioned. Generally speaking, if a pitcher is good enough to start, he will, eventually. In fact, most relievers who may be good enough to start are young pitchers like Neftali Perez or Joba Chamberlain of a few years ago. But the relief pitchers that this article is referring to are veteran relievers who failed as starters at some point in their careers, or were deemed incapable of starting right at the beginning of their careers.

    Comment by Mcneildon — May 12, 2011 @ 6:48 pm

  41. Part of the reason closers are overvalued, I believe, is because of the nature of their failures. To a big picture owner, a couple million bucks lost to salary is dwarfed by his concern for the fans’ perception of the team as a winner or loser.

    Having a middling to poor closer who is perceived by the fan base as sub par or unreliable (an example that hits close to home is Kevin Gregg last year) will lessen fan goodwill by an amount that likely surpasses the inflated closer salary.

    Long story short: poor closers create a lot of anxiety in fans, and owners want to keep fans happy.

    Comment by Powder Blues — May 12, 2011 @ 7:03 pm

  42. Matt: another point–even teams that don’t expect to make the playoffs can sign free agent relievers with the intention of trading them at the deadline. Every year there is a market for relievers, and the returns are sometimes surprisingly strong. Matt Capps for Wilson Ramos comes to mind. And if they don’t get that return, then they can fall back on fishing for compensation picks. I’m not saying that justifies the expenditure, but at least there really is something to be gained for less competitive teams.

    Comment by Newcomer — May 12, 2011 @ 8:32 pm

  43. I’m not sure about mean vs. median – if I get some time perhaps I can try to find out.

    To say that all relievers are failed starters is tautological. ALL major league pitchers begin their baseball careers as starters (e.g. no talented high school pitcher is a relief specialist). The mere fact that pitcher is a reliever is certainly indicative of something about their skill set, but it cannot be prima fascia evidence that they “failed” as a starter. If it is, as I said before, then it must be true the worst starter is better than the best reliever, and I don’t think anyone seriously believes that.

    Also, the fact that reliever-to-starter conversions are rare is entirely consistent with my argument that salaries compensate relievers for not starting – there is SOME value in specialization, so as long as a pitcher is not being disadvantaged economically by their relief role there is little reason to take the risk of switching. I would only expect it to be common in a world where starting pitcher salaries were significantly higher in absolute terms than relief pitcher salaries. Only then would there be sufficient economic pressures for players to switch en masse.

    I think we DO agree that there would be a dropoff in performance for most relievers (even elite relievers) switching to a starting role, so let me see if I can pin down where we disagree. Suppose for the sake of argument that in today’s world, an A+ reliever is paid the same (on average) as a B+ starter. It would only make sense for an A+ reliever to switch roles if they had a reasonably good chance of being an A- starter or better; otherwise they can make the same money doing their existing job with no risk of failure.

    Let’s also suppose that in terms of $/WAR, the “fair” pay for an A+ reliever is the same as a D- starter. Now let’s imagine a world where teams actually did pay relievers this way. Suddenly, the hurdle for an A+ reliever to take the risk of trying to become a starter looks a lot lower – they don’t actually have to be very good to get a big pay raise! Also, the A+ reliever salary they are risking isn’t that high to begin with, so they have less to lose if they fail. Now if they have even a low probability of being a C starter it is a no-brainer – they have to switch. Also consider the development of minor leaguers in this world – every pitcher would resist the transition to the bullpen! Even if a minor leaguer had a clear path to the majors through the bullpen and little chance of being a decent starter, they might still be better served by taking their chances in a starting role – what is the upside to becoming a reliever if the very best they could hope for is to be paid like a lousy starter? Surely every minor leaguer thinks they can at least be a LOUSY starter in the majors, right?

    Ignoring the particular grades (they are just illustrative – I have no intuition for how an average A+ reliever would perform as a starter, though I think you and I can both agree there would be at least some dropoff in relative performance, maybe even a huge dropoff), do you think this framework makes sense?

