FanGraphs Baseball

Comments

RSS feed for comments on this post.

  1. A good closer is part of Nate Silver’s Secret Sauce calculation for post-season success, so I do think there’s relief value above and beyond WAR. Of course, it’s debatable how much value this provides, and you need to make the assumption that the team is going to make the playoffs to begin with.

    Comment by Jayson — January 18, 2010 @ 11:44 am

  2. I wonder if the WAR calc is right and closers are less valuable or whether the closers are valuable and the WAR stat is faulty?

    Comment by Fantasy Alpha — January 18, 2010 @ 11:53 am

  3. Isn’t this exactly what the post is about? Dave does an excellent job of making the case for the former over the latter.

    Comment by notdissertating — January 18, 2010 @ 12:02 pm

  4. Doesn’t it work the other way as well? When you sign a free agent first baseman, you get better at first base. When you sign a free agent closer you get better at closer, set-up guy, middle relief and mop-up.

    Comment by Jason — January 18, 2010 @ 12:21 pm

  5. Wow, if only there was somewhere on the internet where you could find somebody writing a well-reasoned article which made the arguments for one or other side of this debate?

    Comment by Felonius_Monk — January 18, 2010 @ 12:21 pm

  6. This makes no sense… IF WAR is for wins above replacement, and your trying to say that a closers replacement is a setup guy, that means that setup guys have to worth 0 WAR..

    If the setup guy himself is worth 1 WAR, then he isnt a replacement player… the closers value would be above him…

    So you should have to value a closers leveraged innings fully since the guy replaceing him is likely to be worth something above an actual replacement level pitcher..

    Comment by Jeff — January 18, 2010 @ 12:22 pm

  7. The difference is you don’t typically have another first baseman on the roster who is close in value to the guy being brought in. In such cases when a team does, the staff here typically takes that into account.

    Comment by Kevin S. — January 18, 2010 @ 12:23 pm

  8. By this reasoning ALL SETUP GUYS are only replacement level…

    When valuing a 1B’s WAR you value it above a replacement level 1B, NOT someone worth 1WAR…

    Your devaluting of closers WAR is just wrong IMO…. It needs to show the value above replacement level.. NOT THE VALUE ABOVE THE NEXT BEST RELIEVER AVAILABLE!!!!

    Comment by Jeff — January 18, 2010 @ 12:25 pm

  9. Well, I don’t have a major problem with giving a closer half the credit for his leverage index surplus, but i think you are forgetting that setup men also pitch in fairly high leverage situations. The closer should get maybe 25% of the that marginal increase in the leverage index. Then there is the secondary setup man, who also pitches in fairly high leveraged innings, mainly the 7th inning of very close games, sometimes the 8th or 9th inning of tied games. The closer should get some credit for the marginal increase in the leverage index for those innings as well, maybe 12.5%.

    A “classically designed” team will have a drop-off in talent between each reliever. The closer will be the best, the main set-up man the second best, the secondary set-up man the 3rd best, and so on. But when a closer gets removed from the picture, the team not only suffers from the loss in value from the closer to the set-up taking over, but also from the set-up man to the secondary set-up man who has to take over for him, and from the secondary set-up man to whoever else is filling out the bullpen.

    I think this needs to be accounted for as well. It is similar to an offseason where a team replaces a below average OF who is batting 8th in the order to a stud who will bat 3rd in the order. That stud will only be marginally better than the previous #3 batter, but he pushes everyone down a spot in the order (basically), so there are lots of marginal gains that need to be accounted for.

    Comment by Keith — January 18, 2010 @ 12:28 pm

  10. Dave’s article made perfect sense, this comment does not.

    Comment by Not David — January 18, 2010 @ 12:28 pm

  11. Chaining makes sense. Your replacement level player isn’t getting your high-leverage innings, he’s getting the low-leverage ones. There is an argument to say that there’s a better equation for chaining than the way it’s done here, but I don’t think anyone has published anything yet.

    If my first baseman was injured, his replacement would likely have to be the replacement level AAA scrub (or the almost equivalent bench player, since most bench players are close to replacement level regulars). That’s why we give full value to position players.

    I believe Dave covered that already, but it bears repeating.

    Comment by Michael — January 18, 2010 @ 12:32 pm

  12. Right… but he’s saying signing a closer isn’t as valuable as it might seem because what’s being replaced is the best reliever on the team prior to the signing – someone who is (presumably) much higher quality than what we normally think of when we say “replacement-level” player.

    When tabulating the benefits of signing a closer shouldn’t we add the marginal benefit of the newly signed closer over the previous closer PLUS the marginal gain from having the previous closer take over for the previous set-up guy PLUS the previous set-up guy take over… (etc.)?

    Comment by Jason — January 18, 2010 @ 12:32 pm

  13. That goes towards the idea of fixing the chaining equation. For now, this is a good enough approximation, but a study could be done for this (haven’t designed it in my head yet).

    Comment by Michael — January 18, 2010 @ 12:33 pm

  14. Agreed, but until there is a better formula, there will be a lot of debate. Dave might think 50% of the additional leverage index value is fair. Someone else might think 75% of that is fair. It probably makes quite a bit of difference.

    Comment by Keith — January 18, 2010 @ 12:36 pm

  15. Additionally there is the much disputed notion that certain high quality relievers (probably most), have a more difficult time pitching in the closer spot than in the set-up spot, more than the difference in leverage value would suggest (and whether the leverage index would affect pitchers at all is debated as well).

    Comment by Keith — January 18, 2010 @ 12:39 pm

  16. The way they are calculating it makes perfect sense. I don’t think some people understand chaining correctly. The closer’s “replacement” is not a person but a combination of changes throughout the bullpen. The “replacement player” = (Closer-setup) + (Setup-7th inning guy) +…….(Mopup-Real Replacement Player)

    So the idea that a closer is replaced by just the setup man or just the real replacement level player is incorrect.

    Comment by timmy013 — January 18, 2010 @ 12:39 pm

  17. Right, but most of the innings where you’re getting better aren’t as high-leverage as the innings where your closer is pitching.

    Comment by micah — January 18, 2010 @ 12:40 pm

  18. But it doesn’t appear they have actually gone through that calculation.

    Comment by Keith — January 18, 2010 @ 12:40 pm

  19. Good post Dave, and I think your argument is a fair one.

    FWIW, I think the main arguments that WAR under-rates relievers are that:

    1. (as Jason says above) chaining kinda works both ways, and it also assumes that all other positions will be replaced by a replacement-level player (which is not true, as there are more players who are better than 2-wins-below-average than there are roster spots in major league baseball).

    2. (And this, I think, is the argument I’d subscribe to) – the replacement level for relievers is broken. Arbitrarily setting replacement level a run or so higher for a reliever makes sense from the point of view that “starters moving to relief generally experience higher FIPs”, but I think the rationale behind it is hazy at best. I haven’t run any numbers on this, so it’s totally my opinion, but I would imagine your typical replacement-level major league starter (let’s say, Tim Redding) would experience difficulties in moving to a bullpen full-time and putting up a greatly improved FIP (indeed, Redding’s been a bit above-replacement in his career as a starter, but, admittedly in only 50-odd innings, has been below replacement level as a reliever). Also, the transition from starter to bullpen is not necessarily a reversible one and isn’t one that works in the same way for all pitchers.

    3. Leveraging, and the way in which it’s assigned a value, also seems something of an inexact science. I like Tango’s formulae and am not mathematically astute enough to be able to poke holes in it, but as far as I can see (and I don’t think I am mistaken here) the manner in which it is applied to calculations of value is, again, slightly arbitrary and lacks a real proof, especially when using a measure that is entirely based on “win likelihood” (Leverage index) as a multiplier for measures of peripheral performance that are entirely unrelated to “win likelihood” (FIP). I’m sceptical as to whether this argument has been resolved with an accurate answer, or indeed has been chewed over at all in a systematic way in the sabr community.

    Also, as different managers use their relievers in different ways, applying a simple “a closer gets an average of 1.8″ solution appears unreasonable. The value of a correctly-leveraged relief pitcher may well greatly exceed that, or fall below it. If we’re going to attempt to use measurements of leverage at all to value relievers, it seems more consistent to use it in measures of value for ALL players. Why should Todd Wellemeyer get as much credit for throwing three consecutive strikeouts in the 6th inning of a game in which the Cardinals are 20-0 down, as (say) Kyle McClellan does for striking out the side in the 6th inning of a game in which they are 3-2 down?

    I think all of these are valid questions, and people are correct to be answering them. Whilst it COULD be (and I think this is, at least to some extent, the case) that major league GMs (many of whom show proclivities that are pretty stupid a large percentage of the time) are over-valuing late-inning relievers, when teams like Tampa Bay, Baltimore and Atlanta are hugely over-paying for WAR from late-inning relievers, I think it’s reasonable to wonder whether the intrinsic uncertainties in our methods for calculating their value are skewing what we see.

    Comment by Felonius_Monk — January 18, 2010 @ 12:45 pm

  20. But when a team signs a closer, they are paying him for his full leverage. They are paying him for that role and will give him the full leverage, so why should the team take into account bullpen chaining when paying that specific player? Teams have backup infielders and outfielders that are not replacement level, so when a team signs a 2B or SS or CF, they are not necessarily adding the WAR of the said player b/c they would have .5 to 1 win players to replace those guys. If that is not factored in when looking at position player salaries then why should bullpen chaining be accounted for when a team signs a closer? In addition to that, signing that closer replaces a replacement level reliever that would have been in the 7th spot in the bullpen. That would not make that big of a difference, but it should be worth a few decimal points worth of WAR. It has also been shown that good relievers can and often do outperform their peripherals and that won’t be captured by FIP and a RA or ERA WAR may be better suited to look at certain relievers. Finally, this is the article I am most familiar with on bullpen chaining:

    http://www.beyondtheboxscore.com/2009/4/29/856308/bullpen-chaining-and-reliever-war

    However, it assumes that the 2nd best bullpen guy will be a 3.75 ERA-type pitcher. What if said team does not have that guy in their bullpen? Then the value of the signed closer is worth more to that team then it would be to other teams.

