On Valuing Closers as Hall of Fame Candidates

January 12, 2016
 On Valuing Closers as Hall of Fame Candidates


            The sabermetric community places a very different value on Closers than does much of the baseball world.  I am merely trying here to state what is obvious in a neutral, non-judgmental way.   A good many from the traditional sportswriting world feel that Trevor Hoffman is an obvious Hall of Famer.   In our community, we are unable to document the level of value in Hoffman’s career that would justify that.   

            In the recent Hall of Fame election Trevor Hoffman (Closer) received 67% of the votes although his career WAR according to Baseball Reference is only 28.4.    Mike Mussina (Starter) received only 43% of the vote although his career WAR is almost three times as great (83.0).  

            Lee Smith (Closer) received 34% of the vote although his career WAR is only 29.6.   Larry Walker (Outfielder) received only 15.5% of the vote although his career WAR was more than twice as great as Smith’s.   

            Billy Wagner (Closer) received 10.5% of the vote, enough to stay on the ballot, although his career WAR is only 28.1.   Jim Edmonds (Outfielder) received less than one-fourth of Wagner’s vote percentage (2.5%) although his career WAR is twice as high as Wagner’s (60.3).   In Fangraphs WAR, Edmonds is three times as valuable as Wagner. 

            Sportswriters talk about relief pitching as if every pennant race and almost every game revolved around the bullpen, but here is the problem.  Trevor Hoffman in his last ten seasons in the major leagues (2001 to 2010) was credited with 330 saves—33 per season—but pitched an average of less than 51 innings per season.     Mike Mussina in his last ten seasons in the majors (1999-2008) pitched just short of 200 innings per season (199.4).     In his major league career, Mussina pitched 3,563 innings, and Hoffman pitched 1,089.   For Hoffman to be more valuable than Mussina, then, he has to be three to four times more valuable per inning pitched.  

            We in the sabermetric community . . .I am averse to speaking for the town, but just trying here to state what could be agreed upon, I think.   We in the sabermetric community agree that the innings pitched by closers have a disproportionate impact on the won-lost record of the team.   One inning pitched by a Closer counts more than one inning at random pitched by a Starting Pitcher; we agree with that.    What we disagree about is the extent.   The "traditional" camp thinks the ratio is 3-1 or higher.   We think it is lower.

            I am not quarrelling here with the position of the sabermetric side.   I agree with the position of the sabermetric side.    If I were to be gifted with a Hall of Fame ballot, I would certainly vote for Mike Mussina and Tim Raines and the fellas before I would vote for any of Glenn Hoffman’s relatives. 

            However, while I agree with our side of the argument, there are a couple of dissenting points that I think should be made.   First, while the argument against relievers having value comparable to other players may well be solid, it is not intuitively obvious, and I’m not actually sure that I understand all of it.   If I don’t understand the mathematics behind an argument, it seems to me, it is not that likely that the traditional sportswriting world will understand it.    And second, perhaps the discussion about the Hall of Fame should not, perhaps, be turned into a subset of WAR?

            On the first point, look at it this way.   Let us suppose that value for a pitcher consists in having an ERA below 5.00.   Just saying. . .a pitcher with an ERA of 5.00 isn’t really worth very much.   Mike Mussina in his career pitched 3,583 innings with an ERA of 3.68.   That makes Mike Mussina 420 runs better than a pitcher with an ERA of 5.00.     Trevor Hoffman pitched 1,089 innings in his career, with an ERA of 2.87.   That makes him 227 runs better than a pitcher with an ERA of 5.00.     Mussina 420, Hoffman 227.

            Suppose, however, that Mussina has a leverage index of 1.0, and Hoffman a leverage index of 2.0.   Then, applying the leverage, Hoffman would be ahead, 454 to 420.

            Look, I understand generally what is supposed to be wrong with that math.   It assumes that the replacement level for a pitcher is an ERA of 5.00, while Baseball Reference and Fangraphs both apparently believe that the actual replacement level for Mussina, given his league ERAs and the parks that he pitched in, was somewhat over 6.00.   When you make the replacement level ERA higher, that works to the advantage of Mussina as opposed to Hoffman.  But my point is, if the pro-Mussina, anti-Hoffman argument is not obvious to me, then it isn’t obvious.   Expecting your typical baseball fan to get it is like ordering from McDonald’s and expecting to receive a culinary masterpiece. 