    Comment by mcbrown — May 12, 2011 @ 9:42 pm

  44. Is variability of returns a possible explanation?

    I have not done any true analysis on this, but it seems to me that relievers are terribly inconsistent from year to year in terms of their WAR output. It would seem to make sense then that if a reliever were given a large one year, or multiyear contract, and then follow that up with a down year, the appearance would be that the teams had “overpaid” for that relievers marginal contribution to the teams WAR. Anyone who watched Brad Lidge struggle in 2009 in the first year of a new deal, coming off a perfect year in 2008, would see what I am getting at here.

    To me, it seems that relievers end up as relievers because they are too inconsistent to have been starters in the first place. Maybe what this data is showing is NOT that team continue to overpay for relievers, but that relievers, by nature, rarely perform up to the level they are expected to, since by nature they are streaky? Teams should therefore refrain from signing revilers to large, multi-year deals in the first place?

    I don’t know, just kind of thinking out loud here.

    Comment by Ryan — May 12, 2011 @ 10:39 pm

  45. The pitch count and unwillingness of teams to allow their starters to go deep into games have increased the value of middle relievers and closers over the last few decades. Whether they are “overpaid” is up for debate and not well-settled. One major flaw of sabremetrics is the inability to account for intangibles. Yes, intangibles do exist. If any of you have played serious organized sport, then you surely will understand. Not anyone with a live arm can close games. The 8th and 9th innings are often the most important part of the game and to have a pitcher who has the psyche (a word which will likely not be well-received on this site) to close the game and avoid demoralizing the team and that all important fan base is worth infinitely more than your numbers can capture. This is what teams pay for in a closer – skills and makeup. While there is certainly room for statistical analysis in sport, it would be foolhardy to ignore in any fashion the human element of the game. See Hong Chih Kuo.

    Comment by James-NYC — May 12, 2011 @ 11:27 pm

  46. I absolutely agree that the framework you presented makes sense. I think our disagreement stems from the differences in our perspective of how pitchers become relief pitchers or starting pitchers. I believe that you are asserting that relief pitchers choose to be relief pitchers whereas I think that they are basically forced into that role by decisions made by their teams. The decision to make these pitchers relievers, for the most part, is based upon the organization’s belief that they cannot succeed as a starter.

    You said ” Also consider the development of minor leaguers in this world – every pitcher would resist the transition to the bullpen!”

    I think this is in fact what happens. From what I understand, most pitchers do resist the transition to the bullpen.

    I did some brief research and found that in 2008, the average starter made over $4 million. In 2010, the average reliever made just over $2 million. I believe that those averages were mean averages.

    Here are the links:

    That difference in salary is significant enough that the average reliever would probably prefer to be an average starter.

    Additionally, I didn’t say that all relief pitchers were failed starters. I said “Most veteran relief pitchers are failed starters. Some have failed in the Minors, while others have failed in MLB.”

    I think if you were to look at every single Major League relief pitcher’s stats from both MLB and MILB you will find that a majority were starters at some point but did not enjoy success. I have not done this, but maybe I will give it a shot in the next 24 hours.

    Comment by mcneildon — May 12, 2011 @ 11:35 pm

  47. Let me add a small clarification that probably cannot be proved by simply looking at the stats of relief pitchers. I think that some of these “failed starters” I speak of didn’t necessarily perform terribly as starters in the Minors. I think that their teams decided at some point that they didn’t have the pitch repertoire and stuff to succeed as starters in the Majors despite performing relatively well as starters in the Minors.

    For instance, a pitcher who posts a mid 3 ERA in AA but only throws 90 MPH and has average secondary pitches probably can’t duplicate that ERA in MLB as a starter. Or, a pitcher with a mid 3 ERA in AA who has a 96 MPH fastball but very poor secondary pitches, or just one secondary pitch, also is probably incapable of duplicating that ERA in MLB as a starter. However, as relievers, they could succeed because they don’t have to go through a lineup multiple times. From what I’ve seen, a lack of multiple, quality secondary pitches is what separates a reliever from a starter.