    Comment by Scottwood — January 18, 2010 @ 12:51 pm

  21. I agree to some extent, but perhaps this shorthand is over-valuing starting pitchers and position players (and, by that means, under-valuing relievers). The definition of “replacement-level” for these roles is someone who is 2 WAR below average for that position or role, and I believe it has been demonstrated that there are more players who EXCEED that level than there are roster spots in major league baseball (for instance, take a look at CHONE’s projections for 2010; you’ll find each team has considerably more than 25 guys who would be pro-rated to be above-replacement, albeit by only marginal amounts, over the course of a full season).

    Therefore, the average “replacement” for an injured SP or a position player will, in reality, be above the magical figure of “20 runs below average”.

    We can argue that an injured closer is not replaced by a 0 WAR player, but the same is true for ANY player (although perhaps not to the same extent). Arguments over how to assign replacement-level have raged over the years, and I’m not convinced personally that the level set for relief pitchers (and indeed its modification by “chaining”) values them correctly – I’m not saying I conclusively DISAGREE, but merely that this still remains a matter of some conjecture, not objective fact.

    Comment by Felonius_Monk — January 18, 2010 @ 12:52 pm

  22. Well that can’t be right. A sum of all those differences wouldn’t make sense, either.

    Comment by aj — January 18, 2010 @ 12:57 pm

  23. All of this assumes that the only value a player brings to a team is in wins. I think closer value is an area of analysis that is most damaged by the assumption that baseball teams are trying to maximize wins. They simply aren’t. Baseball teams are trying to maximize revenue. That wins correlate so closely with revenue means that we normally don’t have to deal with the distinction. I’d argue that assessing closer value is an area where the distinction is important.

    Management takes a lot of pain from late inning losses. Late inning losses are judged very very harshly by fans. Listen to the call-in shows. Or just note how you feel after watching your team lose in the 9th because Charlie Sheen can’t find the strike zone. I’d think that that fanbase pain would have a not insignificant effect on viewership and ticket sales. I don’t think this is specific to a certain kind of team – I’d argue that Yankees and Pirates fans equally hate watching blown saves.

    I agree totally with Dave that WAR (mostly) correctly assesses reliever contribution to wins. But I’m not so sure about the jump to the idea that there’s a market inefficiency to exploit. Elite closers have value to teams above and beyond their contribution to the win. And (by most reasonable accounts) there are very very few of them. That they are compensated accordingly shouldn’t be all that surprising.

    Comment by jay — January 18, 2010 @ 12:58 pm

  24. Your mistake seems to be coming from a belief that “replacement level” means “the level of the guy replacing him.” Just because a closer is replaced by the setup guy doesn’t mean the setup guy is replacement level.

    Imagine that the entire Yankees bullpen goes on the 60-day DL in a freak stretching accident and are replaced with low-cost pickups of the same AAA-average quality. The difference between Mariano Rivera and that low-cost pickup is your wins-above-replacement, not the difference between Rivera and Phil Hughes.

    Comment by Steven — January 18, 2010 @ 1:01 pm

  25. On AVERAGE, relief pitchers are very interchangeable. Cheap youngsters have good seasons all the time. But others don’t. And some of those that did have one good year don’t have another the next.

    What these teams are paying for in these closers like Valverde is reliability. (Lyon’s fat contract would be a bad example.) Relievers are notoriously unreliable from year to year… except for a select few. And managers want to know what they are going to get when the game is on the line. Call it a confidence score.

    So the dollar value of these high-leverage pitchers is not linear. WAR is drawn up against league averages, and the differences are even more muted when you factor in this chain effect. But even with the chain effect and comparing only to set-up men, there are only so many relievers that managers really trust, that have a high confidence score. Some teams are willing to pay for that comfort. Sure, they could roll the dice with any number of cheaper in-house options and quite possibly get the same result, but the confidence wouldn’t be there.

    In the Tigers’ case, they didn’t want to pay Rodney or Lyon, but minus those two… they did have much of a relief “chain” remaining, nor did they have much to choose from remaining on the FA market, except one high confidence pitcher.

    Comment by aj — January 18, 2010 @ 1:33 pm

  26. Does WAR take into account the scarcity of effective bullpen arms who can pitch in high lev situations…?

    Although their contribution may be overstated, it is still terribly hard to find GOOD relief Aces. Contrast this to guys who can CLOSE – which is something else entirely.

    Although most #3 starters accumulate more WAR per season than a Relief Ace, this doesnt mean that most of them could be converted to relievers and perform well there.

    I also think the conclusion to be drawn here is that small market teams shouldnt spend their money on RPs, but Im not sure that dynamic is the same for big market, big money teams.

    Comment by alskor — January 18, 2010 @ 1:35 pm

  27. edit: “…did[n't] have much of a relief “chain” remaining…”

    Comment by aj — January 18, 2010 @ 1:39 pm

  28. Kindof what I was hinting at… there is a curve for the value of a good closer just like there is a curve for the value of a win. This curve would be dictated by availability of reliably good pitchers and also their value to the team.

    The Valverde deal is frowned upon for being 7MM/yr while people like Abreu and the almighty FraGu make less while accumulating more WAR. But, despite the lack of OF depth on one team compared to RP depth on one team, productive OFs are actually more abundant on the FA market than reliable closers.

    Comment by aj — January 18, 2010 @ 1:46 pm

  29. Isn’t including leverage in WAR for relievers just like including it for hitters. Do we want to give extra WAR credit to the batter who drives in go-ahead run? We seem to be basing it on an assumption that pitching in high leverage situations (based on run differentials, not the quality of the opposition being faced — which is a problem in and of itself…) is somehow different fundamentally.

    When you sign a closer, you are signing him for the skills he brings to the table compared to those an alternate player could bring. The usage of those skills will determine the pitchers impact on win totals, but that usage is not inherent in the player himself unless we believe that it is genuinely more difficult to pitch in high leverage situations — not just that the production in those situations provides a greater contribution to the game’s outcome.

    WAR and it’s valuation should be based on a generic, context neutral production – those aspects of the players performance which are under his control times the frequency with which he provided it.

    Comment by Rick — January 18, 2010 @ 1:48 pm

  30. Uh… just so there’s no confusion, Im well aware WAR doesnt account for scarcity…

    I really do feel WAR alone doesnt account for the value of relievers. There is an inherent volatility in relievers and a scarcity of quality arms.

    The lack of WAR they CAN accumulate is a good reason not to drastically overpay, but its not the whole story.

    Its the same as the defensive value of catchers. Yes, we cant really measure it or its contributions to RS/RA, but that doesnt entirely encompass its value. Since it doesnt appear to show up in WAR it probably is a mistake to put too much money into it… but it would be foolish to ignore it entirely and base our decisions solely on WAR/measurable contributions to RS/RA. This is exactly the case with Relievers, IMHO (albeit they are somewhat measurable).

    Comment by alskor — January 18, 2010 @ 1:56 pm

  31. That’s all true. But how much better the first basemen is versus how much better your closer/set-up/MR/etc. are is the next question. And because of leverage and chaining, the bullpen upgrade is usually less than it appears on the surface.

    Comment by Sky Kalkman — January 18, 2010 @ 1:56 pm

  32. Why is leverage applied to the WAR for closers and setup men, but not applied to #3 and #4 hitters? The principle is almost the same, a manager uses his best players in the positions where he expects them to be most useful, but in one case the players get a bonus and in the other case they do not. Then when your #3 hitter goes down, you don’t replace his spot in the lineup with a guy from the bench or AAA, but more likely move everyone up and have your new guy hit 7th or 8th.

    The 50% thing is arbitrary, at implies a lack of faith in the concept.

    Comment by The Real Neal — January 18, 2010 @ 1:58 pm

  33. Try this for an empirical calculation:

    http://www.beyondtheboxscore.com/2009/4/29/856308/bullpen-chaining-and-reliever-war

    At this point, there’s no perfect theoretical calculation. But if you run a calculation using some common bullpen setups, you find that (1+LI)/2 is pretty good. Certainly room for improvement.

    Comment by Sky Kalkman — January 18, 2010 @ 1:59 pm

  34. To add on to this… we can look at starting pitchers and their roles through a real life example. When the Red Sox added John Lockey, CHONE projects him to be a 3.7 win player and he was paid accordingly. However, if Lackey gets hurt, he won’t be replaced in the starting rotation by a replacement level starter. The Red Sox have a back of the rotation that includes Buchholz, Dice-K and Wakefield. None of those guys are replacement level pitchers. If we assume that Wakefield is the odd man out (I have no idea if that is the case) and he replaces Lackey if Lackey goes on the DL, then the Red Sox don’t lose his 3.7 wins. Wakefield is projected to be a 1.5 win pitcher. The Red Sox also have Michael Bowden and he is another guy who is not a replacement level pitcher. So, if we apply the same thinking given to closers, should we dock Lackey roughly 2 wins b/c his real worth to his Red Sox team is not 3.7 wins? If we take that stance on him, then that changes the whole complexion of that FA signing and makes it look like a terrible overpay. This is essentially what we are doing by not giving closers full credit for their leverage when a team signs them for that role. If we did that for starting pitchers, then a lot of those FA signings would look like drastic over payments, as well. Just like when the Red Sox signed Lackey to be their starter, other teams are signing relievers to be in the closer role and are paying for them to perform in those high leverage situations.

    Comment by Scottwood — January 18, 2010 @ 2:00 pm

  35. That’s exactly what chaining is, yes.

    You can do the same thing for other positions, but it’s usually a much shorter chain, and because of the lack of leverage for positions other than bullpen, doesn’t matter *much*.

    Comment by Sky Kalkman — January 18, 2010 @ 2:01 pm

  36. “Although most #3 starters accumulate more WAR per season than a Relief Ace, this doesnt mean that most of them could be converted to relievers and perform well there.”

    Actually, they typically CAN be converted and be good relievers. Starters, on average, see their ERAs go down by one run when converting to the bullpen. So a 4.50 ERA #3 starter would be a 3.50 ERA reliever. Not an ace, but still pretty damn good. (And a true bullpen ace is more valuable than a #3 anyway.)