            This argument is also flawed because it assumes that the replacement-level ERA for Mussina is the same as it is for Hoffman, which is not true.    Baseball Reference apparently feels (if I understand their math) that the Replacement Level for Mussina is an ERA over 6.00, which seems reasonable, while it (apparently) believes that the Replacement Level for Hoffman is in the fours, which seems. . .well, I’m not SURE that’s reasonable; maybe it is, maybe not.   The American League ERA from 1991 to 2008 (Mussina’s years) was 4.54, while the National League ERA from 1993 to 2010 (Hoffman’s years) was 4.29, so there is that, and the park factors for San Diego were almost always in the 80s, so there’s that.  

            From my perspective, the WAR for Mussina versus Hoffman may be presumed to be more right than wrong, but it is not absolutely, totally, completely and perfectly RIGHT, either.   There are issues.   First, the replacement level is just an estimate.   Mussina’s run advantage over Hoffman is much larger if you assume a replacement level (for both pitchers) of 6.00 than if you assume a replacement level of 5.00—and it may well be that the replacement level should be closer to 5.00 than to six.  

            "Replacement Level" is an important but imprecise concept.   We have fallen into the habit of referring to Replacement Level as if this was a known constant in terms of winning percentage contribution, when in reality it is an unknown variable.   Some times, some teams, the pitching is so good that nobody needs an average pitcher, whereas other times, other teams, there isn’t an average pitcher on the team, and if you could find somebody who had a .250 effective winning percentage, that would be super. 

            In a more sophisticated sabermetric analysis, we would recognize that Replacement Levels vary widely from team to team, and that sometimes they are around .300, but equally often they are over .350 and many times they are over .400.   The Replacement Levels that the analysts are using now ARE too low IF THEY ARE CONSIDERED TO BE FIXED.   I think they’re using a Replacement Level around .310.   The real replacement level is USUALLY higher than that; it varies, but if you are using one figure to represent the whole world, .310 is too low.   Most teams, most of the time, can find a player better than that.  

            Second issue, the "Leverage Index" applied to Hoffman’s role POSSIBLY, in my view, could be larger than the numbers used by current analysts.

            Third issue, not sure about this because I don’t really understand other people’s methodologies, but some of the people who are figuring WAR may be adjusting the replacement level for Hoffman because he is a Closer.   IF they are doing that—I don’t know whether they are or not—but if they are doing that, they shouldn’t be doing it, and they need to take that out of there.   (I may have advocated this myself in the past.   If I did, I was wrong.)  It confuses the discussion.    The leverage index has to be calculated based on normative performance.   If you credit Hoffman with a "leverage index" but then lower his replacement-level ERA because he is a closer, all you are doing is giving him a break on one hand and then taking it away on the other.    In other words, you’re confusing "leverage" with "replacement level", so that (if you are doing that) you’re not ACTUALLY crediting him with a leverage index, at all.  

            Fourth issue, some people who figure WAR may be (again, not certain who is doing what) but some people may be calculating value based not on actual runs allowed, but on formula estimates of the number of runs the pitcher person could have been expected to allow based on his peripheral numbers.    Trevor Hoffman has a 2.87 career ERA, but 3.08 FIP (Fielder Independent Pitching) based on his strikeouts, walks, and home runs allowed, whereas Mike the Moose has a 3.68 career ERA but a 3.57 FIP based on his peripherals.   

            Again, questionable adjustment.   Pitchers do lots of things that have SOME impact on their ERA, other than getting strikeouts and allowing walks, home runs and hit batsmen.    They also:

            a)  Hold baserunners well or poorly,

            b)  Pick runners off,

            c)  Field their position,

            d)  Throw Wild Piches,

            e)  Induce ground balls, and

            f)  Pitch to the situation at a certain level.

            When figuring data for a SEASON, the discrepancies between ERA and FIP are probably mostly random factors which the pitcher does not control, and we should probably prefer FIP to ERA.    But when dealing with CAREER records of several hundred innings—and we are dealing with career records here—when dealing with career records, it is much more likely that the discrepancies between ERA and FIP are created by factors (a) through (f) above, and then the actual ERA is almost certainly the more instructive number.  In a more sophisticated sabermetric analysis, we would rely more on FIP when dealing with small data groups, but we would rely more on actual ERA when the number of innings for an individual pitcher is larger.