    Of course, this is merely my understanding from what I’ve picked up following baseball. I apologize that I am not offering evidence that is more concrete.

    Comment by mcneildon — May 13, 2011 @ 12:06 am

  48. “Yes, intangibles exist, because I’ve played an organized sport, and I say so.”

    Well, i’m sold.

    Later you eggheaded, derek jeter hating losers!

    Comment by Chris — May 13, 2011 @ 12:14 am

  49. These are good points Dan, but I have two counterpoints.

    (1) This article is about “why do teams spend so much money on relievers.” One of your responses is, they do so because good relievers are really important. While that’s true insofar as it goes, one of the core problems here is that no one knows who the good relievers are. Reed touched on this at the end of his piece. Because of the small amount of work relievers do in a season, it’s very hard to predict how good they’re going to be in the coming year (compared to starters). This uncertainty, aka risk, should depress the price for high-end relievers — should make teams less willing to spend “a lot” on any given reliever. But for whatever reason it doesn’t.

    (2) You suggest that the decisions of baseball teams have some authority due to the fact that they are the product of a “free market.” But some markets are better than others at collective rational decision making. This particular “market” has only thirty participants, is closed to other entrants, and its participants have not-insignificant contractual ties to each other. Let’s not pretend it’s the Oracle of Delphi.

    I think I agree with you, though, that seemingly “unexplainable” discrepancies between analytical thought and MLB business practice deserve additional scrutiny on _both_ sides of the disconnect.

    Comment by Leo Martin — May 13, 2011 @ 12:57 am

  50. Macphail, your right, it is about prospect theory, but what is most important to teams is that FANS ALSO FEEL THE LOSES MORE. Owners behavior is more based on tickets sales. If fans feel 9th inning loses worse than first inning loses, then for the sake of revenue owners better pay for a good bullpen!

    Comment by kick me in the GO NATS — May 13, 2011 @ 1:35 am

  51. Bah. The Cardinals had five “devastating” blown saves early this season, but they’re clearly not too upset about it, as they keep winning.

    Comment by Brad — May 13, 2011 @ 2:01 am

  52. Using a FIP based WAR model for this exercise to evaluate relievers is a bit absurd… it’s like measuring a starters yearly value on 2 months of work to determine who’s being overpaid and by how much.

    To assume a 3 outcome model is measuring results accurately over a 60 inning period is a bit foolhardy. You have all sorts of issues over that small a period… strength of opponent, possible serious platoon variation, park effects (that aren’t correctly adjusted) and all sorts of small sample size issues which tend to ‘even out’ over time. And then the usual luck and defense over what amounts to the equivalent of 8-10 starts.

    Was Joba Chamberlin a better pitcher than Daniel Bard last year? (Joba had a lower FIP) Were they both approximately equal in value? (I think Bard was ~0.1WAR better)

    I’m curious as to what type of regression (if any) was done to come up with the $/WAR figures utilized.

    Comment by joe — May 13, 2011 @ 2:32 am

  53. I concur.

    Comment by Barry Zito — May 13, 2011 @ 2:34 am

  54. When was the last time a team won the World Series without a dominant closer? I honestly can’t remember. Off the top of my head there is Rivera, Wainwright, Papelbon, Lidge, KRod/Percival, Wilson, Wohlers, Nenn and Jenks. I can’t remember how good Kim was for the D-Backs(I just remember him blowing some in the World Series but don’t remember how he was during the year) or whoever the 03 Marlins closer was. But outside of that every championship team in recent memory had a dominant closer so I wonder if GMs feel that it is necessary to have one even though alot of those closers either came out of nowhere or weren’t highly paid.

    Comment by Matt — May 13, 2011 @ 3:56 am

  55. There may be some correlation to your closer performing and having a good team, but more to the point for me is that, for most guys, its really hard to ascertain who will have a good season.