    Comment by Sky Kalkman — January 18, 2010 @ 2:08 pm

  37. I know this is probably heresy on this site, but I believe there is a psychological difference between Save outs in the 9′th inning and Hold outs in the 7′th or 8′th inning. Also, the difference in quality between the closer and setup man may be greater than it seems.

    When the Giants lost Robb Nen, they wandered in the Closer wilderness for several years with Tim Worrell, Matt Herges and then Armando Benitez. Even though the Save percentages weren’t all that much different, the timing and quality of the Blown Saves had a tremendous negative effect on morale, at least of the fans!

    It’s hard to put a finger on what’s different about Brian Wilson, but it is different and better. It may seem easy to replace a Closer, but Giants fans will tell you different based on the time frame between Robb Nen and Brian Wilson.

    Comment by DrBGiantsfan — January 18, 2010 @ 2:09 pm

  38. For hitters, there’s really no way to leverage their skills — you can’t choose give a hitter high LI plate appearances. But you CAN do that for relievers, making them more valuable. When a player provides a skill that can be strategically leveraged, it’s valuable.

    Comment by Sky Kalkman — January 18, 2010 @ 2:10 pm

  39. I think this is a good point. Hypothetically speaking, if I have someone like Tyler Clippard as your closer and sign Matt Capps pushing Clippard to SU and slotting every RP down one spot in their role, you have effectively improved every single RP spot in your pen. Now let’s say you callup Drew Storen mid season and slot him in at SU or CL, you have effectively improved every spot in your pen again pushing out your worst RP. Sure you can replace a RP relatively easy depending on your depth (take the Athletics during the past couple of seasons) but that doesn’t make them less valuable or necessarily mean every situation is the same. A team like the Royals for example would have one heck of a time replacing Soria I think.

    Every team needs at least 7 RPs at any time and while most of them wont pitch more than 60 innings, the collective group does pitch a lot of innings. Let’s take the Twins as an example, would you trade their whole pen (Nathan, Guerrier, Mijares, Neshek, Rauch, Crain, and Condrey) which comes out to be 4.4 WAR for Robinson Cano who was worth 4.4 WAR last year?

    It’s sort of funny I was looking at the WAR values for pitchers yesterday and noticed that it really doesn’t think much of even the best closers (a pretty good CL is worth about what a #4 SP is worth according to WAR and the very best RPs are worth about what a #3 is worth which is sort of amazing to me).

    Comment by jfish26101 — January 18, 2010 @ 2:17 pm

  40. Damnit, I absolutely hate the layout of replies on this site.

    My post was supposed to be in reply to Jason’s point about signing a CL improving your whole pen.

    “When tabulating the benefits of signing a closer shouldn’t we add the marginal benefit of the newly signed closer over the previous closer PLUS the marginal gain from having the previous closer take over for the previous set-up guy PLUS the previous set-up guy take over… (etc.)?”

    Comment by jfish26101 — January 18, 2010 @ 2:19 pm

  41. I was agreeing with you. :P

    Comment by aj — January 18, 2010 @ 2:25 pm

  42. Sure, Valverde’s reliable – reliable to not post a 2 WAR season. Reliable to post a not-so-impressive (for a reliever) FIP/xFIP in the mid threes. Reliable to be decent in AAAA. Not reliably good.

    Comment by Kevin S. — January 18, 2010 @ 2:27 pm

  43. Valverde outperforms his peripherals virtually every year and that is not captured in only looking at FIP. It is also not uncommon for good relievers to do that. His WAR the last 3 years at Rally’s site paint him in a much better light.

    http://www.baseballprojection.com/war/v/valvj001.htm

    Comment by Scottwood — January 18, 2010 @ 2:34 pm

  44. Just ask the Phillies how “easy” it was to replace Lidge last year…

    Some guys just can’t close.

    Comment by NEPP — January 18, 2010 @ 2:38 pm

  45. From a Fantasy perspective, relief pitchers have even less of an impact, which is why it befuddles me why people overpay for one category. You don’t see players shelling out top 100 picks for Juan Pierre.

    An extensive look at why Relief Pitchers are overvalued in fantasy is available here:
    http://gameofinches.blogspot.com/2009/03/saves_17.html

    Comment by David MVP Eckstein — January 18, 2010 @ 2:51 pm

  46. Yes, I know it’s not uncommon for relievers to outperform their peripherals, which is why I ranked them over the past three years with a minimum of 150 innings pitched by how much they out-performed their FIP. The guys who out-performed it the most? Tim Byrdak, Pete Moylan, Ron Mahay, Matt Guerrier, Ryan Franklin and JP Howell. Yes, there were some aces mixed in the top of that list, such as Nathan and Okajima, but the vast majority of the players who out-performed it by as much or more than Valverde were your run-of-the-mill guys, so was that all skill, or luck + skill? If we assume that there’s something about relief pitching that allows a pitcher to outperform his FIP/xFIP by a certain amount, we can still control for the luck factor by ranking them by those measurements, since they would all get the same adjustments. By xFIP, he’s 20th. FIP? 34th. He’s the #11 K/9 guy over that stretch, but a brief glance at the top of that leaderboard reminds us that strikeouts aren’t *everything* to a relief pitcher. He’s about 40th (of 93 qualifiers) in BB/9 and his HR/9 lands him solidly in the bottom 3rd of relievers, something that stems from his proclivity for surrendering fly balls. Some of that gets eaten up by Comerica park, but even bringing his HR/9 down from 1.04 to .83 (a 20% reduction) only puts him right around the median.

    Finally, in regards to Rally’s site, that’s all fine and well, but since he uses a different methodology, you can’t just take his Rally WAR and compare it to FG WAR. Sean doesn’t have sortable leaderboards, so if you feel like going through and ranking the prominent relievers by WAR, knock yourself out – that’s your baby, nurture it. I’ll be here if you want to share the results.

    Comment by Kevin S. — January 18, 2010 @ 2:55 pm

  47. The 50% is not arbitrary at all.

    And there is no lack of faith implied. If there is one, then I am asserting there is no lack of faith. No need to imply anything once I outright make something explicit.

    Comment by tangotiger — January 18, 2010 @ 3:00 pm

  48. What?

    Comment by Joe R — January 18, 2010 @ 3:10 pm

  49. But Daves argument does value the setup man as a replacement level player.

    The way WAR is calculated now, it is valueing Mariano above Hughes…

    Comment by Steve — January 18, 2010 @ 3:19 pm

  50. Right, his methodology is different. He has a slightly higher replacement level and he also uses a RA WAR and that will paint those relievers who outperform their peripherals in a better light. Looking at the last 3 years…

    Joe Nathan goes from a combined 6.2 WAR to a combined 10.3 WAR at Rally’s site. Rivera goes from a combined 7.3 WAR to a combined 9.1 WAR at Rally’s site. Joakim Soria goes from a combined 5.9 WAR to a combined 9.0 WAR at Rally’s site. Valverde actually has a higher WAR the last 3 years at Rally’s site than Jon Broxton does (5.6 to 4.6), and he’s 10th in reliever WPA the last 3 years. Carlos Marmol is another reliever who looks a lot better at Rally’s site then when using his FIP.

    I’m not sure if one way is better than the other, but I don’t think one should completely ignore the two different WAR metrics.

    Comment by Scottwood — January 18, 2010 @ 3:24 pm

  51. Fair point, but limiting the conversation just to relievers, if I’m looking to sign a RP in free agency, I have control over the role in which they will be used — assuming that there is NOT a skill to perform better/worse in high leverage situations, I don’t care about how they were used by their prior teams. Sure, in retrospect the performance was worth more in terms of wins & losses than it would have been in lesser leverage, but that value was created by the team who chose to utilize the player in that situation, not by the skill of the player such that it should be priced higher.

    That is to say, the market should not differently value a reliever based on the leverage of the situations in which he was placed unless they believe that the leverage has an impact on the quality of the performance. That closers are paid so much suggests that some organizations do believe in closer clutch and add some multiplier. I don’t think that WAR should be doing that though.

    Comment by Rick — January 18, 2010 @ 3:25 pm

  52. I am not good at expressing my ideas in words… Let me try again…

    Dave arguees that since a closers replacement is a set up man that he shouldnt recieve the same amount or value he would if his replacement was a AAA scrub. But since a setup man HAS value above a scrub, shouldnt WAR acount for that???

    If you have a setup man worth .7 WAR, you can value a closer by the Value he adds above the setup man with out including what the setup man is worth. If you have a 1WAR closer according to the way Dave calculates it, it means he is 1win above the setup man… But shouldnt we also account for the .7WAR the setup guy is worth? Making the closer 1.7WAR!

    Comment by Steve — January 18, 2010 @ 3:25 pm

  53. but the setup guys are also worth something above a replacement level player.. how can they not be??? so if your set up man is above replacement level you cant just value a closer above him, you must look at the setup mans value too..

    Comment by Steve — January 18, 2010 @ 3:27 pm

  54. The people who claim that WAR works just fine for relievers (and closers) seem to just make the following arguments: (1) WAR works just fine for relievers and (2) teams who overvalue closers and set-up relievers are dumb because they pay too high of a rate for reliever WAR. But these are just assumptions. Where is the evidence for these claims? I bet it’s more than a half-dozen teams who pay a premium for late-inning relievers.

    Given that a good portion of teams pay late-inning relievers out of proportion with WAR values, I think it’s the overly reductionist WAR-mongerers’ assumptions that are suspect, not half of the GMs in baseball.

    Comment by WY — January 18, 2010 @ 3:32 pm

  55. PERFECTLY PUT…. I HAVE TRIED TO EXPRESS THIS IN LIKE 4 COMMENTS, BUT WAS UNABLE TO DO IT THIS WELL…

    Comment by Steve — January 18, 2010 @ 3:34 pm

  56. Wait, that makes no sense?

    Lidge was replaced by Lidge. In 2008 he was lights out, in 2009 he sucked. They made no effort to replace him in 2009.

