The All Important Closer

February 3, 2017

The All Important Closer

 2017-8

              It is a staple of October wisdom that you need to have a great closer to win the World Championship.   I have long been skeptical of this notion.   Skeptical is on the up side; I may have been dismissive.   I may said and written things like "of course you need to have a good closer in the World Series, but you need to have a good second baseman or a good right fielder, too.   One is not more important than the other."

              But hearing this wisdom again during the 2016 World Series, I suddenly saw a way to check it out.   Suppose that we rank all the second basemen in the majors, 1 through 30.   How many World Championship teams have a top 10 second baseman?   A top 5 second baseman?   

              In my youth, before Closers were invented, the conventional wisdom was that Championship teams had to be strong up the middle. . .catcher, second base, shortstop, third base.    You still hear that sometimes, not as often as we used to hear it.   But this research addresses THAT issue as much as it does the Closer issue.  

              OK, here is what I did.   I ranked all players from the last 40 years (1977-2016) in two different ways, by Season Score and Win Shares, and formed a compromise of the two; not going to get into the details.   I started in 1977 because:

              1)  I wanted as many World Championship teams as I could get,

              2)  You really can’t go back to 1967; the bullpens are just TOO different for teams from 1967 to be relevant, and

              3)  I sort of wanted to dodge the 1976 Reds, since they are such an outlier team.  

         &nbs​p;    But after I started the work I realized that there is no World Champion in 1994, so I only had 39 teams.   I wanted 40 teams, so. . .OK; I’ll include the 1976 Reds.  

           &nb​sp;  But the BIGGEST reason I started in 1977 is that it’s an expansion year.    The critical question that I am asking is, "How many World Championship teams actually do have a Top Ten closer?   Is that number actually higher than the number of teams that have a Top Ten second baseman?"

              Well. . . .Top 10, you see.   Top 10 out of 30 is a whole lot different from Top 10 out of 16.   If we go back to 1967, we would have 50 years of data rather than 40, which is better, but then you’re choosing the Top 10 out of 20 teams, which is quite a lot different from 10 out of 30.   By 1977 there are 26 teams; 10 out of 26 is enough like 10 out of 30 that I’m not going to worry about it, but 10 out of 20 is a prohibitive problem.   By including 1976 (to get 40 World Champion teams) we’re including one season of only 24 teams, which I guess is OK. 

              Anyway, to my great surprise. . .

 

              The proposition that to win a World Championship you need a great closer, and that a great closer is more important than a great player at the other positions, appears to be true.   According to this study, anyway, it’s true.   Which is a shock to me; I didn’t expect to find that.  

              OK, let me detail the method a little bit.   I ranked the players at each position based on their regular season performance, more or less as you would rank them if you were voting for the Most Valuable Player or a post-season all-star team.   Playing time counts; a player who hits .280 in 300 at bats is not equal to a player who hits .280 in 550 at bats.   A closer who has a 1.80 ERA in 45 games and 27 saves is not equal to a closer who has a 1.80 ERA in 70 games and 42 saves.   

              If the team has the best player in the majors at a position, that’s "1"; if they have the second-best, that’s "2", etc.    You can’t score worse than 30; if the Championship team does not have one of the 30 best players in the majors at a position—which does happen sometimes, you’d be surprised how often—they are listed at "30".     I did not look at the aggregate totals for the position; in other words, it might be that a team used four first baseman but the four first basemen combined hit .318 with 34 homers; I wouldn’t know.   I didn’t look at that.   If a team. . .let’s say, the 1979 Pirates; you can find some sources in which Phil Garner is not listed as a regular on the 1979 Pirates, although he was third on the team in games played and at bats, because one player played more at second base and another played more at third base.   I didn’t do that; that would be silly for this purpose.    A player like Garner has to be considered the regular SOMEWHERE.   

              If a team picks up a player at the trade deadline or even later, if he’s on the roster at the end of the season, he counts.   The 1993 Toronto Blue Jays started the season with Darrin Jackson as a regular outfielder and Dick Schofield Jr. as their regular shortstop, but they ended the season with Rickey Henderson in the outfield and Tony Fernandez as their regular shortstop, so I credit them with Henderson and Fernandez (who were quite a bit better than Jackson and Schofield.)  I didn’t look at career performance norms.   I think one year when the Yankees won, Mariano was like the fourth-best reliever in baseball or something.   Obviously you can argue that he was always the best; he just didn’t happen to have the best numbers that year.   That’s up to you.  If you want to study the same issue, feel free to make up your own rules however you like, but I just used the rules that made sense to me. 

              I evaluated the 40 World Championship teams at eleven positions—the eight field positions, #1 Right-Handed starter, #2 Left-handed Starter, and #1 reliever.   I didn’t do Designated Hitters, because, of course, a 1-to-30 comparison would not be meaningful, since only one league uses the Designated Hitter.  