            There is also a little issue of ERAs of starters vs. relievers. . .when a starter and a reliever share an inning, the starter tends to get charged with the run allowed no matter what, which can queer the ERA.   But that’s not a real issue for modern closers, because modern closers usually enter the game at the start of the inning.    It’s a legitimate issue for pitchers before 1995, and it’s a legitimate issue for 7th- and 8th-inning pitchers.  

            I believe that 30 years from now, when sabermetric analysis is more sophisticated than it is now, our calculations will still show Mike Mussina as much more valuable than Trevor Hoffman.   But I think it is possible, and indeed likely, that it will show a somewhat smaller difference than current analysis is showing.


            Now, the second major issue; I raised two major issues at the start of the article.   I know I am always doing first point and second point and third point and it is confusing; I’m sorry.  Anyway, my second point was that the Hall of Fame discussion, perhaps, should not be treated a subset of the VALUE discussion, or the WAR discussion.  

            The Hall of Fame is not about value—I am just being the Devil’s Advocate here—the Hall of Fame is not about VALUE, it is about EXCELLENCE.    Perhaps his managers should have assigned Trevor Hoffman a different role.   Perhaps conventional wisdom placed Trevor Hoffman in a 55-inning role when it should have put him in a 90-inning role, and perhaps conventional wisdom had him protecting three-run leads because three-run leads will get you a save but are not actually a high-leverage situation. . .perhaps, perhaps.    This isn’t about value, it is about excellence, and about historic standards.    In the role that he was assigned, Trevor Hoffman had historic impact.  

            Here again, a fallacy.    The Hall of Fame is (unarguably) about historic performance.    But we really don’t know yet whether Trevor Hoffman’s 600 saves are or are not a historic performance.   Closers have only been in their modern role, piling up 45 saves a year, since about 1990.   It may be that, in 30 years, there will be 20 pitchers with 600 saves, and it may not be all that notable.  

            I think the sabermetric community is guilty of a confusing what is permanent—Hall of Fame selection—with what is temporary (our current best estimates of player’s value), but I also think the traditional sportswriting world is guilty of the same thing, in assuming that Trevor Hoffman’s 600 Saves are a historic accomplishment, when in fact they may not be.   I’m not sure I would vote for Trevor Hoffman over Lee Smith or Billy Wagner; in fact, I think I probably wouldn’t.   I probably would vote for Smith or Wagner first, despite the career Saves number. 

            But would the Hall of Fame be better, without any Closers?   Would a baseball team be better without a Closer?    I don’t think so.   I think you have to have Closers in the Hall of Fame, regardless of what Baseball Reference thinks their WAR is.   I don’t think you can make the Hall of Fame discussion into a wholly owned subsidiary of WAR.  

            Sophisticated measurements are temporary instruments.   If they weren’t temporary instruments then we should all retire, because we would have nothing more to say to the baseball world at large.   Just run the numbers, dude; that’s all we know.   I don’t see it that way.   I think that WAR (and other measurements) are useful tools that WE use to try to see the truth, but that when they start to dictate to us who we can vote for and who we can’t, then it’s time to bring out the Tasers and drive them back into their cages.  

            One reason that sportswriters believe in the super-importance of Saves (and Closers) is that they believe that certain types of losses are particularly damaging.  You blow leads in the ninth inning, those losses undermine the confidence of the team, and cause the team to lose momentum.

            This is not an inherently unreasonable thing to believe.   In the sabermetric community we are not big fans of momentum, for two reasons.   One, we’re skeptics; we tend to believe only in what we can document to be true, which is sometimes a nasty habit.   Second, there have been thousands of studies of "momentum" in various ways, all or almost all of which have failed to establish that there is any value in momentum.   A hot team has zero tendency to remain hot; a hot hitter is no more likely to get a hit next time up than a hitter who is in a slump.  

            Studies searching for momentum always fail, but this does not entitle us (on our side of the canyon) to dismiss out of hand a possibility such as "certain types of losses are particularly damaging."   We have to assume that that could be true unless or until it is demonstrably shown to be false.  

            If you ask me, do I believe honestly that traditional sportswriters are exaggerating the importance of Saves by asserting things that are not true, such as that late-inning blown leads are particularly damaging?   Yes, that is what I am inclined to believe.   What I am NOT inclined to believe is that we have all of the issues completely figured out, and we should cancel the rest of the discussion right now.   We don’t have the world all figured out, and we don’t have all the answers.   I’m not voting for Trevor Hoffman for the Hall of Fame, but if somebody else wants to, it’s really not my place to pass judgment on his judgment.   