    Comment by snapper — January 18, 2010 @ 3:58 pm

  57. It’s what you call “framing the situation” .

    So, if Chone Figgins gets hurt and gets replaced by a hot prospect, who performs at league average … the Chone’s WAR is calulated by ‘league average’ player, not a replacement level? Of course not.

    Why would closers be the only position to get a “penalty” because their direct replacement happens to be a decent reliever?

    The ROSTER spot is being replaced by a replacement level guy, and isn’t that where the calculation takes place for everyone that is NOT a reliver.

    I understand the point being made, but IMO it creates a unique situation to fulfill a predetermined set of ideas. I’m not saying that this is being done intentionally, but that is certainly appears to be occurring.

    A whole “chain” effect could happen when a position player gets injured. We see it quite often. An IF will move to his 2nd posiiton, an OF will move to another OF spot, someone will move to 1B, allowing a call-up or platoon of a player that overall, makes the team better … without it being a straight “Call-up 3B for an injured 3B” type of situation.

    I don;t know why we pretend chaining only exist for closers/relivers. Isn;t this part of the value of guys like Mark DeRosa? Roster flexibility.

    I completely understand the high quality, less quantity, but I disagree with the chaining penaly in WAR. For example, if Chase Utley gets injured, he’s not replaced by a AAA 2B, he’s replaced by Placido Polanco, who is then replaced with another utility guy or MiLB call-up. Teams tend to put their best talent wherever it fits for the overall best team possible. It’s not always a one-for-one swap/change. Again, this is where some guys like Bill Hall have value.

    Comment by circlechange11 — January 18, 2010 @ 4:17 pm

  58. Steve, that is exactly what chaining is doing. Take a typical bullpen. Compare the new closer to the old closer, compare old closer to the old setup guy, compare the old setup guy to… Add up all those little changes all the way down the line. That’s the value of the closer. That’s “chaining”.

    Comment by Sky Kalkman — January 18, 2010 @ 4:19 pm

  59. How is that any different from what teams can and do with position players. It’s not always a straight call-up.

    See my example above. Chase Utley gets injured. If they have a better 3B than 2B in MiLB, then Polanco moves to 2B and the 3B gets called up or the utility man from the bench goes to 3B.

    So, Utley is replaced by Polanco, and Placido is replaced by the replacement level guy.

    So, do you calculate Utley’s WAR by using Polanco as the “R”. Hell no, and if someone did this whole site would hemorrhage.

    Comment by circlechange11 — January 18, 2010 @ 4:30 pm

  60. I don’t know. I think that the only relievers who ought to be in the Hall of Fame are Mariano Rivera and possibly Wilhelm and Gossage, but WAR doesn’t quite pass the smell test for me. The top 35 pitchers in the majors were, according to WAR, starters. These included pitchers who threw 170 above-average innings like John Lackey and Ted Lilly. When one is comparing starters to relievers, one rightly makes an adjustment for the improvement that starters make when converted to the role. However, it is, most likely, a continuum. It’s probably easier to throw 70 one-inning stints than 34 seven-innings stints, but one guesses that throwing 30 six-inning stints falls somewhere in the middle. I doubt very much that Lackey or Lilly last year added more value than the best reliever in baseball.

    My nose tells me that a top-notch Mariano Rivera season would be in the 4-5 WAR range, rather than just over 3. Fangraph’s WAR justifiably uses FIP for one-season, rather than ERA+ or RA+. However, for a pitcher like Rivera, FIP undersells him because of his low LD/high pop-up rate, and hence the career opposition BABIP of .276.

    Comment by Mike Green — January 18, 2010 @ 4:34 pm

  61. Rick, I agree that past LI shouldn’t matter going forward for value. In a team context, a GM should project where the reliever will be used and how much added value that provides. In a theoretical discussion of future value, a reliever should be given a “deserved LI” — the LI of the role he deserves, on a continuous scale.

    Comment by Sky Kalkman — January 18, 2010 @ 4:36 pm

  62. Fangraphs WAR probably does underrate top closer a bit because of the use of FIP. Ideally, we’d have something like PZR based on hit f/x data. But, well, we don’t.

    Being as nice as possible to closers, we can look at WPA, which will give full credit of results to the pitcher and none to the fielders, and will give full LI credit to the reliever.

    Luckily, replacement level for relievers is basically league average, so WPA is the same as WPAAR (win probability added above replacement).

    Here’s the 2009 leader board: http://www.fangraphs.com/leaders.aspx?pos=all&stats=rel&lg=all&qual=n&type=3&season=2009&month=0

    One huge season from Papelbon, and no other reliever over 4 WPA(AR).

    Once you account for all the things that are inflating many of those numbers (fielding, park, luck, not chaining, etc.), they’ll be even lower.

    Comment by Sky Kalkman — January 18, 2010 @ 4:50 pm

  63. It’s obvious to anyone that actually observes how major league teams deal with injuries. It’s why guys like DeRosa, Hall, and otheres that can start (if need be) at various positions. You don’t have to replace with a replacement level player, you can “chain” so that the replacement level player that fills the roster spot, never sees the field up close. I honestly don;t comprehend how it’s not obvious, unless one is completely submerged in spreadhseets and forgets to take a simple gander at the greatest game.

    Sometimes it feels like FOX News around here: [1] Form an opinion, [2] dig for support and justification.

    We already heard numerous comments in another thread suggested that relievers are just failed starters, which is just absurd when you look at the quality of established closers. It’s more akin to “greyhound v pitbull” than it is to “successful v failed starter”. Closers are failed relievers, the same way 3B and 2B are failed shortstops. How bout we start referring to Chase Utley and Chone Figgins as “failed shortstops”. Geez.

    Think of closers as heavyweight knockout specialists. That doesn’t make them “failed middleweights”, just a different type of pitcher.

    This is the stuff that is obvoius to anyone that has actually played the game at a level where the difference between a good closer and a poor one is night and day. It matters, and much more than WAR says.

    On another note, someone mentioned it earlier, but isn’t it well established that something like 70% of all innings are shutout innings, and we extrapolate this to the idea that any reliever could essetnially rack up 30 saves in a season, and be rather accurate. But, THAT’s the point, an established closer may get 42 or so saves in the same number of attempts. Unlike SP’s when a closer blows a lead, there isn’t much time to recover, so a BS is often an L. Those aren’t theoretical losses, or WAR calculated stats, that are real, devestating, immediate, demoralizing losses.

    Just look at how teams struggle when their automatic closer has an up and down year. Does that look like “marginally important” to you?

    Do teams continually “overpay” closers because they’re just glorified setup men, or perhaps there really is something to the established closer that provides additional value over a common reliever.

    To me, it’s as obvious as stating that certain guys would drive in more runs than other guys if they batted in the same spot of the lineup on the same team. Both guys may drive in 100 runs, but one guy may drive in 145 compared to 112. That’s additional value. Not every reliever succeeds the same in the closer role. LIke, duh.

    Also closers require a different “skill set” (high K/9, usually one outstanding out pitch, often seroius velocity, low BB/9, low contact % … basically the stuff you could not maintain for more than one time through the lineup, unless one is just mega-talented … even among MLBers).

    Comment by circlechange11 — January 18, 2010 @ 5:12 pm

  64. ERA for relievers is often a bad idea. For 3 basic reasons.

    [1] Some of the most important runs they ‘allow’ are inherited runners, which do not count toward their own ERA. (I would favor some type of comination of inhereited runner % plus runners allowed, as some type of stat for valuing relievers).

    [2] Some of the runs that are attributed with were given up by hits allowed by other pitchers.

    [3] Smaller sample size (IP) skews ERA. Like I said, you can throw 6 no-hit innings (all in 1-run games), claiming 6 holds as a RP … and then come in and give up a 3-run HR (1 IP) in a 9-1 game, and your ERA is now 3.86 … which for a reliever ain’t that great. Nevermind that ERA doesn;t give us an idea of well/poor the RP has pitched. For SP’s, I think it’s a decent indicator. Sure, there are btter ones, but generally speaking, it works for the vast majority of pitchers.

    Comment by circlechange11 — January 18, 2010 @ 5:21 pm

  65. I don;t think there’s ANY doubt about it.

    There’s two ways to know …

    [1] Either pitch in the 7th and/or 9th and compare how you felt during both situations. Most guys are not fortunate enough to have this experience.

    [2] Listen to what major league pitchers have to say about it. It’s night and day, as they all say.

    IMO, there is just a strong desire to disregard anything a professional baseball player, manager, or GM says or feels about a situation they experience first-hand in favor of certain metrics, whether they are based on accurate assumptions or not.

    Heck, just read player comments about the adjustment from going from setup man to closer. BIG step.

    Comment by circlechange11 — January 18, 2010 @ 5:26 pm

  66. I always the work Tango did in trying to convert Gagne’s WPA to WAR in this blog post.

    http://www.insidethebook.com/ee/index.php/site/comments/educating_a_fan_on_reliever_impact/

    Comment by Scottwood — January 18, 2010 @ 6:19 pm

  67. That should say, “I always liked….”

    I hate typos.

    Comment by Scottwood — January 18, 2010 @ 6:20 pm

  68. It just seems weird to ever see Mariano Rivera with the same WAR as a pitcher like John Lackey. Mariano Rivera could never ever do Lackey’s job while Lackey could probably do Rivera’s job at a similar level. We see all types of starters from Eric Gagne to John Smoltz change roles and dominate in relief. Very rarely do you ever see a long time closer switching back to a starter role.

    I’m not sure the effect that should have on WAR I’m just saying that the closer role is a lot easier than people here seem to think.

    Comment by recca — January 18, 2010 @ 7:03 pm

  69. “Jason says:
    January 18, 2010 at 12:21 pm

    Doesn’t it work the other way as well? When you sign a free agent first baseman, you get better at first base. When you sign a free agent closer you get better at closer, set-up guy, middle relief and mop-up.”
    —–
    I think this may sort of be the key to the whole issue. I agree with WAR’s theme of relief pitchers being far less valuable than starters, but the numbers just seem hard to believe. It doesn’t mean they’re wrong, of course, but I’ve always thought there has to be something more to it.