    Comment by Chris — May 13, 2011 @ 4:13 am

  56. Except any reliever who is capable of starting almost invariably will. Teams know that starting pitching is much more valuable than relief pitching, so they’re going to pay even more for it. It happens that since starters pitch 2-3 times more innings but aren’t paid 2-3 times as much, the win/dollar ratio for starters is lower, but they do make more money on an absolute sense. I doubt there are many relievers that are fully capable of starting but choose not to; the prestige and the money are better, so why wouldn’t they?

    Any team that’s willing to sacrifice a starter to gain a reliever is most likely going to do so via trade, as that’s going to net a better value, or is going to sacrifice a pretty bad starter and gain a mediocre-at-best reliever.

    Basically the tradeoff for a player is: I can pitch 70 innings for $10M or 200 innings for $15M. You get more money per inning as a reliever obviously, but more money as a starter, and if a player can get more money they will. For a team, it is: I can take a two-win starter and turn him into a one-win reliever, or I can trade him for two one-win relievers, since someone always needs starting pitching. I pay the guy less if a turn him into a reliever, but I pay the same amount for one reliever if I trade him for two, and now I control the rights of another reliever, who I can either keep or flip if I need the salary space. For a team or for a player, it is always better to make starters instead of relievers. They know that; it’s the reason why teams don’t develop a lot of pure relievers.

    Comment by quincy0191 — May 13, 2011 @ 4:25 am

  57. Well said, even though I don’t completely agree with you, your argument is clearly explained and well thought out. You clearly know something about logical argument, my compliments.

    Comment by Dan in Philly — May 13, 2011 @ 8:17 am

  58. The question is not whether there are “many” relievers who are capable of starting, or whether starting pitching is more valuable than relief – of course it is. The question is about why there isn’t a larger pay gap between elite (late inning) relievers and starters. I claim that a large pay gap can’t exist if even a modest number of elite receivers could have been capable of starting had they insisted on it at some point in their career. Someone above cited Joba. Papelbon also came up as a starter and had to be talked into a relief role. If pitchers like these had even a low likelihood of being successful as starters, if the pay gap were HUGE between starters and relievers why would they allow themselves to be made into relievers?

    Comment by mcbrown — May 13, 2011 @ 8:22 am

  59. I think you are absolutely correct about why minor league starters become major league relievers – poor secondary pitches, low velocity, etc., which may not be as much of a factor in relief. However, my point is that if the economic incentives were sufficiently skewed towards becoming a starting pitcher, then it might not matter whether a minor leaguer can duplicate his performance in the majors – even if he will be much worse in the majors than he was in the minors it could still be worth it for him to stick with starting if the pay gap is sufficiently large.

    I think this is getting muddied by the fact that we’re talking about baseball players. Consider me instead. I’m good at math and bad at sales. As it stands now, my non-sales job pays $X. Good sales guys earn X*1.5, and average sales guys earn X, about the same as me. I’m happy where I am because I’m not good enough to reach that 1.5X in sales, and I might struggle to even get X. But if good sales guys earned X*5, average sales guys earned X*2.5, and the worst of the worst salesmen earned X*0.9, then even if I’m lousy at sales it is worth it for me to change roles and take my chances since my downside is so limited and the upside is huge. So in order for my company to keep people like me happy in my role, they can only allow the pay gap to be so large. The question is not whether I can be as successful a salesman in relative terms as I am as a quant (I certainly can’t), it is whether I can be a lousy salesman and still earn more money by making the switch.

    By the way, I think those links cite mean salaries, and the ehow data is for all relievers, not elite relievers. I certainly agree with you that the “roster fillers” in the bullpen are lower quality pitchers overall than either starters or late-inning relievers.

