Relief Pitchers and the Hall

January 4, 2016
  
I’ll be posting the results of the 2016 BJOL HOF vote tomorrow, right in line with the BBWAA announcement. Just before we get to that, I’d like to touch on one of the questions that our HOF ballot shares with the BBWAA ballot.
 
How should we consider the candidacy of relief pitchers for the Hall of Fame?
 
It’s a vexing question. Lee Smith had a record of 71-87 with 478 saves and a 2.98 ERA. John Franco had a record of 90-87 with 424 saves and a 2.89 ERA. Lee Smith has been on the BBWAA Hall of Fame ballot for thirteen years, where he’s gotten as much as 50.6% of the vote. John Franco, a direct contemporary of Lee Smith, received 4.6% of the vote in his first year of eligibility and was promptly dumped off the ballot. How does that made sense?
 
Bruce Sutter was 68-71 with 300 saves and an ERA of 2.83. Doug Jones is 66-78 with 303 saves, and an ERA of 3.29. They weren’t direct contemporaries, and offensive contexts probably get their ERA’s a little closer. Bruce Sutter is in the Hall of Fame. Doug Jones isn’t.
 
So how do you sort through the modern closers, and determine which candidates to support for the Hall?
 
Part of the challenge is that it doesn’t seem particularly hard to be an effective closer: every year we see a dozen reclamation starters switch to the bullpen and post ERA’s that would make Bob Gibson jealous.
 
There’s a sense that good closer are failed starters….that they are lesser than starters. Wade Davis might be the best closer in baseball, but he wasn’t great as a starter. Luke Hocheaver was great in the bullpen, but couldn’t hack it as a starter. Ryan Madson has a 6.82 ERA as a starting pitcher, 2.94 out of the bullpen. That’s just one team….I haven’t left Kansas City.
 
Another part of it is that it’s a high turnover position: if you want to know who the best 1B in baseball is going to be in 2016, you’re going to pick from a small group: Goldy or Rizzo or Votto or Miggy. Try picking the best closer, and you’d have to cast a wider net: Kimbrel and Chapman are obvious choices, but Kimbrel ranked 18th in fWAR last year, while there’s a good chance Chapman isn’t the best relief pitcher on the Yankees in 2016. Cody Allen led the majors in relief pitcher WAR, and Melancon led in saves. Roberstson, Familia, Betances, Giles, Rosenthal, Davis, Jansen, Britton….take your pick.
 
The third part is that the position is still in flux. The usage patterns of closers have evolved drastically over the last fifty years, and there’s strong evidence that that evolution is continuing.
All of those elements work against the candidacies of closers. It’s tough to care about a failed starter when you are considering a candidate’s career against the careers of successful starters. It’s tough to judge a pitching career that is sometimes 800 or 1000 innings in length. And it’s tough to judge someone by what the position was in their era, instead of by what the position is now.
 
It’s a difficult process.
 
*             *             *
 
There’s another issue, which is that the Hall has seemed a little iffy about which closers they elect. Rollie Fingers was elected because he was, for a brief moment, the all-time saves leader. And because he won an MVP. Then they elected Sutter, who seemed to be a little short on ‘career length’. Then they went with Gossage.
 
The message seemed to be that 300 saves was a standard of greatness: get to 300 and you’re a candidate.
 
It is easy to see, in retrospect, why 300 saves wasn’t a particularly good metric for pitching greatness. Hell, it was easy to see at the time why that was going to wind up a poor benchmark. A good closer will save 40 games a year, which means that an eight year run as a good closer was a Hall-of-Fame career.
 
If we upheld 300 saves as the benchmark, we’d have 27 closers either in the Hall, or posing for their plaque. Jason Isringhauser has 300 saves. Jose Mesa has 321. John Wettland has 330. Randy Myers has 347.
 
And saves are a stupid stat. They aren’t a particularly useful way of judging performance. We all know this.
 
*             *             *
 
Because of this, a lot of us throw our hands up and ignore closers. We’ll vote for Mariano Rivera, but that’s it. We’re going to ignore Lee Smith, or Hoffman, or Wagner, because there’s just no way to differentiate them. If you vote for Hoffman, don’t you have to vote for Lee Smith? Asking that differently, what characteristics would make you support just one of those pitchers?
 
I think this is a mistake. Like it or not, relief pitchers are hugely important figures in the game. If the Hall of Fame exists to honor the best players of each era, it should honor the best closer of this era.
 
So how do we decide which candidates to support, and which ones to ignore?
 
*             *             *
 
You ever make up junk stats?
 
I do. It’s just something fun to do, some way to fiddle with the numbers. Bill makes up real numbers, numbers that attempt to answer something decisively. I just do it to have a little fun. I don’t put any stock in any metric I invent, and I wouldn’t encourage you to do it, either. We’re just having some fun.
 
So junk-stat time.
 
I was trying to figure out how to evaluate closers….how to figure out what was a reasonable benchmark to judge a relief pitcher.
 
I started with WAR…with the FanGraphs version. I did that because it’s really easy to build spreadsheets with their database. God bless everyone at FanGraphs.
 