              As I said, the Championship teams were more outstanding at Closer than at any other position—actually, quite a bit better.    Of the 40 World Championship teams, 31 (or 78%) had a top ten closer—four more than any other position.    20 of the 40 teams, 50%, had a Top-5 closer, which is as many as any position, and 11 of the 40 teams had a Top-3 closer, which, is the second-best of the positions, behind left-handed starter.    The average rank of a Closer on a World Championship team was 7.9 among the 30 teams, which is much higher than any other position.  

              Only two of the 40 World Championship teams were ranked as having the best Closer in the majors—the Tigers in 1984 (Willie Hernandez, the MVP) and the Royals in 1985 (Dan Quisenberry).  Four teams were listed as having the #2 closer in the majors—the 1976 Reds (Rawly Eastwick), the 1977 Yankees (Sparky Lyle), the 1978 Yankees (Goose Gossage), and the 1999 Yankees.   In 1977 Sparky Lyle, the American League Cy Young Award winner, ranked second behind Goose Gossage, then with the Pirates.   In 1978 Gossage, now with the Yankees, ranked second behind Kent Tekulve, now with the Pirates.   In 1979 the Pirates won, but Tekulve ranked third, behind Jim Kern, who had an all-time great season with the Rangers, and Senor Smoke, Aurelio Lopez.   In 1999 Mariano ranked second behind Billy Wagner. 

              Only 3 World Championship teams had Closers who ranked lower than 20th; that is, 31 teams in the Top 10, only 3 in the bottom 10.   Those three were the 1987 Twins (Jeff Reardon, with a 4.48 ERA), the 2006 Cardinals (Jason Isringhausen, walking 38 men in 58 innings), and the 2014 Giants, who used Madison Bumgarner as their Closer in the World Series because they didn’t really have one.  

              The second-strongest position for World Championship teams was #1 Left-Handed Starter.   While six teams really did not have a left-handed starter to speak of, and thus drew a "30" in this slot, six World Championship teams had the best left-handed starter in baseball—the 1978 Yankees (Guidry), the 1980 Phillies (Carlton), the 1996 Yankees (Andy Pettitte), the 2001 Diamondbacks (Big Unit), the 2009 Yankees (Sabathia) and the 2016 Cubs (Jon Lester).   Four other World Championship teams had the second-best Left-Handed starter in baseball, and you could probably argue that any of those four was really the best (Fernando Valenzuela, 1981; Scott McGregor, 1983, Frank Viola, 1987, and Bumgarner in 2014.)   

              The point is, though, that you can’t win a World Championship without a great closer; well, you CAN, but it doesn’t happen a lot.   In terms of Top 10-Bottom 10, the Closers are 31-3; the Left-handed Starters are 27-7, and they’re in second place.   The 1984 Tigers, 1989 Athletics, 1991 Twins, 2004 Red Sox, 2006 Cardinals and 2007 Red Sox really had no left-handed starter most of the year.  (The 1991 Twins did have one, Allan Anderson, but he was completely ineffective.)   Jon Lester joined the 2007 Red Sox late in the season, recovering from cancer, and made one start in the post-season, which was the final game of the World Series, but made only 11 starts during the season.

              Of course, the pitcher’s record is inextricably linked with the performance of his team, and one can never be TOO certain that you have the ratings right.  Can I give any weight to the won-lost record, or do I have to throw that out?   The problem is that if you throw that out, the things that remain are still colored by the performance of the team.  

              The other problem is that, of course, a team doesn’t have two starters; they have five or four or seven or eight; we COULD rank teams by their #1 starter, #2, #3, etc.   That doesn’t actually work; the pitchers form a line, so that the World Championship team would virtually always have a Top-10 3rd starter or 4th starter.   Every team has a #1 catcher, a #1 second baseman, a #1 reliever—and a #1 right-handed starting pitcher.   You can argue the point, but I think it is reasonable to parallel the rankings of starting pitchers, in this form, with the other positions.  

              These are the average ranks of the players starting for World Championship teams, by position:

              Closer                                  7.9

              Left-Handed Starter       10.2

              Catcher                            ​; 10.8

              Right-Handed Starter     11.1

              First Baseman                  11.1

              Center Fielder                  11.5

              Second Baseman            11.8

              Third Basemen                12.4

              Shortstop                         12.5

              Left Fielder                       12.7

              Right Fielder                     13.2

 

              The overall average rank of a regular player on a World Championship team is 11.4 out of 30. . .well, out of 29.2 or whatever it works out to.  