    Comment by mcbrown — May 13, 2011 @ 8:43 am

  60. Gambling is illegal at Bushwood sir.

    Comment by Neuter Your Dogma — May 13, 2011 @ 8:50 am

  61. While removing long relievers and relievers who signed for less than #1million accomplishes the desired effect of limiting the scope of the sample to your target group, it also has the effect of skewing the results in your favour and in the end proves nothing. WAR is based on $ and results of all players, not just on a particular group. You could perform the same excercise that you did for players at any position and come to the same conclusion. What if you chose outfielders and limited the sample to starting OFs making over $1M, you would find that outfielders are overpaid. Same for any other position because the excluded group is factored into WAR. Generally, an analysis at any position will show that players who have completed 3 years of service are overpaid compared to WAR, unless you remove fringe players (such as long relievers, 5th starters, back-up catchers, 4th outfileders and utility infielders). Likewise, an analysis of players with fewer than 3 years of service (or who signed a long-term deal prior to completing 3 years of service) will show those players of being underpaid vs. WAR (i.e., the players we refer to as being bargains).

    Comment by Steve — May 13, 2011 @ 10:26 am

  62. I agree. Most of these relievers are signed for what they’ll provide in the regular season. No one will argue that people like Jesse Crain or Matt Guerrier, to name just 2, are “shut down relievers.” At best, they’re nice relievers to have in the pen.

    I think a lot of this is based on the notion that people, GMs and fans, prefer certainty to uncertainty. It’s the same idea that informs the decision by many to go w/ veterans over rookies. Someone like Crain is seen as a sure bet to be decent whereas going w/ someone unproven may yield awful results. Unfortunately, the “certainty” often isn’t nearly as “certain” as many believe it to be.

    Comment by chuckb — May 13, 2011 @ 11:05 am

  63. What is the replacement level that is used? Replacement level in reality would be the actual replacement. IE If Russell Martin dies, the Yankees… Bring up Montero…? etc etc. But that type of nuanced analysis is cumbersome and unique.

    Comment by Garrett — May 13, 2011 @ 12:01 pm

  64. Mariano wants to mention he hasn’t started walking to Cooperstown yet.

    Comment by Garrett — May 13, 2011 @ 12:05 pm

  65. None of this touches on why WAR/$ is a pretty terrible way to analyze FA signings. Clearly if the Yankees sign Pujols and pay him 25m a year… OMG GREAT FUCKING IDEA.

    Comment by Garrett — May 13, 2011 @ 12:08 pm

  66. Ok, I think you’re analogy made your point clearer for me. I believe you are right in asserting that an elite relief pitcher will make more than a lousy starting pitcher. However, I don’t know if that’s the best way to look at it because there isn’t a market, really, for lousy starting pitchers. What I mean is that if a pitcher is consistently lousy as a starter, he will either be out of the Big Leagues or he will be converted to a reliever. Therefore, I don’t think we should look at it in terms of could an elite reliever make an equivalent amount as a lousy starter.

    However, I just did a little research by looking at Cot’s Baseball Contracts page. What I found there suggests to me that an elite reliever would choose to be a starter if he thought he could be at least an average to somewhat above average starter.

    I found that there are 23 relief pitchers currently signed to deals that average at least $5 million a year. The mean average of those annual figures is just under $8 million. The median is $7 million. For the sake of argument, let’s go with the mean as a reasonable estimation for what an elite reliever can expect to make.

    However, a starter doesn’t have to be anything close to elite to equal or exceed this level of annual compensation. Here are some examples:

    Joe Blanton-$8 million/year
    Carl Pavano-$8 million/year
    Jake Westbrook-$8 million/year
    Scott Kazmir-$9 million/year
    Ricky Nolasco-$9 million/year
    Randy Wolf-$10 million/year
    Kyle Lohse-$10 million/year
    Aaron Cook- $10 Million/year
    Ted Lilly-$11 million/year
    Bronson Arroyo-$12 million/year
    Carlos Silva-$12 million/year

    None of those pitchers would be considered elite starters and some of them might even be considered average or worse. Kazmir and Silva would be considered lousy.