    Maybe the problem is that, based on the average leverage index of each role, we should have different replacement levels. After all, a replacement level mop-up guy would hardly be suitable for the closer’s role, and would probably be considered below replacement level for what we’d expect of a closer. Of course, different relievers’ roles are ambiguous, but they could probably be broken up in much the same way that spot starters have separate SP and RP runs now. If that were the case, a closer’s WAR, for example, could be calculated by his WAR (as it is presently), plus the difference between a replacement level closer and set-up man, plus the difference between a replacement level set-up man and middle reliever, etc…

    Of course, then teams would have to evaluate the marginal impact within their own bullpens, but that’s always true of WAR. I’m hardly an expert on developing any new metrics, but does anyone think this may work?

    Comment by Whartonite — January 18, 2010 @ 7:30 pm

  70. While relievers are almost universally overvalued in fantasy drafts, real life is not quite like fantasy. I have trouble telling if closers are overvalued or adequately valued in real life. I feel like they have these pros and cons, for an ideal closer:

    Pros:
    - High reliability to not blow a save
    - High quality innings on average (low ERA, etc)
    - High ability to beat the best batters
    - Does not pitch worse in a higher leverage/pressure situation (i.e. playoff game)
    - Lower injury risk (?) due to lower use

    Cons:
    - Low number of innings than a go-to reliever (i.e. 7th inning guy)
    - Difficult to measure most of the above things you want in a closer, due to sample size splitting.
    - Can’t easily compare non-closing relievers to ones with closing experience, as they lack the situational samples you are interested in

    In my opinion, a closer is a specialist which requires not just mean performance to be good but really requires a certain adjusted median or mode performance. When it comes down to it, you mainly want him to be able to finish the game. Additionally, it becomes more important in playoffs when you’re likely to be facing the top hitters.

    Basically, I just don’t see that mean performance is a good metric for closers and one that makes it hard to tell what their true value is. It’s obvious that a closer who can prevent any replacement level hitter from ever getting a hit (ex. 0.050 BAA) but gives up a 0.5 BAA to cleanup hitters is probably a worse closer. Likewise, I would rather have a guy that gives up under 0 runs 80% of the time and 5 runs the remainder than a guy who gives up 0 runs 66% of the time and 3 runs the remainder. The ERA is the same, but the extra 2 runs lost have no impact if you’re always in high leverage situations (3 runs or less).

    So for my money, if you really want to evaluate the role of “closer” and see what the real WAR-type value is- aggregate stats are losing important information about the distribution. I think that is also why teams are willing to pay a premium for closers. High leverage and playoff situations are a very small sample size, and one in which some players do seem to perform worse than their historical averages in. It’s a different and potentially different set of samples, and there’s no sure way to tell if your star reliever will be a dud closer (except the risky process of using them in those situations). For this reason, closers seem to be compared in value primarily only against other players with closing experience.

    So in my opinion, part of the cost of a closer is paying for their sample set of closing experience to compare them against other closers. The remainder may be due to factors which are not well represented by mean-value analysis. This isn’t to say closers aren’t overvalued, but it is to say that I don’t think mean value analysis can prove that at all.

    Comment by B N — January 18, 2010 @ 7:34 pm

  71. Great article. I think many that look at closer WAR as too small let traditional, non calculable notions affect their reasoning. The way I see it, if a closer has a 1-2 run lead almost every appearance, even with an 6.0 FIP they should be able to do a pretty good job on most nights. It doesn’t take a Rivera to close games.

    Comment by neuter_your_dogma — January 18, 2010 @ 7:49 pm

  72. Man that is sooooo true. Fox News is the only one who does that. All the rest of us are very reputable and balanced news organizations.

    Comment by Keith Olbermann — January 18, 2010 @ 8:06 pm

  73. I’m not sure they’re as scarce as you might think. Hence we can often convert a mediocre-to-failed starter into quite the effective reliever.

    (Of course, if you get a handful of GM’s who think they’re terribly scarce, and terribly important, you get BJ Ryan-esque contracts.)

    Comment by Jason B — January 18, 2010 @ 8:08 pm

  74. Not true. They used Madsen off-and-on, and he was pretty ineffectual in the closer role.

    Which, of course, does NOT definiteively prove or imply that he can’t be an effective closer. (Something our manager, announcer, and TV bloviator friends need to be reminded of, occasionally – “well he blew that one save opportunity we gave him, so let’s never speak of THAT again and pretend this whole sordid affair never happened!!”)

    SSS, SSS…

    Comment by Jason B — January 18, 2010 @ 8:12 pm

  75. It seems like the frankenmetric master plan should be to have the relief pitchers throw the 1st 3innings, then have the starters come in and throw the last 6 innings.

    If relievers aren’t important then why should they pitch in the highest leverage situations?

    Comment by Linuxit — January 18, 2010 @ 9:20 pm

  76. I don’t watch news, I just went with a name I hear people complain about.

    Political talking heads and know it alls (without actually serving in a position of public service or politics) are more annoying than any biased sabermatician. *grin* Everyone seems to have all the answers, they just don’t have to make any of the decisions.

    Present company excluded, of course. *cough* *cough*

    Comment by CircleChange11 — January 18, 2010 @ 9:31 pm

  77. To circlechange11: Your statements are true, but not relevant to my point. I was simply using the ERA scale to discuss the talent levels of hypothetical pitchers.

    Comment by Sky Kalkman — January 18, 2010 @ 10:02 pm

  78. Don’t worry about the nesting, Fish, your post was almost perfect. That along with high leverage innings are why Top. Closers. Are. Worth. The. World. But since none of that can be explained with statistics, Stat sites (like FG) refuse to acknowledge their value. All they see is 60 IP.

    And Dave, ‘some clubs still value late inning relievers enough to pay a premium’? Really, just ‘some clubs’? As far as I can tell there are but two GM’s that devalue closers: Beane and Sabean, and needless to say, neither win all that often. Those team’s that expect playoff baseball every year (think, BOS, NYY, PHI, LAA, ATL, LAD, STL) will pay whatever necessary to have one. And in the years they don’t have one, they lose.

    I love FG but this opinion of theirs is just wrong.

    Comment by Dirty Water — January 18, 2010 @ 10:06 pm

  79. Nice post, CC. Very true.

    Comment by Dirty Water — January 18, 2010 @ 10:25 pm

  80. Madson tried several times and he imploded each time. It wasn’t pretty.

    Comment by NEPP — January 18, 2010 @ 10:32 pm

  81. I understand what you were saying, and don;t disagree.

    My major point was that as a reliever if I “stink it up” in a game where I enter with 1st and 2nd with 1 out, and give up 2 hits, allowing both inherited runners to score, walk a guy to load the bases, and then get pulled and the next reliever gets a double play … my ERA for that game is 0.00

    As a starter you don’t get that luxury. All of the earned runs that score when your pitching are yours AND all of the runners you leave on that score on hits given up by the relief guy.

    My commentary was to state that a starter could go to the bullpen, pitch worse, and yet (theoretically) have a lower ERA.

    ERA is essentially, IMO, useless for relievers … and I’m not in the anti-ERA crowd.

    I still come back to my point that the best relievers are guys who have one (maybe 2) outstanding pitches and have the mentality and physical attributes to “throw their balls off for one inning”. The mindset and physical attributes for starters is almost the exact opposite. Think: marathon v. 100m sprint.(1.2 marathon v 800m would probably be a better analogy)

    That’s not directly to your point either, but hey I ‘m on a roll

    Comment by CircleChange11 — January 18, 2010 @ 10:39 pm

  82. Chaining? Really?

    (A-B) + (B-C) + (C-D) + (D-E) + (E-F) is still A-F, no matter how many parentheses you throw in there.

    Comment by ToddM — January 18, 2010 @ 10:50 pm

  83. Because they may be as good as the starters….they will just never pitch enough innings to be worth it in value.

    Comment by timmy013 — January 18, 2010 @ 10:53 pm

  84. Not when you account for leverage…. Its really (1.5A-1.5B) +(B-C)+(.75C-.75D)….

    Numbers are just guesses to represent leverage

    Comment by timmy013 — January 18, 2010 @ 10:55 pm

  85. That’s funny, because all signs coming out of Boston are they have no interest in paying Papelbon once he gets expensive and are going to hand the role over to Bard.

    Comment by Kevin S. — January 18, 2010 @ 10:56 pm

  86. Except… that’s not true. He had six blown saves last year, but three of them were in the 8th inning, i.e. not in the mythical “closer’s role.” In Save Situations, he allowed a .261/.312/.373 line, with an 8.71 K/9, a 2.18 BB/9 and a 0.87 HR/9. That doesn’t sound like imploding each time to me.

    Comment by Kevin S. — January 18, 2010 @ 11:08 pm

  87. Which I am sure is true (although it might not be if Papelbon did not have shoulder issues). But two facts exist here; 1, they are paying Papelbon a lot presently and, two, if Bard (or whomever) does not work, Epstein will pay for another closer who does, like he did before with Foulke.

    Comment by Dirty Water — January 18, 2010 @ 11:20 pm

  88. Didn’t Dave write in this very article that leverage index is accounted for in WAR?

    Comment by Andy — January 19, 2010 @ 12:27 am

  89. Hi,

    To broaden the discussion slightly, I have a big problem with Fangraphs’s WAR for pitchers in general, not only relievers. This is due to its exclusive reliance on FIP. It seems to me that FG is trying to accomplish 2 things with the WAR:

    1) Estimate what the pitchers value actually was – retrospectively determining value
    2) Provide a baseline for a projection as to his expected value in the future.