Career WAR, for relief pitchers, is pretty unsatisfying.  Trevor Hoffman has a career WAR of 26.1, which is about what Mike Trout put up in his first three years in the majors. That seems like very little, but it ranks Hoffman third among closers since 1950, behind Mariano and Goose.
 
Just an aside: Mariano ranks first in everything. Any way you try to look at closers, with junk stats or fancy metrics or just ‘saves’, Mariano is ahead of everyone, by a huge margin. He almost ruins the conversation…no one compares to him, and no one really comes close.
 
Anyway, I adjusted the WAR of relief pitchers to ‘per-60-innings-pitched’, but that didn’t really help. I got a list with 1.8 and 2.2 and 2.5 and 1.9. How you do separate the great pitchers from the pretty good ones. The number was too small to ‘see’ the mark.
 
So I jumped it to WAR/200 IP. Counting only players who pitcher 600+ innings as a reliever, Here are the top-ten:
 
Name
WAR/200 IP
Mariano Rivera
6.4
Billy Wagner
5.4
Dennis Eckersley
5.1
Trevor Hoffman
4.8
Lee Smith
4.1
Doug Jones
4.0
Goose Gossage
3.7
Fr. Rodriguez
3.7
Bruce Sutter
3.7
Tom Gordon
3.7
 
This is a pretty good list. Mariano is first, and it’s not surprising to see the likes of Wagner, Eck, Hoffman, and Smith on the list. If you told me those were the top-five relief pitchers in baseball, I’d listen.
 
But it’s not a perfect list. You can probably guess why.
 
Tom Gordon has a WAR/200 of 3.7, but his career as a relief pitcher was 857 innings long.
 
Rollie Fingers had a WAR/200 of 3.3, but his career as a relief pitcher was 1553 innings long.
 
Who you gonna take?
 
You’re taking the moustache. No hesitation. Rollie Fingers might’ve been the wrong candidate for the Hall, but he’s a better candidate than Tom Gordon.
 
So we need to balance things out.
 
The tension with relief pitchers is similar to the ‘peak-versus-longevity’ debate that we have about position players. Should Nomar Garciaparra be in the Hall of Fame because he was really great for six years, or should Rafael Palmeiro be in the Hall of Fame because he was pretty good for twenty years?  
 
I had already calculated WAR/200 innings pitched. To balance things out, I backed up to find out just how many 200 IP sections a pitcher had. Really simple: I divided each pitcher’s innings pitched by 200, and rounded to the first decimal.
 
Here’s the top-ten on that list. I’ll give you a dollar if you can guess #1.
 
Name
IP/200
Hoyt Wilhelm
9.2
Lindy McDaniel
9.0
Goose Gossage
7.9
Rollie Fingers
7.8
Gene Garber
7.5
Kent Tekulve
7.2
Tug McGraw
7.0
Sparky Lyle
7.0
Clay Carroll
6.6
Don McMahon
6.6
 
Rollie’s on the list again. So is Hoyt Wilhelm, the one HOF reliever who didn’t make the first list.
 
You know what I did next?
 
I said this was a junk stat, and it’s about to get very junky. Really simply, I added the first column to the second column. I did addition. That’s horrible. That’s kid math.

And I got this list:
 
Name
WAR/200
IP/200
RP Score
Mariano Rivera
6.4
6.2
12.5
Goose Gossage
3.7
7.9
11.5
Hoyt Wilhelm
2.1
9.2
11.3
Lindy McDaniel
2.2
9.0
11.1
Rollie Fingers
3.3
7.8
11.1
Lee Smith
4.1
6.3
10.4
Trevor Hoffman
4.8
5.4
10.2
Billy Wagner
5.4
4.5
9.9
Doug Jones
4.0
5.5
9.5
Gene Garber
1.9
7.5
9.5
Kent Tekulve
2.0
7.2
9.2
Dennis Eckersley
5.1
4.0
9.1
Sparky Lyle
2.1
7.0
9.0
Tug McGraw
2.0
7.0
9.0
Bruce Sutter
3.7
5.2
8.9
 
For such a dumb stat, it’s an oddly perfect list. It gets all of the pitchers elected to the Hall of Fame: Goose, Wilhelm, Fingers, Eck, and Sutter. And it gets all the guys who have come closest to getting elected: Smith and Hoffman (if he isn’t elected tomorrow).  
 
The list passes the smell test. Mariano Rivera shows up as the best closer…by a good distance. Behind him are Hoyt Wilhelm and Goose Gossage.
 
You know why that’s a good top-3? Because it identifies relief pitchers from completely different eras….three pitchers who were used in differently capacities. So our junk metric doesn’t seem to be a bias towards older relief pitchers, or modern ones.
 
This metric also gives us a satisfyingdemarcation: 10. If you cross over 10 by this metric, you are a serious candidate for the Hall of Fame. Seven relief pitchers are over 10.0…three of those seven are already in the Hall (Goose, Fingers, Wilhelm) and one more is a shoo-in when he gets on the ballot. Trevor Hoffman is getting a lot of love this year, and Lee Smith did really well by the voters. The only guy we’ve ‘missed’ is Lindy McDaniel.
 