            &n​bsp; So is the proposition that Championship teams are strong up the middle true?   Well, maybe.   The average rank at catcher/second/short/center is 11.6, whereas the average rank at the corner position is 12.1.   I don’t know that I would want to base any conclusions on that.  

              Whenever you rank players you are walking on thin ice, but I have more confidence in the rankings for closers than for almost any other position.   I am confident of the rankings for first basemen, because first basemen are basically hitters and it is easy to recognize a good hitter, whereas for catchers and shortstops more of their job is defense, and fielding is difficult to measure reliably.   Relievers are fairly straightforward.   I have a lot of confidence that the rankings are justifiable, although of course you might prefer one guy over another. 

              The other key metric for me was the percentage of teams which had a Top 10 player at the position, and here the data is a little different:

              Closer                                78%

              Left-Handed Starter       68%

              First Base                          65%

              Second Base                    60%

              Right-Handed Starter     55%

              Center Field                     55%

              Catcher                             55%

              Shortstop                         53%

              Right Field                        53%

              Third Base                        48%

              Left Field                           48%

 

              Overall, 58% of the regulars on World Championship teams are Top 10 players at their position. 

              OK, three things to tie up here before we move on.  

              1)   What were the greatest teams of that time period, compared by this method?

              2)   Which teams had #1 players?

           ​   3)  What did I learn from doing this that surprised me?

 

              1)  The Greatest Teams.

           &​nbsp;  Analyzed in this way, the 1976 Reds are the greatest team of the last 40 years.   The Reds have not only top 10 players, but top FOUR players at every position except center field, right-handed starter, and left-handed starter.    Their center fielder, Cesar Geronimo, hit .307 in 1976 and was outstanding defensively, so he ranks 10th in that season, his best season, so the Reds rank in the Top 10 at 9 of the 11 positions.  They rank first at four positions, two of which are surprises.   Joe Morgan, National League MVP, ranks #1 at second, and George Foster, the 1977 MVP, led the majors in RBI in 1976, and ranks first in left field.    Hall of Famers Johnny Bench and Tony Perez don’t rank first, and Pete Rose doesn’t, but they rank highly, while non-Hall of Famers Dave Concepcion and Ken Griffey Sr. do rank first at their positions.   Concepcion hit .281 with 9 homers, 69 RBI.   The 69 RBI led all major league shortstops in 1976, and only one shortstop (Toby Harrah) was in double figures in homers; Concepcion was second there.  Ken Griffey in right field hit .336, stole 34 bases and scored 111 runs; with Reggie Jackson and Dave Parker having less than stellar seasons, Griffey ranks #1 at the position. 

              The ’76 Reds have an average position rank of 4.6, the lowest of any World Championship team in the last 40 years (OK, 41).   That’s not a surprise, but second on the list is, or might be.    The 1978 Yankees rank 8th at catcher (Munson), 8th at first base (Chambliss), 2nd at second base (Randolph), 3rd at third base (Nettles), 8th in left field (Lou Piniella), 6th in right field (Reggie), 8th at right-handed starter (Ed Figueroa), 1st at left-handed starter (Guidry), and 2nd at closer (Gossage).  

              The ’78 Yankees thus match the ’76 Reds (and a few other teams) in having Top 10 players at nine of the eleven positions, although their average rank is higher:

Year

Team

C

1B

2B

3B

SS

LF

CF

RF

RHS

LHS

Cl

Avg

1976

Reds

3

4

1

2

1

1

10

1

15

11

2

4.6

1978

Yankees

8

8

2

3

16

8

12

6

8

1

2

6.7

 

              Mickey Rivers, the center fielder on the ’78 Yankees, had a down year in ’78, although he was a good player and would have been top 10 in other seasons.   So the irony is that the Yankees’ weak spot in ’78 was shortstop Bucky F. Dent, who

              a) hit one of the most famous home runs in baseball history in that season, and

              b)  was also the World Series MVP in 1978, hitting .417 in the World Series and driving in 7 runs in the six-game series.  

              Bucky was a good player, but the question I am trying to get to is why we have largely overlooked the greatness of that team.    I think there are two reasons:

              1)   A third great team, coming on the heels of the great Oakland team (1972-1974) and the great Reds team (1975-76) just doesn’t fit in our minds.   The ’77-’78 team—although a legitimately great team—were in the shadow of the other two great teams of that decade.

              2)  The Yankees had to beat the Boston Red Sox—ALSO a really good team—just to reach the post-season.   The ’78 Red Sox were probably good enough to beat 80% of division-winning teams in that era.   The fact that the Yankees needed a 163rd game to win their division gave the impression that they were lucky to reach the post-season, whereas the reality was that they were pushed to the wire by another very strong team.  