COMMENTS (29 Comments, most recent shown first)

Just thinking out loud here, but as a warning, I don't like closers, and I don't like comparing them to starters. Mariano Rivera is probably accepted as the best closer ever, and he couldn't make it as a starter. I fear that he will be first player ever voted unanimously into the Hall of Fame.

The closer is, to my mind, more comparable to the career pinch hitter-- Manny Mota, Jerry Lynch-- than to a starting pitcher, despite the obvious disparities. Curiously, though, no one has ever, to my knowledge, argued that Manny Mota belongs in Cooperstown.

Mariano seems most often to be characterized as a one-pitch pitcher. I am not aware of ANY successful one-pitch starter, ever. As a closer particularly, his job might be to get the 9-1-2 batters one day, 4-5-6 the next, then not face that team again until a month later. He didn't really have to be a "great pitcher." And he wasn't. He was "only" a great closer (I'm not trying to say that he wasn't a great closer).

For me, while I understand that closers' innings are to be leveraged, it's going to take considerable calculation to resolve all the relationships... 1-run lead, heart of the lineup, September; 2 run lead, game 3 of the season, top of the order. As Chuck Berry might say, "too much monkey business."

My "first final" thought-- rather than include closers in the regular Hall of Fame voting, have a special election in every year that ends in 5 and select the closers from the previous decade who have demonstrated the appropriate excellence, obviously allowing for a little "decade stretching" as needed, in something that amounts to a separate vote.
10:21 PM Mar 2nd
Why wouldn't we have different replacement levels for SP and RP? No full-time RP has more IP than any full-time SP, even back in the 70s when relievers went 3+ innings. So with that concrete divide, we can (and should) assign different replacement values for each.

It's a sample-size thing.
It's a pitch-to-pitch effort thing.
It's a 'the ERA stats show this difference' thing.

I haven't researched it, but I'm fairly confident the league-average RP ERA has never been higher than the league-average SP ERA. Another divide.

I see no reason to fight what seems to me as something plainly obvious.
7:19 AM Jan 24th
As Tango has mentioned, there's a lot more to this honest mess even than has been discussed here. Leverage and inherited baserunners aside, relievers--especially closers--usually pitch only an inning, which means they go all out. In addition, a closer who enters a one-run game can't afford to let a run score. A starter is often in a situation where the team will trade a run for an out. (First inning (or later with a lead), runner on third with none or one out, teams often concede the run. That affects the starter's ERA, where for a closer in a close game the team may have to avoid that. On the other other hand, pulling the infield in increases the chances of prolonging the inning if the play isn't made to cut off the run.

It's a swamp.
10:36 AM Jan 20th
I'm curious about the fact that all of these comparisons between starters and relievers are based on innings. True, an average starter probably pitches two or three times as many innings as the average reliever. But the reliever pitches in two to three times as many games. The object of the competition is to win games, not innings. Is this meaningful? Just asking: I don't know.
4:17 PM Jan 15th
By the way, that's a big compliment that Lewis gave James. It indicates that there's a lot of considerations, and there's a lot of work left to do, so whatever we do, it's always a work-in-progress.
9:06 AM Jan 14th
There's a concept called "chaining" that is more appropriate for relievers than other spots. But that's basically what happens. When a starting player goes down, he gets replaced by a bench player, who himself gets replaced by the AAA guy.

It really gets into the intricacies of trying to create such a system, which makes it complex and not transparent. If someone wants to do that, go right ahead. You'll gain 1% for an extra 50% of effort.

This is similar to creating a forecasting system. I created Marcel as the absolute simplest forecasting system you can envision. You can certainly make one better by increasing the complexity by a factor of 10. As others have done. But the gain is so small as you can't really see it unless you take out a magnifying glass.

As Michael Lewis noted of Bill James: better to put out an honest mess than a tidy lie.​
9:04 AM Jan 14th
I hate to annoy people with my ignorance, but it makes more sense to me to have a separate replacement level for bench players than for relievers.

Working on a project about the greatest infields of all-time, I noticed many of them didn't have a single player on the bench with a positive WAR (using Baseball-Reference out of ease of reference more than statistical preference). The vast majority of back-up infielders had negative WAR. You would think that those infields would be so hard to crack, that they would have above average back-ups. They might - just saying that based on these samples, the replacement values don't appear to be too low at all.