    FIP is an excellent methodology for future projection, but I think it misses out big time on as to assessing the value actually provided. My problems with FIP are:

    a) FIP does not account for sequencing. bb, bb, HR, k, k, k (3 runs scored) is valued exactly as HR, bb, bb, k, k, k (1 run scored). ERA or RA of course accounts for this.
    b) FIP views identically all balls in play, whether it is a dribbler to the mound or a line drive double smashing into the monster (actually with Posada those are always only singles…).
    c) FIP condenses the difference between the best and worst pitchers

    While I agree that the ability of the pitcher to repeat such “skills” as sequencing and quality of balls in play are doubtful, there is no doubt in my mind that a pitcher who has done a good job in this regard, has provided value. Whether it was due to luck or not is irrelevant, one way or the other the pitcher has provided value. As Napoleon allegedly said “Give me lucky generals”. I do not think it is just luck that Rivera’s career ERA is 2.25 while his FIP is over 1/2 a run higher at 2.78.
    Now, I do agree that the pitcher should not be credited with the value his fielders provided, and this is the rationale behind FIP. But to ignore the above points is even a bigger mistake IMHO and I’d prefer some combination of RA/FIP/WPA.
    I also agree with the previous posters as to the unfair penalizing of relievers as it relates to chaining with the free pass that everybody else gets. Also intuitively, I can’t accept that Vazquez was worth more than 3 times that of Rivera last year (WAR of 6.6 vs. 2)

    Comment by Jud — January 19, 2010 @ 4:10 am

  90. Vazquez was significantly more awesome than usual last year. I’m not sure what’s so difficult about that for people to get. Even by a traditional metric, ERA, he was superb. Rivera simply couldn’t make up the massive innings gap.

    Comment by Kevin S. — January 19, 2010 @ 4:18 am

  91. Sky, I think most of us understand that; the main argument is that “we’re quite probably not doing it right”, i.e. the way it’s worked out is rather arbitrary (as far as I can tell having read some stuff on it), in much the same way that the application of Tango’s LI to bullpen pitchers seems to be done somewhat inelegantly and (IMO) taking the stat somewhat out of context. When we then apply those two slightly hazy applications of non-win/run-based concepts and attempt to derive a definitive WAR figure (especially when the WARs we’re dealing with are rather small), those of us who don’t buy 100% into the reliever WAR metric are suggesting that there’s potentially a LOT of error and unincorporated variables into this way of doing it.

    I think that’s a fair argument and I’m yet to see an argument in favour of the validity of the bullpen WAR formula that patches all those holes.

    Comment by Felonius_Monk — January 19, 2010 @ 5:35 am

  92. I think you’re over-simplifying Dave’s (and fangraphs’) argument somewhat, Steve. It is not valuing the set-up guy as a replacement level player.

    The true “replacement level” for a bullpen pitcher is set (somewhat arbitrarily) higher than that for a starter, and the extra wins the leverage-index stuff adds are ON TOP of the “regular” WAR (as a multiplier of that WAR).

    Comment by Felonius_Monk — January 19, 2010 @ 5:42 am

  93. “Actually, they typically CAN be converted and be good relievers. Starters, on average, see their ERAs go down by one run when converting to the bullpen. So a 4.50 ERA #3 starter would be a 3.50 ERA reliever. Not an ace, but still pretty damn good. (And a true bullpen ace is more valuable than a #3 anyway.)”

    But Sky, is this not inherently biased (and I think the -1 run thing is something of a rule of thumb, rather than an idea that has a consistent, contemporary proof – please correct me if I’m wrong)?

    Presumably, the starters in the history of baseball who have converted to relief are guys who are deemed to be either not good enough to start or who have other issues with starting – they will, in general, be poor starters, for one reason or another (not enough secondary pitches, poor stamina, etc etc), so immediately this sample is biased in the direction of “poor starters”.

    Secondarily, the starters who have moved to the bullpen and put up better ERAs are the guys who will have STUCK in the bullpen. If a guy can be a competent #5 starter, then moves to the bullpen and puts up a similarly mediocre ERA (or FIP), he’s a long-man. These sort of pitchers will either fall out of the bullpen and into AAAA duty or the minor leagues, or be moved back into the starting rotation, a large amount of the time, certainly more so than poor starters who move into the bullpen and dominate. So the study (if it’s an empirical one) is ALSO biased in the direction of “starters who successfully move to the bullpen”.

    Given that any empirical study of starters moving to the pen will be biased both in favour of bad starting pitchers, and comparatively good bullpen pitchers, is it any wonder that the bulk of the data in that study will suggest that a given pitcher will put up better stats in the bullpen than the rotation?

    Also, as circlechange mentions above, ERA differentials between SP/RP are unfair because a percentage of the ERA that starters give up will be attributed to relievers allowing their inherited runners to score, whereas the same cannot be true in the opposite direction.

    I am yet to see anything convincing to suggest that a typical starting pitcher is one whole run per 9 innings worse off in the rotation than in the bullpen, which makes the “replacement level for relievers” thing a questionnable concept at best, IMO. I just don’t think we’ve answered so many questions about the true value of relievers, and I think that makes giving them a hard-and-fast numerical figure for their value inherently difficult.

    Comment by Felonius_Monk — January 19, 2010 @ 6:13 am

  94. How often has this happened in recent history, though?

    I’m struggling to think of more than half a dozen failed starting pitchers who were not tarred with the “likely future RP” tag in the minors, who were not good enough to be #5 men in their rotations and who ultimately became consistently good relievers. Certainly not too many who were ~1WAR starters (i.e. a 5th starter, more or less) and then became >1WAR relievers (i.e. a typical closer/bullpen ace type). Of course, there’s the Joe Nathans of this world, but they seem to be more the exception than the rule as far as I can see. It’s far more often your typical crappy #5 starter gets bumped out of the pen and immediately moves into middle relief or longman/swingman work, is equally as sucky as he was in the rotation, and ultimately gets displaced out of the team or into AAA when something better comes along.

    Comment by Felonius_Monk — January 19, 2010 @ 6:18 am

  95. “The 50% is not arbitrary at all.”

    A link would be useful. I’ve been reading a bit on chaining and LI as concepts applied to value and I don’t think I’ve come across the proof for this yet. Thanks!

    Comment by Felonius_Monk — January 19, 2010 @ 6:21 am

  96. Off the top of my head I’ve got Eck, Mo and Nathan. Another reason a starter can fail is health, which explains the moves by Wood, Pap and Smoltz (temporarily). Chan Ho Park’s not a Closer with a capital C, but he’s another who washed out of the rotation and has found a niche as a very good short reliever. If I wasn’t completely exhausted, I could probably give you a better list.

    Comment by Kevin S. — January 19, 2010 @ 7:04 am

  97. i’m sensing some sarcasm

    Comment by Steve — January 19, 2010 @ 8:25 am

  98. This is exactly what Cameron & Co. are saying.

    How many great relievers had great careers as starters to go off of? Eckersley?

    Gossage had a year of starting, and rocked a 91 ERA+.

    Rick Aguilera started 89 games in his career. 4.07 ERA, 1.307 WHIP. Good for ~ a 103 ERA+. That might be high-balling the split, too.

    And SSS, but Rivera in 10 starts rocked a 5.94 (!) ERA. I wouldn’t read TOO much into that, but the principle still holds. A struggling starter can slot into the bullpen (ex: Wakefield, 1999, holy shit he’s been on the Red Sox forever), but not vice versa.

    Comment by Joe R — January 19, 2010 @ 8:50 am

  99. I LOVE it whenever a guy wastes an early pick on a reliever other than Rivera for saves.

    Please, build your team around volatile stats with no predictive value like BA/RBI/Wins/Saves, I’ll go get my HR/OPS monsters and worry about that later.

    Comment by Joe R — January 19, 2010 @ 8:55 am

  100. I’d never want Wakefield as a reliever.

    Comment by neuter_your_dogma — January 19, 2010 @ 9:06 am

  101. They’re paying Papelbon $6.25 million. He’s constantly at 2-3 WAR, Dirty Water.

    The point isn’t that you should just throw around league minimum contracts to relievers and hope 1-2 a year pan out, the point is that it’s silly to pay extra because one has the word “closer” attached to his name. Papelbon for 1 year / $6.25 mil is fine.

    The mind-boggling save fetish going on in Los Angeles is not okay. Not to mention what the hell is wrong with just using Kevin Jepsen (2.86 K/BB) as a closer if you’ve no faith in Fuentes, rather than bring in another mediocre pitcher with a good fastball? Is it because Jepsen had the audacity to be unlucky (.299 BABIP against for the league vs. a .363 for Jepsen, the nerve)?

    Does anyone think Jepsen just has some total inability to control where balls land in play? Look at this ridiculousness:

    Jepsen

    vs.

    All of MLB

    If Jepsen’s even close to a 5.00 ERA in 2010, it’s either because he fell apart as a pitcher, or the most improbably unlucky crap in a long time.

    Comment by Joe R — January 19, 2010 @ 9:08 am

  102. God dammit, why do people argue using the counter-intuitive?

    That is not a point.

    Sequencing? That’s the same argument Joe Morgan uses to argue a .300 hitter sucks balls (“if he makes his hits only with no one on base, he’s not a valuable hitter”). Over the long run, that balances out.

    Some pitchers just suck in certain aspects of the game (like holding base runners) and underperform their FIP a bit. Some are better. So sure, bump Rivera up a little in terms of performance. Vasquez was still much more valuable than he was in 2010.

    Comment by Joe R — January 19, 2010 @ 9:12 am

  103. SSS ambush coming, but:

    1999, out of the pen:

    3.50 ERA, 1.38 WHIP, 1.73 K/BB

    And he was pitching bad in 99-00 as a starter, giving up long balls and having trouble controlling his stuff. I thought he was going to wash out of the league.

    Here we are a decade later and he just got a 2 year contract. I love the fact that he’s going to last until he’s 45.

    Comment by Joe R — January 19, 2010 @ 9:18 am

  104. Joe R – A pitchers job is to limit the number of runs he allows. In terms of value provided, the “balancing out” is irrelevant. The question is whether he limited runs or not. If he suceeded in doing so by not giving up the big hit, he should be credited with it. Please note that I’m not claiming that this has any predictive value.
    I also did not claim that Rivera was more valuable than Vazquez. What I said was that a 3x difference does not seem reaosnable to me.