And the guys behind those guy…the guys in the 9.0-9.9 range…are candidates to consider. Sutter is in…he’s not a great choice by this metric, but he’s not so bad. Billy Wagner is a very good candidate, though not as good as Lee Smith. Eckersley shows up….he gets in on the extra years as a fine starting pitcher.
 
What’s startling is that the list suggests that the BBWAA voters have actually gotten closers right, a statement I never thought I’d ever put to paper. As strangely quixotic as their selections seem, they’ve mostly picked the best relief pitchers…the relief pitchers who have the most convincing blend of peak performance and longevity. There’s a good case that the relief pitchers with plaques in Cooperstown are the guys who most belong there.
 
*             *             *
 
The one pitcher who missing…the one guy I was surprised didn’t make either list…is Dan Quisenberry. Dan Quisenberry was one of the great relief aces of the 1980’s, so how come he’s not on our list?
 
Simple: FanGraph’s version of WAR gives a lot of weight to strikeouts, and Quiz didn’t get too many strikeouts.
 
FanGraphs credits Quiz with 14.5 WAR, which puts his WAR/200 at 2.8. That's not quite in the top-ten for relief pitchers. I think he ranked 14th or 16th, but I'm too lazy to check that. 
 
What if we substitute in his Baseball-Reference WAR, instead? Baseball-Reference credits him with a WAR of 24.9.
 
That jacks his WAR/200 all the way up to 5.2. His career value is 4.8. So we get:
 
Name
IP/200
WAR/200
RP Score
Mariano Rivera
6.2
6.4
12.5
Goose Gossage
7.9
3.7
11.5
Hoyt Wilhelm
9.2
2.1
11.3
Lindy McDaniel
9.0
2.2
11.1
Rollie Fingers
7.8
3.3
11.1
Lee Smith
6.3
4.1
10.4
Trevor Hoffman
5.4
4.8
10.2
Dan Quisenberry
5.2
4.8
10.0
Billy Wagner
4.5
5.4
9.9
Doug Jones
5.5
4.0
9.5
 
 
If you use the version of WAR that is less dependent on strikeouts, Quiz becomes a very good candidate for the Hall of Fame.  
 
So that's what our list gets us: the relief pitchers already in the Hall, and the relief pitchers who have the best cases to make the Hall. It's a junk stat, but it's also true, somehow. It's a True Junk Stat. 
 
 
*             *             *
 
We can use it, too, to consider modern relievers
 
Name
WAR/200
IP/200
RP Score
Craig Kimbrel
7.4
1.7
9.2
Jonathan Papelbon
5.5
3.4
8.9
Aroldis Chapman
7.1
1.6
8.7
Francisco Rodriguez
3.7
4.5
8.2
Kenley Jansen
6.2
1.7
7.9
Koji Uehara
5.6
1.6
7.2
Sean Doolittle
6.1
1.0
7.1
David Robertson
4.8
2.3
7.1
Dellin Betances
6.1
0.9
7.0
Ken Giles
6.4
0.6
7.0
Wade Davis
5.7
1.1
6.8
Jonathan Broxton
3.6
3.0
6.6
Huston Street
3.4
3.3
6.6
Joakim Soria
4.3
2.3
6.6
Joaquin Benoit
2.9
3.5
6.4
Sergio Romo
4.3
2.0
6.3
Ryan Madson
3.2
3.0
6.2
Jake McGee
4.8
1.3
6.1
Trevor Rosenthal
4.8
1.2
6.0
 
Craig Kimbrel, at this point, is amassing a tremendous WAR/200 innings pitched. So is Chapman. The challenge, for each of them, is whether or not they can healthy long enough to get their innings pitched up enough to be a serious candidate for the Hall of Fame. Jonathan Papelbon hasn’t been quite as effective as Chapman or Kimbral, but his career length is twice theirs, so he’s building a case. K-Rod’s case is more longevity than peak performance, and I don’t see him as a particularly strong candidate.
 
*             *             *
 
This is a junk stat. But as non-scientific as it is, I like it. It seems to work, better than anything else I've found.
 
 I’ve been on the fence about Lee Smith and Trevor Hoffman, but I think they’re deserving candidates now. I’m sold on Quiz. I think Billy Wagner has a good case. I’m going to have to look into this Lindy McDaniels fellow.
 
We’ll find out tomorrow if the BBWAA voters continue their surprisingly good sense when it comes to relief pitchers, and elect Trevor Hoffman into the Hall of Fame. And we’ll see if the BJOL readers have come to any conclusions about Hoffman, Lee Smith, and Billy Wagner.
 
Dave Fleming is a writer living in Wellington, New Zealand. He welcomes comments, questions, and suggestions here and at dfleming1986@yahoo.com. 
 
 

COMMENTS (31 Comments, most recent shown first)

KaiserD2
Thanks for the info. I had never noticed that on retrosheet.