However, as Bill has demonstrated in a post here last year, that replacement level is somehow low when comparing starting players. I don't seriously advocate having separate replacement levels for subs and starters, because it is inelegant and too difficult to separate the two. However, the standards are clearly different. Further, a bench player needs to be able to play multiple positions and deal with playing infrequently. Those skills don't show up in WAR.

We don't want a negative value for so many Major Leaguers, so merely raising the replacement level won't work. What we need is an increasing bonus for how far above replacement level a player gets. Unfortunately, that makes the stat more abstract - an excellence index rather than simply wins above replacement.
8:52 AM Jan 14th
mweddell: Thanks. I didn't think it was far from the theme of Bill's article to ask whether Ortiz was the greatest DH ever because the historical problem applies to relievers as well, just not as starkly as it does to DHs, who didn't exist at all before 1973. The discussion of what 600 saves means historically was what got me thinking about how the DH question is similar and I was trying to illustrate that. The way relievers are used, it seems to me, continues to evolve, which makes putting any individual reliever into some kind of context challenging. The way second basemen are used, not so much. Mariano was John Wetteland's set-up man for a while in NY. What if John Wetteland had been Trevor Hoffman and got 600 saves, and Mariano had been his lights-out set-up man for 15 years? So far fewer saves for Mariano, but the same number of innings pitched, strikeouts, ERA, and so on.....Would he even be in the conversation? Could a reliever as good as Mariano even have been a set-up man for that long? What if Mariano and Wetteland had shared set-up and closer duties?
7:26 PM Jan 13th
"It gives them both an advantage (lower ERAs) and disadvatage (fewer innings)."

Right, that's exactly correct. The question is how do you incorporate that. It's not clear that using the same ERA as a comparison point should be the same for RP and SP. If we go ahead and use an ERA that is 1.00 lower for RP, then you make a good point that perhaps being forced into lower IP should be considered as well.

And yes, leverage is yet another variable.
2:53 PM Jan 13th
Woops, no spell checker: I meant "tangental".
2:16 PM Jan 13th
Tango, that's why I qualified it with "I think" - leaving it open to other opinions. I don't understand why you fouind that annoying.

It is tangetal to the point I was trying to make - that closers are used very differently than other relievers and that should be considered with regards to replacement level RA.

I'd be happy to reopen the discussion on the other reasons relievers should have a different replacement value. I don't see how the fact that they go fewer innings and therefor can go all out should matter. It gives them both an advantage (lower ERAs) and disadvatage (fewer innings). It is up to their manager to manage that as best as possible. If it is more important to control the end game, then leveraging comes in, which is a different part of the formula, isn't it?

If relievers should have a leveraging advantage, though, then there should be some way to differentiate between set-up men and mop-up men. As a group, I THINK these should cancel each other out: relief innings are not more important that starter innings. In the old days (pre late '80s), I suggest that relief innings were less important on the whole than starter innings. Anyway, even if you are factoring in the leverage differences as part of the replacement RA, the difference as a group should not significantly favor starters. That's my opinion based on what I know and have observed.
2:14 PM Jan 13th
"which is the only legitimate reason, I think"

I get annoyed when I see something like that. There are legitimate reasons beyond what you are saying. We've said it on this site. You may not agree with those reasons, but they remain legitimate.

1:17 PM Jan 13th
We in the sabermetric community (if I may include myself) have long accepted the difference in replacement value between starters and relievers. However, as Bill points out in his paragraph about that difference, it should not apply to modern closers - at least, not for reasons having to do with inherited runners (which is the only legitimate reason, I think, for having a difference in replacement value between reilevers and starters), because closers generally come in at the start of an inning. It sounds so obvious, but it is a huge take-away.

To be really fair, we should have just one replacement level for starters and reliever alike, but charge a percentage of inherited runners who score to the reliever and a percentage to the relieved - based on the number of outs and places on the basepaths (or a simplified version that splits them in half).
12:30 PM Jan 13th
Studes, great timing!

I just posted this on my blog:
10:33 AM Jan 13th
Not to confuse the issue, but you also have the concept of chaining going on here. Relievers are fairly unique in baseball because the context in which they perform isn't random, but chosen by the manager. You don't replace your closer with your replacement level guy--you replace him with your setup man. The new LI situations go to the setup guy and his game impact increases from where it was (but not to the level of the previous closer). His LI situations go to the next-best reliever, and so on. This somewhat mitigates the loss of the closer.