              With an average position rank of 6.7, the ’78 Yankees tie with a probably more popular candidate for the position of Greatest Team Ever:  the 1998 Yankees.  

Year

Team

C

1B

2B

3B

SS

LF

CF

RF

RHS

LHS

Cl

Avg

1976

Reds

3

4

1

2

1

1

10

1

15

11

2

4.6

1978

Yankees

8

8

2

3

16

8

12

6

8

1

2

6.7

1998

Yankees

7

10

6

5

3

19

3

5

5

4

7

6.7

 

              The 1998 Yankees are the only team in the study who have Top-10 players at 10 of the 11 positions.   But what you will notice above is that, whereas the 1976 Reds have four #1s, the 1998 Yankees actually don’t have anyone on their team who ranks #1 or #2 at his position for that season.   Posada obviously was not on the level of Piazza or Pudge that season; Tino Martinez, hitting .281 with 28 homers, was nowhere near the level of Mark McGwire or Mo Vaughn.     Second base, Craig Biggio and Chuck Knoblauch are somewhat similar players, but Biggio out-hit Knoblauch that year by 60 points.   Third baseman Scott Brosius was a good player, but no Chipper Jones or Scott Rolen.   Jeter ranks third in that season behind A-Rod and Nomar, who both beat him by about 40 RBI.    Center fielder Bernie Williams was great, but not the equal of Ken Griffey Jr., who hit twice as many homers as Bernie and drove in 50 more runs.   Right fielder Paul O’Neill was terrific, but not near the level of the two 1998 Most Valuable Players, who were both right fielders.   The Yankees top right-handed starter (David Cone) was outstanding, but ranks below Kevin Brown, Roger Clemens, Greg Maddux and Pedro Martinez; their top lefty, David Wells, ranks below Tom Glavine, Randy Johnson and the Mets’ Al Leiter.   Closer Mariano Rivera (36 saves and a 1.98 ERA) cannot reasonably be rated ahead of Trevor Hoffman (53 Saves and a 1.48 ERA) or Robb Nen (40 Saves and a 1.52) ERA.  

           &n​bsp;  The argument you would have to make—and maybe can make, I don’t know—to get the ’98 Yankees up to the level of the ’76 Reds is starting pitching.   The ’98 Yankees certainly had better starting pitching than the ’76 Reds, and it is a reasonable argument that this accounting method does not give sufficient weight to that.  

              The 2009 Yankees have high-ranking players at catcher (Posada), first base (Teixeira), second base (Cano), third base (A-Rod), short (Jeter), left field (Damon), right field (Swisher), left-handed starter (Sabathia) and closer (Mariano); their only two weaknesses are center field (Melky Cabrera) and right-handed starter (A. J. Burnett and Joba Chamberlain), and those aren’t real weaknesses; they’re just not strengths.     1993 Blue Jays are fifth, 1979 Pirates sixth:

Year

Team

C

1B

2B

3B

SS

LF

CF

RF

RHS

LHS

Cl

Avg

1976

Reds

3

4

1

2

1

1

10

1

15

11

2

4.6

1998

Yankees

7

10

6

5

3

19

3

5

5

4

7

6.7

1978

Yankees

8

8

2

3

16

8

12

6

8

1

2

6.7

2009

Yankees

4

4

4

3

3

5

17

9

23

1

3

6.9

1993

Blue Jays

20

1

1

20

7

4

6

4

12

4

4

7.5

1979

Pirates

16

8

3

8

5

13

10

2

10

8

3

7.8

 

            ​;  The six WORST championship teams, by the same method, are these six:

Year

Team

C

1B

2B

3B

SS

LF

CF

RF

RHS

LHS

Cl

Avg

2014

Giants

1

30

28

12

9

13

26

7

30

2

24

16.5

2006

Cardinals

14

2

22

5

19

24

18

15

3

30

27

16.3

1988

Dodgers

25

30

7

29

30

2

11

13

1

9

11

15.3

2015

Royals

8

6

10

19

18

16

4

30

21

28

7

15.2

1997

Marlins

9

27

30

6

9

5

30

12

6

17

9

14.5

1987

Twins

24

7

20

8

9

28

4

14

15

2

24

14.1

 

              The 2014 Giants had two great players, Posey and Bumgarner; otherwise they are kind of a mess.    An average team would rank by this method at 15.5, so BY THIS METHOD, it appears that the 2014 Giants and the 2006 Cardinals were actually below-average teams; statement is not offered for the truth of it.  

 

#1 Players on Championship Teams

              The only championship teams in the last 41 years which had THE best catcher in baseball that season were the 2012 and 2014 Giants, with Buster Posey.   Two teams had the best first baseman in baseball, the 1983 Orioles (Eddie Murray) and the 1993 Blue Jays (John Olerud).   The 2006 Cardinals had Pujols, back when Pujols was Pujols, but the MVP that year was another first baseman.