7:12 PM Jan 8th
 
MarisFan61
Thanks!
And let's see if those data match what I said......

Would you believe......yes!
(Although I was as sure as I can be that I had it right, I have to admit I didn't expect the data to match exactly.)
3:52 PM Jan 8th
 
those
It is readily available through Retrosheet. Just go to any date on the schedule and the standings will give you each team's total runs and runs allowed through that day.

Here's September 13, 2000, for example:

http://www.retrosheet.org/boxesetc/2000/09132000.htm
3:44 PM Jan 8th
 
MarisFan61
One additional thing about that Yankee season: If it's true that in those last 18 games, the reason they did so world-class terribly, after having the division title essentially wrapped up, was that they 'let up' and were coasting toward the post-season.......

That's sort of a macro version of "playing to the scoreboard" -- which this field mostly says doesn't exist.

I think there's little doubt that it exists.
The field just hasn't found a way to demonstrate it. It has only found ways that seem to refute it -- emphasis on seem to.

Why would I doubt the seeming refutation?
Because to me the idea that it doesn't exist is an absurdity.
2:50 PM Jan 8th
 
MarisFan61
It isn't that readily available. You need to do some work -- but it took me about a minute and a half, including the time for getting the right webpage.

What I did was to get the page on baseball reference.com for the Yanks' "Schedule and Results" (basically the game log), and since I was only looking at the specific thing you had raised (which was how many runs the Yanks lost on the run differential during that period at the end of the season (well......you said "in a month" but remember, I was talking actually about an even briefer period, and BTW I made a little mistake on the total number of games; I said it was the last 17 games, actually it was the last 18, and the team's record was 3-15, not 3-14.... so those last 18 games were what I checked), what I did was to total up (mentally) what the run differential was for those last 18 games.

In case you want to check my 1-minute mental math (which isn't infallible) :-) or just want to take a look at that page anyway, here's a link for it:
www.baseball-reference.com/teams/NYY/2000-schedule-scores.shtml

BTW I just double-checked on the arithmetic, and it looks like I had it right: 89 runs, in just the last 18 games.
2:44 PM Jan 8th
 
KaiserD2
Yes, it was implicit. May I ask where that data is readily available?
2:17 PM Jan 8th
 
MarisFan61
You know what they say about "assume"..... :-)

" I suppose the Yankees could have lost 70 runs of their differential in a month but that's an awful lot."

You're right, it is an awful lot -- and it happened.
They lost 89.

Anyway, wasn't the info I posted before enough to tell you that the run differential at that time would have been top-level?
It was absolutely implicit.
12:21 PM Jan 8th
 
KaiserD2
p.s. I just figured my stats on Trevor Hoffmann. He topped 3 WAA just once, 3.1 in 1998 in 73 innings. He topped 2 WAA only twice. If you compare that to the stats I listed for Rivera, Sutter, Gossage and others below, it isn't very impressive. the most innings he ever pitched in a season was 88 and he was over 70 just three times.
11:45 AM Jan 8th
 
KaiserD2
p.s. I meant to say, regarding the strong hitting in pitcher's parks--everyone has always had an idea of what an acceptable level of hitting is. The point is, that same level was always much better, in reality, for a Yankee than a Red Sox. So a hitter that looks good enough (based on his average) to play in Boston, wouldn't look good enough as a Yankee, and they would find some one better. A great many below average hitters have hit .280 in Fenway.
11:38 AM Jan 8th
 
KaiserD2
Dear MarisFan61,

Don't worry about it, it happens.

I don't know where one can go to find what run differentials were on September 1, 2000, for instance If you do please let me know. I suppose the Yankees could have lost 70 runs of their differential in a month but that's an awful lot.

Regarding ballparks. I think Bill wrote in an Abstract that teams in pitcher's parks tend to win more--not in postseason specifically, but in general. And that's true--Yankees in old Yankee Stadium, Orioles in Memorial Stadium, Oakland As. But the reason, I've found, is really interesting. It's because they have better hitting. Every Yankee dynasty until the late 1990s was hitting heavy, pitching light. So were the Weaver Oriole teams and the Oakland 1970s dynasty. (Not the Oakland 2000s one though.) I mean, the pitching on those teams was decent, but it was not the key to their success. We all talk about the great pitching staffs those teams had but it's not too hard to win 20 games when your lineup scores 1-200 extra runs for you.

On the other hand, when teams like the Red Sox or Cubs are successful the reason tends to be their pitching. In the late 1970s the Red Sox had much better pitching than the Yankees but a weaker lineup. The ballparks made it look like the opposite. The same was true for the Red Sox and Yankees in 1949-50--Boston pitching was better, Yankee hitting was better. And an amazing example is the `1969 Cubs, who had significantly better pitching than the Mets. (The Mets had tremendous fielding that year.)