This is a different situation than that with starting pitchers, because each game can be presumed to be as important as any other game (at least given the way we manage rotations these days). So when you lose your #1 starting pitcher, he is effectively replaced by your sixth starter in terms of game impact.

Not sure if I've explained this well, but WAR factors it into their calculations.
9:10 AM Jan 13th
Flying Fish, your question about whether David Ortiz is the greatest DH ever might be moving pretty far away from the content of this article. I would rank them Edgar Martinez first, David Ortiz second, Paul Molitor third and Harold Baines fourth. However, an argument could be made that Ortiz' post-season accomplishments push him ahead of Edgar.
8:23 AM Jan 13th
I am not sure I understand everything in Bill's article or the comments but I found both useful in helping me think about the issues. I had concluded that the discussion applies also to DHs, and so I was pleased that Tom and others raised that point. I also think the "historic" question is important. Is David Ortiz the greatest DH ever? I think he is, but "ever" is only 42 years if my history and arithmetic are right. I don't think anyone would have thought of putting a DH in the HOF whose career ended in 1983 as soon as he became eligible purely on his standing as a DH (is there anyone who was primarily a DH in the period 1973-1983?). I don't know if 42 years is long enough, but I also think that Ortiz--while a slam-dunk if you consider only whether he was the greatest DH ever--also merits consideration just as a hitter in general. Then of course there are the matters of there being no NL DHs, and that the way the way managers use the DH continues to evolve. I think I'm glad I don't have a HOF vote.
10:08 PM Jan 12th
Manny Parra might be a good example to the extent we want to have a face of a recent player, without necessarily being beholden to this one guy.

He's a poor pitcher. As a SP, career 5.44 ERA. As a RP, career 4.01 ERA.

If we want to compare him to "league average", do we just go with whatever it was for the seasons in question, say 4.00 ERA? Maybe we do, and maybe we give him an individualized W/L record of 7-13 as a SP and 6-6 as a RP. This I think is Bill's point. This "conserves" the wins, and it keeps the win% aligned with the runs allowed. And we can stop here. (In WinShares LossShares parlance, that's 21-39, 18-18 respectively.)

But then, do we want to go a step further and say that a poor pitcher would have a 6-14 "individualized" record as a SP and a 5-7 "individualized" record as a RP?

This last step is the one that is the contentious step.
5:20 PM Jan 12th
OK, I'll try to be more articulate in my position. . .think it through better, and be open to reconsidering. That's a goal.
3:20 PM Jan 12th
I agree that we've got the winning percentage issue to contend with that we won't be able to get everything to balance out. From that standpoint, I am not 100% married to my position. Just say 90%.

If we try to learn from other sports, we can take football, and the idea of halfbacks v fullbacks, or wide receivers v tight ends. Because they are given different roles, different handcuffs, could we just simply compare a running yard to a running yard and a passing yard to a passing yard?

Or, would we take a "replacement level receiver", and put him as a TE and put him as a WR, and then compare Jerry Rice to the "repl-level receiver as WR" and compare Brent Jones to the "repl-level receiver as TE"?

That's where I take the analogy for the pitchers.


However, as someone else noted, you have the number of innings messing us up. After all, if you had 9 replacement level pitchers as relievers pitching in a game (with average offense), this team would win say 47% of the time. But 1 replacement level pitcher as a SP (and 3 repl level pitchers in relief) in a game would win say 35% of the time.

This is basically where Bill is going, that things won't add up.

As I said, I'm not convinced that I'm right, and I agree there are a lot of considerations to balance.
1:45 PM Jan 12th
The best closers and designated hitters should be in the Hall of Fame. We should ask what they could do, not what they couldn't do. Like it or not, these are legitimate positions.
12:36 PM Jan 12th
From Tom's first post:

This is based on the position that a "replacement level pitcher" AS A RELIEVER would put up a much lower RA/9 than if that same replacement level pitcher were asked to start.

I believe that I have advocated this point in the past, in private discussions with you and with others, but this is wrong and we absolutely should not be doing it. Why? Because the replacement level is tied to a winning percentage. . . a .310 winning percentage or something, which should be higher than that but that's another issue. But while it may be true that a replacement-level closer can post an ERA not that much above league average, THIS IS IRRELEVANT because that pitcher will not have a .310 winning percentage (assuming that .310 is the replacement level.) So what you are in effect saying is that the replacement level is .310 for every player EXCEPT RELIEF PITCHERS. By doing this, YOU entangle the issues of replacement level and leverage index in such a way that you are not, in fact, applying any leverage index at all--thus underestimating the value of a Closer.