    To use your grading terminology, (and we both understand that these grades are arbitrary) I think these figures demonstrate that an A reliever could expect make as much as or more money per year if he was a B- or C+ starter. Obviously, my starter examples were cherry picked to some degree, but nonetheless, they demonstrate that you don’t have to be a great starting pitcher to make more money than the average elite reliever. Therefore, I believe that this would be strong enough motivation for a reliever to want to be a starter if he could.

    Comment by mcneildon — May 13, 2011 @ 12:43 pm

  67. While relievers are clearly overpaid, I think it’s also pretty clear that the sabermetric community undervalues the importance of having an “anchor” for your bullpen, a guy like Rivera who is so dependable and intimidating that he literally changes the psychology in any game both for the rest of his bullpen and for the other team as it enters the late innings of a close game.

    Maybe we can actually call the whole phenomenon of overpaying relievers as the “Rivera” effect. Teams are so desperate to find their own Rivera, that they’ll ante it up for anyone who shows that kind of potential at any point. What they don’t realize is that there’s only one Rivera. Maybe if Rivera had blown his arm out 10 years ago, we don’t see so many GMs dreaming on other guys being able to imitate him.

    Comment by everdiso — May 13, 2011 @ 1:33 pm

  68. Your analogy, not “you’re analogy.” How embarrassing.

    Comment by mcneildon — May 13, 2011 @ 1:39 pm

  69. JimNYC’s comment is really interesting in light of MacPhail’s prospect theory explanation for reliever salaries. Psychological research has shown that we are significantly more sensitive to changes in probability at the margins (ie 95-100%) than we are in the middle of the spectrum (40-60%). As a result of being risk averse, we do not like uncertainty in the slightest, and any change from certainty to uncertainty or vice versa registers on our psychological radar. To bring this back to baseball (and propose another potential explanation for this effect), as JimNYC points out, closers and late-inning relievers are the ones who bring a team’s win probability from uncertainty to certainty (going from something to 0% or something to 100%). By changing win probability at the margins, closers and late-inning relievers register more strongly on the psychological radar of GMs. As a result they place a greater value on them than they deserve.

    One other thing related to prospect theory: MacPhail’s point about the availability heuristic is really interesting. This reliance on availability is probably driven by base-rate neglect. GMs neglect to account for the average performance of a reliever in their evaluations, and instead rely only on the mentally available, high-quality performances. They take these, as well as small sample size stats, to be representative of general reliever performance and overpay as a result. If they were to take into account the base-rate numbers of the general population of relievers (or that of an individual reliever, given he has thrown enough innings in his career to constitute an acceptable sample size) they would be more likely to see the volatility of relievers and the risks associated with multi-year deals.

    Comment by Adam K. — May 14, 2011 @ 12:27 pm

  70. When I go to games and my team gives up 6 in the first inning then loses 6-2, I remember that game as a boring bummer. When my team enters then ninth up 2-1 then loses 6-2 it makes me mad, and I leave the game pissed off. I take that games as a hard loss. I do not believe I am alone in feeling these reactions. Plus, this is a clear example of prospect theory.

    Now owners know that a late inning loss makes fans more unhappy than and early inning loss all else the same, so owners will pay extra to avoid late inning losses. This is the entire reason why relievers are paid more than this site thinks. And frankly, that is an efficient outcome if owners are maximizing fan pleasure rather than wins.

    Can this discussion now be over. The reliever market is very efficient if owners are maximizing fan pleasure rather than wins.

    Comment by kick me in the GO NATS — May 15, 2011 @ 3:19 am

  71. I am really not as well familiar with this matter but I do wish to stop by blogs for layout tips and fascinating topics. You honestly described a topic that I frequently really don’t care considerably about and crafted it quite fascinating. This is a good weblog that I’ll be aware of. I currently bookmarked it for long term reference. Lung Cancer Symptoms

    Comment by Lung Cancer Symptoms — May 30, 2011 @ 6:34 am

Leave a comment

Line and paragraph breaks automatic, e-mail address never displayed, HTML allowed: <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 *

Close this window.

0.121 Powered by WordPress