Whether "wearing down" is key I don't know. I do think big hitting teams, or teams in hitter's parks (whose hitting is almost never as good as it looks), do much worse in September when the weather cools down. That has been documented, I'm pretty sure. It could be that "wearing down" is just regression to the mean. We've all noticed that the leading hitter on August 1 is likely to have a higher average than the leading hitter at the end of the season.
11:36 AM Jan 8th
 
MarisFan61
P.S. to Kaiser: Sorry, hysteria got the better of me in that reply to you.
But I'm not sure if it'll be much better without the hysteria:

"It is hard to believe that the Yankees stood out with just three weeks left in the 2000 season"??

The thing is, they did, in the manner I indicated, and I gave enough detail for you to see that it was so. Like, what I said about the Pythgorean projection, which of course is just a function of runs-scored and runs-allowed.

Either you didn't fully read what I said (which of course would be forgivable), or you didn't grasp it. But it was right there.
10:15 PM Jan 7th
 
MarisFan61
P.S. Off the subject, but.....revisiting that old "doubles" thing, I wonder if something else was involved. And it wouldn't be too hard to check a little on this, at least to start getting an idea if this might have been so *:

I'm guessing, first of all, that teams with more doubles tend to be teams whose home parks are hitter's parks.

And -- I think I also got this from a thing by Bill in one of the old Abstracts; wherever I got it, I think it makes sense (and I hope I'm not misrepresenting it -- please, anybody say so if you think I am):
At least as of when this thing was written, the kind of home park that was most conducive to post-season success was slightly-pitcher's parks.

If that's all true, it's understandable why higher totals of doubles would be negatively correlated with post-season success. I don't mean to be saying this was necessarily a large portion of it, just that it would probably be in there.

About that old thing that I read about it being advantageous to have a home park that slightly favors pitchers, I don't remember if the article talked about why this might be so, but I can easily imagine one reason why (maybe because I read it there): :-)
Pitchers in a hitter's park would tend to wear down more as the season goes on. Pitchers in a pitcher's park would tend to wear down less, and to have more left in October.

Of course it's the other way around for position players. But I can imagine that this factor would loom larger for pitchers -- and I'd guess it would.

I remember another article -- this one for sure by Bill, in an Abstract -- about the Cubs' frequent late-season meltdowns, saying (if I remember right) how it might have been related to playing all those day games, which meant warmer weather -- and that it may have caused the players to wear down more. Assuming it was correct that "wearing down more" was involved, I would wonder if it wasn't just due to the day games and warmer weather, but also to this other factor I'm talking about here: being in a hitter's park, and the effect of that on the pitchers. I can well imagine that both of those factors worked together.

BTW, an additional issue here is, why slightly-pitcher's park? Why not just pitcher's park, period? I also can imagine why this would be so (and I think I did get this from that article, rather than coming up with it myself) but I won't take up space with it here.

* The way I'd start checking on it would be to see, first of all, if teams' doubles totals are highly correlated with Park Factor, which I think would be an easy part of this; I can't imagine that they wouldn't be correlated. And then, to look through some results of post-season series, to see if Park Factor did seem to correlate with winning, in the stated manner.
8:48 PM Jan 7th
 
MarisFan61
Kaiser:You completely missed it. There's nothing that's hard to see; I explained exactly what it was. If you want to see it, all you need to do is understand what I said -- and I was about as specific as one could be.
8:29 PM Jan 7th
 
KaiserD2
To MarisFan:

It is hard to believe that the Yankees stood out with just three weeks left in the 2000 season. They finished the season with a park-adjusted run differential of +58; Cleveland, Seattle, Oakland and the White Sox were all over +130. Over their season their lineup, by my calculations, was perfectly average, and their pitching staff was six games better than average. They were in fact six games over .500.

It is true, of course, that teams like the 1973 or 2015 Mets can be significantly better in October than their record indicated, either because of injuries or late-season acquisitions. But in general I think run differential is the best measure of a team's overall quality. The Blue Jays were 100 runs better than anyone else in baseball in 2015 but couldn't make it to the World Series.

I do remember Bill's article about his post-season prediction system. Doubles were a negative indicator because teams that hit lots of doubles took lots of chances on the bases, he hypothesized, and those didn't pay off in the post season. Home runs and power pitchers were advantageous as well, because (I would say) in cold weather it's very hard for long-sequence offenses to score. But Bill didn't ask whether his system was more accurate than simply using run differential would have been.

It is at least theoretically possible that one could design a team to do well in post season, and if there is a specific argument to that effect I'd be interested to see it. But nothing is foolproof. The 2011 Phillies had one of the great pitching staffs of all time, including Halladay, Cliff Lee, and Hamels. They won 102 games and lost in the first round of the playoffs. The reason we all watch baseball is that the outcome of an individual game--or 3 out of 5 or 4 out of 7 of them--is almost impossible to predict.
8:19 PM Jan 7th
 
steve161
Maris, if somehow you succeed in winnowing the arrogance out of sabermetric discourse, you should win the Nobel Prize for Peace.

Don't hold your breath.
1:49 PM Jan 7th
 
MarisFan61
But how do you decide which team is "the best" of the season? How did Pete do it? Are you sure he had it right? Are you sure there's a clear way to have it right?