12:18 PM Jan 12th
But if you say, as Tango does, that the conditions and circumstances under which a closer performs allows him to perform at a better ERA level, can't you also say those same conditions and circumstances limit the total innings he can pitch? Closers are sprinters, and starters are distance runners. It is the nature of the closer's role that they pitch less innings, and it is the nature of the starter's role that they pitch more. So you assign a lower ERA to the closer's replacement level, thereby reducing that half of the equation, and then multiply it by their naturally smaller innings total. Doesn't seem fair to the closers.
12:17 PM Jan 12th
The "role" has an impact on replacement level in other parts. For example, a player, acting in the role of PH, performs far far worse than he would as a starting player. In wOBA parlance, such players take a 30 point hit, which is equivalent to 30 OBP points and almost 40 SLG points. Of course, since PH are players we don't generally care about, in addition to not having the quantity in times at bat, we don't care much either way.


DH are also in a similar position, taking a 15 point hit in wOBA (15 in OBP, 20 in SLG). So, a player performing as a 1B will hit better than if he performs as a DH. In other words, he's got one hand tied behind his back.

This is DIFFERENT to comparing the hitting of SS to other SS and 1B to other 1B. This is a role adjustment.

DH complicates us a bit because you could argue that "lifers" might have adapted to the role of DH, and so, aren't operating with a hand tied behind their back. Not to mention that as DH, you can argue they get to bat more often than if they were forced to play the field.


So, that's where the role adjustment comes for SP v RP. The baseline level for John Smoltz, on a rate basis, has to be different if he gets a 2.50 ERA as a reliever compared to a 3.25 ERA as a SP.


Leverage is a separate discussion point.

11:35 AM Jan 12th
"But would the Hall of Fame be better, without any Closers?" No, but some of us wish that we had gone more slowly in inducting relievers to make sure they were historically excellent. I have no problem with Hoyt Wilhelm, Dennis Eckersley and (when he becomes eligible) Mariano Rivera in the Hall of Fame. The problem is that when Rich Gossage, Rollie Fingers, Bruce Sutter and any of Hoffman / Wagner / Smith go in then it is difficult to clearly distinguish between that group and pitchers not in the Hall of Fame without inducting 15 other relievers. Your mileage may vary.

Bill's article mentions that it is unclear whether 600 saves is a historically significant total. At one point, I recall that folks thought that 300 career saves was notable, publicizing it as "a club" like the 500 home run club. Today, it's obvious that 300 career saves is no longer very notable.
8:47 AM Jan 12th
I agree with Tom's point too. More specifically, I disagree with Bill's statement:

If you credit Hoffman with a "leverage index" but then lower his replacement-level ERA because he is a closer, all you are doing is giving him a break on one hand and then taking it away on the other. In other words, you’re confusing "leverage" with "replacement level", so that (if you are doing that) you’re not ACTUALLY crediting him with a leverage index, at all.

It seems to me that the replacement-level ERA for a reliever is lower than that for a starter (the fact you can take a failed starter and make him a useful -- and sometimes a star -- reliever speaks to this point), and this is totally separate from leverage index.
8:09 AM Jan 12th
Agree with Tom. The reason we talk about "replacement level" is because we are trying to imagine would have happened to the team if Borderliner had not been playing for them. If Mussina and Hoffman had not been there, it seems reasonable to think that Mussina might have been replaced by someone with an ERA around 6, while Hoffman's replacement might have had an ERA around 5 (and a lot of saves).

This is completely separate from the issue of leverage- the lower replacement ERA would apply to all one-inning pitchers, even to Wade Boggs pitching in a blowout.
7:43 AM Jan 12th
Generally speaking, the replacement level of a SP is about 6.00 RA/9 and a RP is 5.00 RA/9 in a league average of 4.70 RA/9.

This is based on the position that a "replacement level pitcher" AS A RELIEVER would put up a much lower RA/9 than if that same replacement level pitcher were asked to start.
7:12 AM Jan 12th
Just a quick follow on point to Bill's question - "But would the Hall of Fame be better, without any Closers?" - it would seem somewhat strange to try to describe the game over the last 20+ years without representing one of the more prominent positions.
5:57 AM Jan 12th
© 2011 Be Jolly, Inc. All Rights Reserved.