    Comment by Jud — January 19, 2010 @ 9:25 am

  105. Post 6:

    http://www.insidethebook.com/ee/index.php/site/comments/valuing_relievers/#6

    Comment by tangotiger — January 19, 2010 @ 10:35 am

  106. In case it got lost in the threading, read this please (post 6):

    http://www.insidethebook.com/ee/index.php/site/comments/valuing_relievers/#6

    Comment by tangotiger — January 19, 2010 @ 10:38 am

  107. Yes, you don’t need to tell me what a pitcher’s job is.

    And in the case of Rivera, he constantly outperforms his FIP mainly due to having a devastating non-FB pitch (his cutter). That’s not commonplace for a reliever. Just because there’s an anomalous result, doesn’t mean a metric straight up doesn’t work.

    And I love the implication that WAR undervalues relievers. It still says that the best ones deserve big contracts (in the $8 – $12 mil a year range). Would you really want your team locking up 1/8th of their payroll for a guy scheduled to pitch 5% of your team innings? The best relievers get their money, and marginal wins say they deserve to get their money. The argument is against giving guys like Fernando Rodney big money for being “bona fide” closers.

    Comment by JoeR43 — January 19, 2010 @ 10:45 am

  108. So a 3.00 ERA (FIP) max-leverage closer ~ 2.4 marginal wins?

    Or $8-$10 million a year in FA?

    Makes sense to me.

    Comment by JoeR43 — January 19, 2010 @ 10:47 am

  109. Just to go into point number 2 – I think the logic behind that is that a given pitcher will perform better as a reliever than a starter, for a number of reasons (doesn’t go through the lineup multiple times, doesn’t get tired from the workload, can throw harder, etc), right? Over at Inside the Book, Tango and mgl have studied this and found it to be true – performance does increase moving from starter to reliever, by something like .5 FIP or so, on average. So given that, it does seem to make sense to me that they’d set replacement level higher for a reliever.

    Comment by B — January 19, 2010 @ 11:18 am

  110. Madson in 2009
    8th inning, he was a .257/.318/.339 against guy, .333 BABIP against
    9th inning, he was a .267/.313/.467 against guy, .328 BABIP against

    That’s 1 HR every 19.2 PA, or 5.21% HR/PA rate in the 9th. Coming into 2009, his HR/PA rate in the 9th was 1.86% in 161 PA. His career HR/PA is exactly 2.5%. So, want to chalk this one up on Madson not having that “umph” factor, or bad luck? Does he magically melt down in the 9th inning because he’s not a clutchy mcclutcho (ignoring that his career “clutch” rating is actually positive)?

    How about we talk about Heath Bell? Earned $1.255MM in 2009. 2 SV coming into 2009. 42 in 2009. What made his 2009 (3.02 xFIP, FIP overrates SD pitches imo since they play in a dungeon) better than his 2007 (24 more IP, 2.82 xFIP). Coming into 2009, he had 151 PA against in the 9th inning for his career. He was slotted into the 9th inning in 2009 and seemed to have no problem becoming a “bona fide” closer.

    Is Jenks > Matt Thornton because Jenks is “bona fide”? Thornton’s been a much better pitcher two years running, yet Jenks gets the $ and the save situations because of some mystical clutch factor. Jon Broxton seemed to do just fine when he went from setup guy to closer.

    So sure, maybe some guys can’t close. But we’re analyzing Madson’s “closing ability” based on a small, anomalous sample that was completely different from his past performance? That is biased at best, and insane at worst.

    Comment by JoeR43 — January 19, 2010 @ 11:42 am

  111. That all being said, I fully expect Charlie Manuel to keep Madson (and Chan Ho Park for that matter) as setup guys because of their lack of “9th inning experience”.

    Comment by JoeR43 — January 19, 2010 @ 11:47 am

  112. “Either pitch in the 7th and/or 9th and compare how you felt during both situations.”

    How you “feel” is only relevant in what effect it actually has on changing the outcome. To that degree, it’s quite measurable – and that’s the biggest reason it’s discounted among people who actually look at the effects, because they really don’t find much (if any) difference. If it “feels” different but produces the same results, then frankly, who cares about the mental aspect? All we’re interested in is the performance. I’m not going to speak for the majority and say the FG crowd (or whatever other population subset you want to talk about) is or is not too quick to disregard what those guys say…I just want to point out that these are the types of things we should be able to objectively measure/quantify, so for your point to be relevant, you’d have to have some real, hard evidence. Maybe I’m wrong – I’m not the world’s expert on sabermetrics – so if there is some evidence that supports your point (or even if you have good reasons to disregard the evidence against your point), by all means, share it….I just don’t think pointing to the fact that something “feels” different, when it’s pretty well known how awful we (people in general) are at objectively observing anything, really, means a whole lot.

    Comment by B — January 19, 2010 @ 12:47 pm

  113. “Over the long run, that balances out.”

    Of course this is right. However, what’s the purpose of the stat we’re looking at? In terms of looking backwards at what a player did to help win games, something like sequencing is important, because it has a real value on winning games or not. Get a big enough sample size and it’s not important because it will eventually regress to the expected level, of course…but if we’re looking to see who contributed to wins over a sample of a season….I can see a good argument to take it into account – because, again, it did contribute to winning or losing games in the past. Now, if we’re talking about quantifying how much a player should be paid, it shouldn’t be taken into account, because it’s not a repeatable skill going forward. Again, in some circumstances, it makes sense to credit/value it, though.

    Comment by B — January 19, 2010 @ 12:58 pm

  114. Aren’t you just arguing for WPA, though?

    A WPA where pitchers and fielders get proper credit for their roles, but still.

    Comment by Kevin S. — January 19, 2010 @ 1:06 pm

  115. “A WPA where pitchers and fielders get proper credit for their roles, but still.”

    Sure, I think a stat like that would fill the “looking backwards at what a player did to help his team win games” role pretty well.

    Comment by B — January 19, 2010 @ 1:42 pm

  116. I agree.

    FiP doesn’t distinguish a pitcher who gives up 2 HR’s in a game vs one that gives up 8 or 9 Walks. The formula for FIP also implies that K’s have the ability to negate HR’s and BB’s. Strike Outs are good and all, but they’ll never be able to erase something that already happened. It’s just a poor stat for determining value.

    Comment by Linuxit — January 19, 2010 @ 1:50 pm

  117. /facepalm
    No, it is not Linuxit.
    Among MLB starters who pitched the league minimum in both 2008 and 2009 to qualify for the ERA title, their 2008 FIP had double the predictive value to their 2009 ERA as their 2008 ERA had.

    FIP is the result of statistical regression, of how much each individual element helps prevent, or hinders, run prevention.

    Please, find me a pitcher in history that only racks up strikeouts after allowed a whole crapload of HR and BB. I dare you to. Then you might have a point.

    According to fangraphs, 1,137 pitchers have amassed at least 1,000 innings in MLB. 3 of them are outside of 1 ER difference between their ERA and FIP (about a quarter of 1 percent). 352 are within 1/10th. 661 (over half) are within 2/10ths of their career FIP in ERA.

    The formula for FIP does not imply that K’s undo previous HR and BB, you strawman-arguing clown. It implies that if you have two pitcher who’ve each thrown 100 innings in the first 81 games, pitcher A has 100 K, 25 BB, 10 HR, but an ERA of 4.00, while B has 60 K, 30 BB, 10 HR, but an ERA of 3.00, that pitcher A is probably going to pitch better in the 2nd half than pitcher B. This “well stat geeks don’t get that all that matters is what happens” crap is exactly what was said during 2005 when the Washington Nationals were 52-36 going into the ASB despite having a +/- of -4. Just because YOU don’t understand it, doesn’t make it wrong.

    Comment by Joe R — January 19, 2010 @ 2:52 pm

  118. Rockies extend street for 3 yrs, 22.5 mil.

    Comment by aj — January 19, 2010 @ 3:00 pm

  119. No, we’re not looking back on what a pitcher actually did. The case in this essay is whether relievers are valued fair or not, whether WAR “underrates” relievers.

    If a pitcher goes K, K, HR, BB, K to pick up a save, there’s no promise that’s the order he always goes.

    Obviously all that matters in the end is how they actually did. FIP provides a baseline to predict how he will do in the future. Better than ERA, at least.

    Comment by Joe R — January 19, 2010 @ 3:07 pm

  120. Also, after you give up a BB, and a HR, then you get a K and two outs for the inning, your profile goes:

    ERA: 18.00
    FIP: 17.20

    Horror. Total inaccuracy.

    That strikeout also helped improve his ERA. So does the formula for ERA imply that outs negate runs?

    Comment by Joe R — January 19, 2010 @ 3:51 pm

  121. The argument for low WAR values for relief pitchers rests squarely on chaining. Closers are not replaced with replacement players but rather setup men. Since setup me are not much worse than closers it lowers the value of closers. Yet for regular players we assume they are replaced with replacement players even though in reality that does not often happen. An injured OF is usually replaced with a 4th OF not a minor league player. Most 4th OFrs given daily playing time are 1-2 WAR players not 0 WAR players. Should we not treat relivers and regular players the same way in our modeling?

    Comment by PhD Brian — January 19, 2010 @ 4:00 pm

  122. I was wrong btw.
    Only one pitcher w/ 1000+ IP has ever deviated more than 1 run from ERA to FIP for his career

    Luckiest pitcher ever

    Comment by Joe R — January 19, 2010 @ 4:08 pm

  123. I meant to say, unluckiest pitcher ever.

    Comment by Joe R — January 19, 2010 @ 4:09 pm

  124. Here’s my issue with this:

    I think we’ve lost sight of our objective while making these adjustments for relievers. I think we basically want to create metrics for 2 things, how well did a player preform on the whole in a context neutral situation (what WAR does) and how much did a player actually contribute to a team’s chances of winning (and what WPA does). From these things we then make a simple conversion to estimate how much those performances where worth in a dollar values. Naturally, we then use these metrics and the conversion to compare players with one another and to judge trades/signings, MVPs, etc.

    With relievers what we’ve done is created a metric that sort of combines the context neutral metric with the context specific (meaning we’re using average LI of the roster spot over the rest of the pen to weight the context neutral metric, but not fully delving into the specific LI of the player) and kept the context neutral name, WAR. So we’ve lost part of the utility of WAR because we can no longer make a context neutral comparison between what players have done.