              Second base, there are three teams in this time period which have won the World Series with the best second baseman in baseball, the 1976 Reds (Joe Morgan), the 1993 Blue Jays (Alomar) and the 2008 Phillies (Chase Utley).    Third base, the only two I have marked are the 1985 Royals (Brett) and the 2016 Cubs (Bryant).    In 1980 the two MVPs were third basemen, and they met in the World Series; I have Brett (who hit .390 and drove in more than a run a game) ranked ahead of Schmidt for that season, not for his career.   You can argue that one all night, and also it is not absolutely clear that Bryant is the #1 third baseman in 2016; you can go with Donaldson or Arrenado. 

              Shortstop, there are three World Championship teams with the best shortstop in the game that year, the 1976 Reds (Concepcion), the 1983 Orioles (Ripken) and the 1999 Yankees (Jeter).    The only championship team that had the majors’ best left fielder was the 1976 Reds, George Foster; left field shows as a very weak position in this survey.  

              No team in the last 41 years has won the World Series and also had the best center fielder in baseball that season.     The 1999 Yankees are the closest (Bernie Williams).    In right field, the 1976 Reds (Griffey) and the 1977 Yankees (Reggie) are the only two; Parker ranks second in 1979.     Four teams have had the best right-handed starter in baseball that season, the 1988 Dodgers (Orel Hershiser), the 1995 Braves (Greg Maddux), the 2001 Diamondbacks (Curt Schilling) and the 2004 Red Sox (Curt Schilling again.)    Schilling is remarkable in that he appears twice on the list, the only player who does, and he did it with different teams, and the 2001 Diamondbacks are remarkable in that they had both the best right-handed starting pitcher in baseball (Schilling) and the best left-hander (Randy Johnson).   Six teams have won the World Series with the best left-handed starter in baseball; I listed those earlier. 

              Teams with outstanding weaknesses are "30s" in this list; sometimes the player there would rank lower than 30th, but I mark it as 30.    Catcher, nobody; all teams have at least a semi-legitimate catcher.   First base, the weak ones are the 1988 Dodgers (Franklin Stubbs) and the 2014 Giants; Brandon Belt is a good player, but he had a very poor season.  Second base, the 1995 Braves, 1997 Marlins and 2011 Cardinals had holes at second base.    The 1992 Blue Jays had an issue at third base, Kelly Gruber hitting .229 with no positives on the side.  

             ​; Four almost consecutive teams in the 1980s had real problems at shortstop:  the 1985 Royals, the 1986 Mets, the 1988 Dodgers and the 1989 Athletics all had shortstop shortages.   Buddy Biancalana briefly became a celebrity after he played well in the 1985 series.    No team had a huge hole in left field.     The 1997 Marlins and the 2010 San Francisco Giants were weak in center field, while six teams—all from the 21st century—have won the World Series despite being weak in right field:   the 2001 Diamondbacks, 2004 Red Sox, 2010 and 2012 San Francisco Giants, 2015 Royals and 2016 Cubs (Wayward Heyward).   The 1996 Yankees, 2008 Phillies and 2014 Giants had no very impressive right-handed starting pitcher, and six teams (listed before) had no good left-handed starter.    No team has won the World Series without some kind of a closer, although, as I discussed earlier, there have been three cases where the closer was pretty weak. 

 

              3)  What did I learn from doing this that surprised me?

              The thing that surprised me (other than the basic conclusion, which surprised me) was the 2013 Red Sox.    At least three times I have been speaking to an audience somewhere, and an audience member has suggested that the 2013 Red Sox, while that was a thrilling season and a wonderful run, were not really as good as the other Red Sox World Series teams.    I agreed with them; I don’t know, but it sounds true.   We weren’t good the year before or the year after.

              This is not at all the way this system sees it.    By this method, the 2013 Red Sox rank as MUCH better than either the 2004 or 2007 Red Sox.   They rank as the 8th best among the 40 Championship teams.

          ​;    See it the way you want to see it, but this is why this happens.   First, the 2013 season was a pitcher’s year.   The major league OPS was down 40 points from some of the surrounding years, so the numbers have to be re-calibrated.    Second, the 2013 Red Sox had a bunch of players who may not have been great players, but who did have very good years in 2013:

              Jarrod Saltalamacchia has not had a great career, but he was a top ten catcher in 2013.   He hit 40 doubles in 121 games, had an .800+ OPS; his defense was good enough that you could live with it.

              Mike Napoli was struggling with his sleep disorder thing, and we never could get the great hitter that we thought was in there to come out, but he did drive in 92 runs with an .846 OPS.   There are not more than six major league first basemen who were better.