I don't have to just be rhetorical. How about these things:

-- Suppose a team either makes major changes in mid-season that improve it or just gels for whatever reason; the team does far better in the latter part of the season with a better record than any other team while the overall record wasn't. Is some other team "better"?

-- An example that looks superficially like the opposite of the above: In 2000, the Yankees didn't have one of the best overall records. They had the worst record of any division winner, far worse than 4 of the other 5, and I'd guess they'd look like the worst on any overall metric analysis. But: 3 weeks before the end of the season they did have one of the best overall records, not 'the' best but in an essential clump of 5 teams with very similar outstanding records-- going by run differentials they were even a little better than that -- and they had the division wrapped up although not clinched. Their record the rest of the way was terrible: 3-14. As a fan, I wasn't particularly worried (but yes, a little), because it seemed like they were coasting toward the post-season; the season was essentially over in early September. After winning the division series with (indeed) some luck, they won everything. I imagine that in an analysis like what Pete did, they'd be shown as not even among the several "best teams," but are you sure some other team was "better"?

-- There's a principle that I think in itself undoes your point by a fair amount: What if some kinds of team aspects are more important in the post-season than in the regular season? And it seems that in fact they are.

We've seen more and more work on this in recent years. I'd say that Bill kicked it off (quietly) in one of the early Abstracts, by looking at what kinds of statistical factors correlate with winning in the post-season (the only factor I remember being the odd one of doubles being negatively correlated). There's been some work on Reader Posts suggesting that strikeout tendencies are more important for the post-season than for the regular season. Bill has talked about top-line pitching being more important in the post-season. Perhaps most famously -- perhaps the main underlying factor about the team in the above example -- a top-level closer is thought by many to be more important in the post-season.

So, this is yet another instance where it seems far too pat, sort of arrogant, to view something like the cited work of Pete as "showing" what was stated. It shows something; if I'm correctly imagining the essential kind of thing he did, it shows how certain whole-season factors are correlated with winning the World Series, and it shows that they aren't particularly. That's not the same as showing how often the best team wins.
10:46 AM Jan 7th
 
KaiserD2
In response:

The post season is basically a crap shoot, that is, luck is by far the most important single factor in determining the result of three rounds of playoffs. Pete Palmer has shown that since 1995 the best team in baseball over the season wins the World Series less than 25% of the time. Yes, the Kansas City Royals have won the last World Series and nearly won the one before that, but without ever posting an impressive record over the course of a season.

Unfortunately, I have not figured out how to prove this mathematically, but I believe (and some other knowledgeable students agree with me) that the whole "leveraged situation" argument is bogus. Runs are runs, and runs scored or given up in the first inning count just as much as runs scored or given up in the ninth. In fact, they count more, because in reality there are more of them. The rest of the team has to create the situation to allow the relievers to save games.

To be continued.
8:17 AM Jan 7th
 
Gfletch
KaiserD2, not saying you are right or wrong about the actual value of a relief pitcher, but this is an old, ongoing, and unresolved argument. Yes, not enough innings pitched to have anything close to the value of a great starting pitcher, or a great every day position player. But then we have the whole “leveraged situation” counter-argument.

My feeling about relief aces in the HOF…there are a handful of guys who were so impressive that you knew the game was over if they came into the game after 7 innings with the lead. Gossage was like that. Henke was like that. Today I think Wade Davis is like that. Betraying my age, guys like Lee Smith or Jeff Reardon…good players, but just not in that class. Bruce Sutter was like that for parts of some seasons, but never had a multi-season run like, say, Gossage. The Goose, to me, is the quintessential closer.

Of course, Rivera. Not as impressive to watch, but his record can’t be denied.

12:40 PM Jan 5th
 
OldBackstop
Shin makes a good point....we say "closer" but a lot of these guys did stints as setup or long men, or came in to the occasional mop up.

I wonder how the numbers look restricted to Save Opportunities?​
11:53 AM Jan 5th
 
MarisFan61
Kaiser: This article doesn't "show" it; it suggests it, and supports it very well. I'm skeptical that your work will be able to "point out" what you say it will, but it might well also suggest and support it.

Look: Hasn't sabermetrics had an evolution in how it sees differential values of different innings and situations, and accordingly, in how it sees the values of closers? And, what gives you such confidence that the evolution is done?

And after all, hasn't sabermetrics, at least public sabermetrics, been behind the curve on relief pitching? The advent of 'multi-headed monsters,' the Royals' success with it, the imitation being sought by other teams.....was any of this foreseen by (public) sabermetrics? I think not; it seems to have taken the field by surprise and left it a little slack-jawed. BTW I keep specifying "public" sabermetrics because we (most of us) don't know what goes on privately to front offices, and I wouldn't at all be surprised if we eventually learn that the Royals (for example) had the benefit of private sabermetric research suggesting that relief pitching, cultivated and used in a certain way, could have value beyond what was suggested by 'state-of-the-art' metrics.
11:52 AM Jan 5th
 
OldBackstop
HeyDave, where you been, dude?

Nice read as always.