    I understand the temptation do make this conversion because one of our end goals is to use WAR to evaluate value on a dollar basis. But now this value is tied to usage patterns, of which the player has no control. So, one team could sign a reliever to pitch only the 7th inning and he’d be worth less then he would to a team that would use him in the 9th (assuming he’s a good pitcher here). So, we’ve muddied the waters. I can’t just look at WAR to get a good context independent measurement of how valuable say Mo was compared to Brad Ziegler or even maybe Dan Haren or Nick Swisher. I guess I could go back and do the conversion myself from FIP and IP, if I cared enough, but frankly its a hassle. I’d rather just keep the bar the same across WAR.

    If you want to up the replacement bar for relievers because of the .5 FIP advantage, fine, but we don’t do these chaining issues or average LI issues with other positions, so just in the name of consistency, I don’t think we should do it with relievers/closers either. I’d be much happier if we just created a new statistic like reliever WAR or something (rWAR is kind of already taken, so I don’t know maybe closer WAR and call it cWAR?) to base dollar values off of.

    Lets not forget our goals for these statistics while creating and editing them. We want context neutral stats to judge individual performance, context dependent stats to judge a player’s contribution to win probability and dollar values based on pretty much anything in between. I fully understand why we’re making this closer adjustment, I just think we’re confusing ourselves by still calling it WAR, and people that are picking up on that are dissenting because of the name. So, just keep WAR the same and make a new name, ie. cWAR. And maybe have it available for all relievers.

    Comment by Wally — January 19, 2010 @ 4:45 pm

  125. “the point is that it’s silly to pay extra because one has the word “closer” attached to his name”

    I agree with that statement, Joe. A lot of teams do pay far too much for ‘closers’ who are nothing more than highly paid ‘setup guys’. But my point is that a stud closer, like 2009 Broxton, or Mo, Papelbon, is worth a hell of a lot more than, say, Washburn, who WAR values more. And Why? simply because he pitches a bunch more crappy innings.

    That makes no sense because a guy like Washburn could be replaced by a bar of soap, with roughly the same results, but top closers – and I mean real top closers, not someone who takes the mound in the ALCS and literally shakes in fear, like the overrated Nathan – are irreplaceable. Leverage may get notice with FG, but in no way is it fully respected, or, I suspect, understood.

    ‘The mind-boggling save fetish going on in Los Angeles’ is okay. Like Fish posted, top closers literally create the bullpen. So although Fuentes didn’t work out well for his 1st year in the AL, LAA still does understand that the dynamics of top bullpen starts from the top, and so were willing to pay to make that happen.

    BTW, if Paps earns less than $10 mil this year I would be shocked. Arb judge or not, Theo will be paying 8 figures for his services.

    Comment by Dirty Water — January 19, 2010 @ 8:28 pm

  126. $9.3 mil. Mind if I round that up ;)

    Comment by Dirty Water — January 19, 2010 @ 8:56 pm

  127. Ahh, almost any closer early is a waste in the fantasy realm. I personally don’t think Rivera is any exception. While Rivera is consistent and has the advantage of nice peripherals, he doesn’t actually net as many saves as many other closers annually. I’m pretty sure this is due to the Yankees winning by higher run differentials on average, when they win.

    Comment by B N — January 19, 2010 @ 9:04 pm

  128. I would argue that while most pitcher’s jobs are not to allow runs, a closer’s job is to ensure wins. I would rather have a closer that always gives up all but the tying run, but wins the game 100% of the time, than any closer in existence. In that respect, FIP is in fact a bad metric for closers. I think that’s pretty obvious to anyone who has seen the strategy of pitching as a closer. With smaller leads, you’re fine with loading the bases with 2 outs if it’s to your advantage. In some circumstances, a pitcher may be best off walking 2 extra guys in order to get the best chance to win (i.e. runner at 3B). This will be bad for FIP, needless to say.

    However, I don’t think this means FIP is necessarily a bad metric for relievers. If Rivera were just some guy in the relief corps, like Phil Hughes, would it still seem unreasonable that Vazquez is worth 3x as much? Because in terms of limiting runs, Hughes was about as good as Rivera last year. (shrug).

    Comment by B N — January 19, 2010 @ 9:37 pm

  129. A couple little points. Firstly, Rivera definitely outperforms his FIP. And the implication of my other post would be that Rivera could probably do that and lower his FIP if he were a 7th inning guy who only had to worry about giving up expected runs. What if FIP is not just anomalous for some results, but that it is actually misleading because that pitcher’s role is not to maximize FIP but to maximize retained wins and ties?

    Also, guys like Rodney shouldn’t get big money because they were historically awful at closing ;). What I don’t get is why a guy is given credit for closing experience when he still blows 25% of his saves. If mean-value analysis says that the average reliever should still save about 75% of opportunities- shouldn’t that be a warning sign rather than a great resume point? Rodney has blown about 26 saves against 70 saved. For comparison, Todd Jones saved 319 against just 33 blown. Rodney’s FIP is 4.15 career, while Jones is 3.95. While blown saves have too low of an N to be a good metric in most cases, there has to be a proxy metric which works at a smaller N.

    Tell me how a minor difference in FIP can lead to a massive difference in the ability to close out games, and we can all have a cookie. Considering that you’re saying FIP is good because it gets within about 0.2 of ERA in most cases, which is totally believable, I think the answer is that ERA is also a bad metric for closers. I mean, would you sign Jones if you knew his ERA would be 3.95? What about if you knew he’d close out higher than 90% of his ops? If I knew that, I’d basically say “Well, who cares about the ERA then?”

    There is an argument to made that teams are dumb about how they evaluate closers, which is a separate issue, but I would definitely state that WAR (due to using FIP) is not evaluating the correct metrics for closers.

    Comment by B N — January 19, 2010 @ 10:08 pm

  130. B N: If you are looking at Rodney’s saves versus blown saves throughout his career, it is bound to be misleading. He hasn’t been a closer for much of his career, right? Working as a set-up man, you get a hold if you do your job and a blown save (not a “blown hold”) if you don’t. That leads to unfairly low “save percentages.” A fairer (if still flawed) way of measuring these things for non-closers would be a save/hold percentage.

    Comment by WY — January 20, 2010 @ 2:05 am

  131. My order of preferance in evaluating closers
    1) Strikeout rate
    2) Walk rate

    I kind of get away from K/BB a little bit, here’s why:
    2 pitchers, both have a 3 K/BB. One is 12 K, 4 BB per 9, the other 6 K, 2 BB per 9. Both BABIP-against of .300.

    Pitcher A needs 15 outs in the field. Hits/(Hits+15) = .3, or about 6.43 H/9. + 4 BB/9 = 10.43 WH/9, or 1.159 WHIP.

    Pitcher B needs 21 outs. Hits/(Hits+21) = .3, or 9 H/9. + 2 BB/9 = 11 WH/9, or 1.222 WHIP.

    And of course, a hit is at worst as damaging as a walk. At best, it’s way more damaging. So of course, the first pitcher is better than the 2nd.

    xFIP > FIP, though.

    Comment by Joe R — January 20, 2010 @ 10:16 am

  132. I (as I noted below) still have a couple of problems with this.

    It’s biased in favour of guys who do well in relief. Let’s say you’ve got a guy who struggles as a starter, and he gets moved to the bullpen. If he continues to struggle in the bullpen, he either gets cut or he ends up in AAA or barely throwing any innings as a longman. Now, if the same guy struggles as a starter, then moves to the pen and becomes Joe Nathan, he’s going to pitch a lot of innings, and pitch them very well.

    The guys who have started, and then moved to relief will, overwhelmingly, be guys who didn’t do well enough as starters, but who did well as relievers. Sampling their ERA/FIP/whatever as starters and comparing it to what they did in relief is inherently biased and statistically sucky. Guys who suck as starters and suck even worse as relievers (I used the probably poor example of Tim Redding, as he’s not really thrown much relief) might as well stick around as 1-WAR 5th starters, where they at least have some vague value.

    Also, the two skillsets are inherently different. That DOES mean that yes, to some extent, starting is a more difficult job, but it also means it’s not apples vs apples. Relievers tend to have fewer plus pitches, but those plus pitches are often better than the average starter’s. Relievers may have more significant splits, which means they can be leveraged better in innings with a lot of LHB or RHB up. Yes, most (all) good starters could be very good relievers, but a lot of mediocre starters with several mediocre pitchers move to the pen and become, surprisingly, mediocre relievers.

    If we look at the number of relief pitchers who posted positive WARs last year, compared to every other role in baseball, the number seems to be crazily low, which suggests to me that either the job is much more difficult than we assume, or that the replacement level is not set as well as it might be: In the NL in 2009, there were only 70-odd relievers who put up positive WAR last year (30+IP), and of those, only 38 were better than 5 runs above replacement level. Intuitively, does that sound dubious?

    Comment by Felonius_Monk — January 20, 2010 @ 12:37 pm

  133. He didn’t do it! I adore The Sheen Man. How do you think he is dealing with the Rich, fameous and married? I truly don’t understand how celeberities get caught .

    Comment by Yuk Labreque — June 14, 2010 @ 9:27 pm

  134. i imagine it was a bit rushed, and overlooked some features that not many individuals speak about. Everybody is aware of that the majority new smartphones have web, so why present that primary perform at its bear minimum. Scroll up scroll down zoom in zoom out. Actually? which new telephone doesnt try this? How bout talk about how the text rearranges itsself. Also the texting, very poor review. Why didnt you mention you need to use the mic and text together with your voice? Left out ALOT of other more necessary features

    Comment by LGDare hacks — July 22, 2010 @ 5:07 pm

  135. Mark Eichhorn had a 7.4 WAR in 1986 without starting a game, so it certainly is possible.

    http://www.baseball-reference.com/players/e/eichhma01.shtml

    Comment by toycannon81 — October 4, 2013 @ 11:44 pm

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.

0.229 Powered by WordPress