              Stephen Drew had his last good year at shortstop.   He was a Top 10 Player.

          &​nbsp;   Daniel Nava was a player with a long, long road to the majors and a fairly short career once he got there, but in 2013 he flipped between left field and right, and hit .303—the highest average of any major league left fielder.     It wasn’t an empty .303; his .385 On Base Percentage was the highest of any major league left fielder except Matt Holiday, and he had a little power.  

              Koji Uehara, our closer at the end of the season, was 38 years old then and is not a Hall of Famer, but he had a Hall of Fame season, pitching 73 games, 74 innings, with a strikeout/walk ratio of 101/9 and a 1.09 ERA.  

              Those were all Top 10 players, as were, of course, Dustin Pedroia and Jon Lester, and Jacoby Ellsbury was.     It’s actually a really solid team, based on how they played that season.    The 2013 Red Sox had a Top 10 player at nine out of 11 positions—the same as the 1976 Reds.  

              And our DH was pretty decent.  

             

  
 
 

COMMENTS (76 Comments, most recent shown first)

Guy123
So to sum up, we have an assurance (but no evidence) that the closer rankings are not biased by team strength. And it's unclear what evidence to the contrary, if any, is admissible. So I guess we will have to leave it there.....
7:30 AM Feb 14th
 
bjames
A lot of traditional forms of evidence appear to be out of bounds. So let's try another approach. Can you tell us how many Win Shares the average top 10 closer has? That must be easy to calculate given your methodology.



My apologies, but. . . .it would take me MUCH longer to figure that than it did to do the original study. Sorry.
5:58 PM Feb 13th
 
Guy123
A lot of traditional forms of evidence appear to be out of bounds. So let's try another approach. Can you tell us how many Win Shares the average top 10 closer has? That must be easy to calculate given your methodology. Then we will know how many more wins we can expect these teams to have just by virtue of their closers' performance. If it turns out the teams win many more games than that, then we know that either A) the closer rankings are biased by team wins, or B) Win Shares greatly understates the true value of closers. We'd still have to sort that question out, but I think the issues would be clarified somewhat.
5:50 PM Feb 13th
 
bjames
Can we focus on the issue of whether team wins significantly influence the *opportunities* that a closer has to record a save?


That issue has no relevance to the validity of the study. So. . . .you focus on it if you want to. But it has nothing to do with the validity of the study.
5:07 PM Feb 13th
 
Guy123
Can we focus on the issue of whether team wins significantly influence the *opportunities* that a closer has to record a save? A winning team hits more HR, but it does not have substantially more HR opportunities. I think that is the best chance for reaching a common understanding.

As I look at the data, it looks like top10 closers have 40%-50% more Save Opportunities than other closers. Since a closer can't create his own Save Opportunity, that must mean the team is doing it by winning more games (prior to closer's entry). Tango looked at 9th inning leads rather than Save Opportunities, but finds the same relationship: teams that win more are much more likely to have a lead of 1-3 runs after 8 innings.

But you know for sure who the top 10 closers are, so you can tell us: do they in fact have more Save Opportunities?

4:27 PM Feb 13th
 
bjames
Saves increase with team wins. So do home runs. In fact, home runs increase with team wins in essentially the same proportion as Saves do. By your logic, we can't use home runs in calculating value, because they increase with team performance.
9:50 AM Feb 13th
 
bjames
Save totals are higher on good teams than on bad teams, but

a) Good teams have better players than bad teams, and

b) Save Totals are a minor element in how Closers are evaluated.
9:40 AM Feb 13th
 
Guy123
The issue is whether Saves bias the study, and cause a false result—a proposition for which there is no evidence.
I'm puzzled that this remains a subject of debate. Tango demonstrated below that every team win generates .36 9th-inning save situations. You said you accept that result. And we also know that closers of all calibers convert the vast majority of these save chances. So a closer's save total *will* reflect the overall strength of his team to a considerable degree, and not only his own performance. Other factors matter too, of course, including the quality of his performance. But closer saves clearly are significantly a function of team wins, and so this must bias your study. So, far from there being "no" evidence, we have conclusive evidence.

Does this bias explain 100% of your finding? I suspect so, but perhaps not. If someone wants to demonstrate otherwise, it shouldn't be hard to do: just use leverage-weighted runs prevented to rank closers, and see what happens.

6:48 AM Feb 13th
 
bjames
Guy 123
Perhaps you could simply report the regular season win totals for the teams of your top 10 closers? Then we would all know the answer, and there would be no need for theory or speculation.