I saw a panel on the MLB channel talking about closers and the HoF. It was mostly centered around postseason performance, which I think has some validity. Usually to me the argument is it is "extra credit" because an individual can't really be blamed if the team never gets him postseason opportunities.

But to me the postseason is what it is all about with closers. Unlike position players, an inability to get it done in the postseason just might lose you your job.

Billy Wagner, for one example, was a train wreck, giving up 21 hits and 13 earned runs in 11.2 innings over 14 appearances. Mo, of course, was money, 8-1 with 42 saves and a .70 ERA.

So I think whereas the postseason is sometimes merely a bonus for a career, I think for closers it should be given considerable weight. If, for instance, Mo and Wagner had identical regular season numbers somewhere in between their actual ones, I would say Mo was a shoot-on to go right in, and Billy not so much.


11:48 AM Jan 5th
 
shinsplint
Dave, engaging article as usual, thanks. Your stat inadvertantly, and I think justifiably, gives a leg up to relievers from the era when they weren't necessarily 'closers'. It favors peak over career (relative to fWAR) only when the WAR/200 for the player who played longer gets above a certain point relative to the pitcher who pitched fewer innings. When the WAR/200 for the pitcher who plays longer goes below that it values career over peak, more so depending on how low the WAR/200 is for the player who played longer.

So, we see Wilhelm and McDaniel moving way up relative to Rivera and Hoffman for example, closing the gap much more than even their career fWAR does. As it turns out, I think that's serendipitous because Wilhelm and McDaniel were not used in such a way that they would be in a position to have a super-low WAR/200 because they had to pitch more innings per start, unlike later closers who can concentrate on throwing gas for one inning and thus have a higher WAR/200. So your stat turns out to be a good compensation for that disadvantage.
9:24 AM Jan 5th
 
KaiserD2
34 years after the birth of modern Sabermetrics (with the publication of Bill's first commercially published Baseball Abstract), we are still largely the prisoners of popular prejudice, especially when it comes to the Hall of Fame. The book I am in the middle of will point this out but there's no way it will do anything about it.

We decided, at roughly that same time, that "closers" were important. That meant that a really good one was a huge asset and belonged in the Hall of Fame. And a few of them have been elected, and there is pressure to elect more.

Yet I would suggest that, as this article shows when it turns to WAR, it is simply impossible for anyone who pitches 70-80 innings a year to consistently have the impact on his team's fortunes as a great hitter or starting pitcher. My book is based on WAA, not WAR. It defines a superstar season as one of at least 4 WAA. I have not got a full count in front of me, but I am pretty certain that there have not been 10 relief pitchers, much less closers, who have had 4 WAA in a season in the entire history of baseball. Since 70-80 innings is equal to about 8 or 9 games, it's easy to see why. You would have to give up so few runs as to be virtually certain of going 8-0 or 9-0 to do so.

Now I do have a spreadsheet showing WAA totals for a number of famous relievers. (But not all by any means.) Bruce Sutter topped 4 WAA once (in 1977 when he pitched 107 innings) and nearly did it again in 1984 in 122 innings. Rollie Fingers topped 3 only once. and that was in 1981--he probably would have dropped below 3 without the strike.
Mike Marshall was over 3 only once. Hoyt Wilhelm topped 3 as a reliever just once, in 145 innings in 1965. (In his one year as a starter, 1959, he had 6.2 WAA in 226 innings. That is a fantastic season. Why he was promptly sent back to the bullpen I have no idea.) Gossage had 4.7 WAA with the Pirates in 133 innings in 1977 and 3 for the Yankees in 1982 in about 90 innings.

Now Rivera is indeed incomparable. He too topped 4 WAA only once, with 4.4 in 1996, when he pitched a career high 107.2 innings. But he had four seasons over 3 WAA and seven more seasons over 2 WAA. No one else comes close to those figures. For a starting pitcher, that is not an overwhelming record for the Hall of Fame, but there are starting pitchers with lines like that in the Hall of Fame.

I personally think that one of the best tests for the Hall is one of Bill's questions in the "Keltner test": if this guy were the best player on your team, could you win the pennant? I have discovered that there were at least two occasions in which the answer, for a reliever, was yes. One was Rivera in 1996--no other Yankee had more than 4 WAA in that year. The other, amazingly, was Willie Hernandez in 1984, who had 4.5 WAA for the Tigers in 140 innings or so. He was elected the league MVP and he was in fact the team MVP, barely beating out Chet Lemon (who had a great year in the field.) He was far from the best player in the AL but he was the best player on the Tigers. But I do not believe that being good enough to be the MVP on a pennant winner just once should be enough to get anyone in the Hall of Fame.