That would be a considerable investment of time, and also, it would do nothing at all to enlighten us about the issue I believe is at the heart of this. The issue is whether Saves bias the study, and cause a false result—a proposition for which there is no evidence. This is merely another statement of the false assumption which you have persisted in repeating since the beginning of the discussion. You are implicitly arguing that IF the top closers are concentrated on the best teams, that would be proof that Saves have a disproportionate impact on the evaluation of closers. This is obviously untrue. IF the top closers are concentrated on the best teams, there could be many different reasons why that is true.

3:54 PM Feb 12th
 
Guy123
But it is not true that the ratings of closers depend to any significant degree on team wins. They do NOT depend to any meaningful degree on team wins. This entire approach is nonsense, because it based on a false assumption.
Hmmm. We know that *saves* depend on team wins to a very substantial degree. Tango showed that below, and you indicated agreement, so I don't think there is any disagreement there. So for this statement to be true, it must be the case that your closer ratings "do not depend to any meaningful degree" on saves (or wins and losses). Is that the argument -- that saves, wins, and losses have almost no influence on your closer rankings? I'm struggling to see how that could be the case, but we don't know the exact methodology.

Perhaps you could simply report the regular season win totals for the teams of your top 10 closers? Then we would all know the answer, and there would be no need for theory or speculation.

8:53 AM Feb 12th
 
bjames
Guy123
Apologies, I described Bill's findings incorrectly in that last paragraph (should have re-read first). Closers, whose ratings are most dependent on team wins, were clearly at the top. RSP and LSP were also ahead of most other positions, but not by a lot.


But it is not true that the ratings of closers depend to any significant degree on team wins. They do NOT depend to any meaningful degree on team wins. This entire approach is nonsense, because it based on a false assumption.
7:15 PM Feb 11th
 
Guy123
Apologies, I described Bill's findings incorrectly in that last paragraph (should have re-read first). Closers, whose ratings are most dependent on team wins, were clearly at the top. RSP and LSP were also ahead of most other positions, but not by a lot.
8:04 AM Feb 11th
 
Guy123
what Guy123 is arguing is that Closers are less important than I think they are because the closers on bad teams blow a lot of save chances.
My point was that *non-closers* are responsible for about 75% of blown saves. While blown saves count as save opportunities for the team, most are not save opportunities for the closer but actually the exact opposite: they are "anti-opportunities," potential closer save opps that were annihilated in innings 6-8, before the closer could enter the game. Team save opportunities are thus irrelevant to understanding how many save chances a closer gets -- so let's ignore them.

What matters for this discussion, I think we can agree, are the save opportunities of *closers*. As Tango shows -- and Bill agrees -- there is a strong correlation between 9th inning save chances and team wins. And we also know that a closer's save total is largely a function of his opportunities. Over the past two seasons, top10 closers have averaged 68% more saves than other closers (42 to 25), mainly because they have 59% more opportunities (46 to 29). Their conversion rate is only a bit better (91% to 85%).

So: 1) wins cause save opportunities, 2) save opportunities cause saves, and thus 3) wins cause saves. Top10 closers appear to be more important than position players because playing on winning teams literally causes a pitcher to be ranked as a "top closer." It's circular logic in action. Alternatively, you can believe that the three positions Bill found to be most important just happen to also be the only three where the ranking depends on team wins and losses. That would be quite the coincidence.
11:20 PM Feb 10th
 
tampabob
Would it be helpful to rank relievers on some combination of "save independent" measures and run the same ranking analysis, or does the low number of batters faced require reliance on saves to measure performance?
11:09 PM Feb 10th
 
bjames
Also, I would point out to you that what Guy123 is arguing is that Closers are less important than I think they are because the closers on bad teams blow a lot of save chances. Which seems to me to be a self-defeating argument.
2:02 PM Feb 10th
 
bjames
Well, actually, Tango's study confirms what I was saying, which is that Saves do not increase proportionally to wins, but only in proportion to the square root of wins.
1:57 PM Feb 10th
 
KaiserD2
Interesting results, Langotier, and they certainly seem to confirm what I was arguing. But why not include visiting teams with leads going into the bottom of the 9th as well?

DK
7:56 AM Feb 10th
 
tangotiger
I did this very quickly, so I didn't have a chance to verify the results. I limited only to home teams so I could do this quickly. I tracked how many home wins each team had, 2010-2016. I also tracked how many times they entered the top of the 9th with a 1 to 3 run lead. The correlation was r=.50 (.4995 actually), with a regression equation of: .36*HomeWins + 6. That's how many save situations they entered the top of the 9th.

The 35 winningest home teams (51+ wins) averaged 52.7 wins and 24.4 save sits entering the 9th. The 35 losingest home teams (36-wins) average 33.7 wins and 17.7 save sits entering the 9th. This gap matches the regression equation.