When you rely on stats like saves, and when you rely on career totals for anything, to determine who belongs in the Hall, you are not really looking at what this player did to help his team win games and pennants. Mariano Rivera is probably the only reliever in history whose contribution to winning games and pennants could be compared to some of the hitters and starting pitchers in the Hall of Fame, and even his case is not overwhelming.
8:52 AM Jan 5th
 
MarisFan61
singular, "McDaniel"
He wasn't conjoined twins. :-)

(Although, he did have a brother who also pitched in the majors, who had an interesting but much less happy history himself.)​
2:47 AM Jan 5th
 
DaveFleming
Just a note on Wilhelm and McDaniels....while FanGraphs WAR suggests that they were similiar players (2.2 WAR/200 for Wilhelm, 2.1 for McDaniels), Baseball-Reference says that they're a bit more far apart:

Wilhelm: 4.5 WAR/200 IP
McDaniels: 2.7 WAR/200 IP

One of the problems with FanGraphs WAR for relief pitchers is that it does seem to show a bias towards more modern players/RP's with high strikeout rates. Comparing McDaniels to Wilhelm is probably a bit of a stretch.
12:44 AM Jan 5th
 
DaveFleming
Yeah, I think Henke must've fallen off when I cut the list to RP's with 800 IP's....that's got to be what happened, because he shows up at 9.2 on my list.

I did the spreadsheet once, got the formatting messed up, and did it again. I think the second time I must've used 800 IP as a cut-off for HOF candidates.

I loved Henke. The 80's had some fun closers. My kid was running around with a Dave Smith card the other day.....I wonder if he still holds the 'Stros team record for saves.
12:24 AM Jan 5th
 
DMBBHF
Hi Dave,

That was a nice read. I, too, am drawn towards measurements that attempt to combine peak and longevity. If you blend it right, the cream does tend to rise to the top.

And, I agree, it seems like a reasonable ranking. I was surprised to see Garber rank so high, but then again, very few pitched more career relief innings than he did.

Quick question: Shouldn't Henke be in the top 15 of your first list? I followed your steps, and have him at 9.2. Did you set a minimum inning requirement (say, 800 relief innings)? If so, that would explain his absence, as he was a little bit short of that threshold.

You also got me to thinking....what if we took your results and did a Power-Speed type of number, using a harmonized mean approach, so that it rewarded the balance or blend of the two, so that you had to do pretty well on both scales to score well? I tried that, and then multiplied the result by 2 to put it back on a similar scale as yours (although the numbers tend to be a little lower doing it this way):

Name - Harm Mean (then x 2)
Mariano Rivera - 12.5
Trevor Hoffman - 10.2
Rich Gossage - 10.0
Lee Smith - 9.9
Billy Wagner - 9.8
Rollie Fingers - 9.3
Doug Jones - 9.2
Tom Henke - 9.0
Dennis Eckersley - 9.0
Joe Nathan - 8.6
Bruce Sutter - 8.6
Jonathan Papelbon - 8.4
Robb Nen - 8.3
Francisco Rodriguez - 8.1
Dan Plesac - 7.9

Lots of the same names towards the top on both lists, but this one puts Henke, Nathan, Papelbon, Nen, K-Rod and Plesac in the top 15 instead of Wilhelm, McDaniel, Garber, Tekulve, Lyle, and McGraw, who tended to be really strong on the IP scale.

I like both lists....subjectively, I like seeing Wilhelm on your list because I think he deserves to be mentioned among the top relievers, and your method recognizes that. It was just another "fun" way to look at it.

Thanks!
Dan
10:49 PM Jan 4th
 
MarisFan61
I sent a link for this article to Lindy's 'official website.'
(And BTW here's a link to that site:
www.lindymcdaniel.com/Index.html)

Lindy is still with us, and writes essays on the site. I'm hoping he himself will be able to take a look at this article.

As I mentioned in the note that I sent to his website, I had the pleasure of meeting him during his time with the Yankees, and was able to experience first-hand what a gracious gentleman he is.
10:25 PM Jan 4th
 
evanecurb
McDaniel had a long career. He pitched over 2,000 innings, or 137 per 162 game season. He did have 74 starts, so in order to get an accurate count, you'd need to take that out of your relief pitcher data (which it appears you've already done). He was really good for a really long time, but was not particularly dominant. He was probably never the best relief pitcher in baseball at any point in his career, but he was probably in the top ten in over half of his seasons.
10:16 PM Jan 4th
 
DaveFleming
Mike Marshall did pretty good (8.5). If you sub in his BB-Ref WAR he jumps to a 9.1 score overall. His career was longer than Sutter's (by IP), and longer than most of the modern guys.
9:40 PM Jan 4th
 
evanecurb
Interesting. You've taken an approach that attempts to drive out the distractions of ERA and Saves, which are more context dependent than WAR. And you're correct: it passes the smell test. One suggestion: what would happen to the list if you used reverse hitting stats, such as batting average against, slugging average against, adjusted for context?
Oh, and Lindy McDaniel was better than most people remember. He pitched for a long time, and very effectively. Great forkball. The Dwight Evans of relief pitchers, maybe?

Surprised that Ron Perranoski and Mike Marshall don't make the list, but they were both guys with high peaks and not so long careers, so it makes sense.
9:30 PM Jan 4th
 
 
©2019 Be Jolly, Inc. All Rights Reserved.|Web site design and development by Americaneagle.com|Terms & Conditions|Privacy Policy