If someone wants to double-check my work, I have: 2016 Cubs with 57 home wins, and 21 save sits entering 9th. 2016 Braves with 31 home wins and 9 save sits.
3:53 PM Feb 8th
 
Guy123
I think I figured out where Bill went astray. He says below:
In the American League the A's went 69-93, and had 65 Save Opportunities; the Indians went 94-68 and had 48. In the aggregate Save Opportunities do increase slightly the performance of the team, but there are a lot of other variables.

The problem is that the "save opportunity" statistic does not actually measure opportunities for a team's closer to gain a save. It is simply the sum of Saves plus Blown Saves. So a guy who pitches the 8th inning with a lead and give up no runs did not produce a "save opportunity, " but if he allows the tying run to score then he has. So the stat is a weird hybrid of 9th inning successes and failures, plus a bunch of failures earlier in the game. And so weak teams are credited with plenty of "save opportunities."

Oakland was a leader in Save Opportunities because it had 23 blown saves. The closer Madson had 7, but so did Axford. In none of those 7 games did Axford even appear in the 9th inning and in two of them he entered in the 6th -- these games tell us nothing about how many real chances Madson had to record a save. CHW led the AL in save opportunities, largely because Nate Jones (middle reliever) blew 9 saves.

The real question is, do winning teams create more save chances for their closers. Let's define "Save Chances" simply as games in which the team leads by 1-3 runs after 8 innings. Are Save Chances correlated with team wins? The question answers itself....

1:00 PM Feb 7th
 
KaiserD2
In one way or another, I know, I keep raising similar issues here.

I have argued repeatedly that pitchers tend to be judged based on the quality of their teammates. Any judgment based on won-loss records does this, because a pitcher bears less than 50% of the responsibility for whether he "wins" a game.

In the case of closers this is even more striking. The closer only gets into the game if the rest of his team has outperformed the other team for eight innings. That's the only thing that will enable him to have a save opportunity and earn a save.

Bill specifically wrote that two otherwise identical pitchers were ranked in his model according to how many saves they had. I would suggest that if you have two identical pitchers, one on a team that wins 70 games and one on a team that wins 92, both operating as closers, the one on the team with 92 wins is going to have significantly more saves--because there will be, inevitably, more opportunities.

Thus,. it is not in the least surprising that the closer on a championship team (which often wins a good deal more than 92 games) will by this method rank in teh league's top 10. Indeed it would be quite extraordinary if he didn't. It's not that having a great closer allows you to win a lot of games. Winning a lot of games makes your closer look great.

Postscript: Mariano Rivera was in my opinion the most valuable reliever ever by a considerable margin over the course of his career. But the biggest contribution he ever made to the Yankees' success was in 1996, when he was not the closer, and therefore pitched more.

2nd postscript: I totally agree, and my book says so, that the 2013 Red Sox were stronger than any of the other 21st century Red Sox winners.

David K


12:37 PM Feb 7th
 
tangotiger
Just to be clear, the only assertion I'm making is that there is a selection bias to consider, one that may or may not impact the study. It simply adds an uncertainty level to any conclusion.

I agree that now that Bill has laid the groundwork, anyone can replicate the study, and make modifications, and see how that affects the results.
11:52 AM Feb 7th
 
Guy123
But in your posts, guy and tango, you want to assert that I have made this mistake, overrating saves, when you have no real reason to believe that I have made this mistake. If you study the same issue with a parallel study of your own construction, weighting saves as you want to weight them, I believe you will find that my conclusion is valid given the question that I asked

Well, this doesn't really count as a "study," but it should do the trick. I looked at the top 30 closers, and ranked them by the W/L/Save components of season score (3*S+10*W-5L). My goal is not to replicate Bill's full method, but to look at the impact of using saves (and W/L) to help identify top-10 closers. In 2016, the teams of Top10 closers won an average of 88.4 games, and in 2015 they won 86.2. So overall, let's say Top10 closers' teams won about 87 games, compared to 79 wins for the other 20 teams -- a gap of 8 wins.

How much better were the Top10 closers than the other 20? Well, the top10 had an average fWAR of 1.5, compared to 1.0 for the next 20 -- a difference of half a win. Perhaps that slightly understates the value difference, but even if we double the WAR difference, it explains only 1/8 of the gap in team strength. Pretty clearly, ranking closers based on saves, wins, and losses elevates those who play on winning teams, far beyond the closer's own impact.

Would a measure of closer performance unbiased by game outcomes still show a positive association with world championships? Maybe! Couldn't hurt to look. But even if it does, that still leaves the question of which way the causal arrow points. Maybe elite closers are just an indulgence that prospective champions (often wealthy teams) are more likely to indulge in. After all, we would probably find that teams with very expensive corporate suites win more championships, and that girls with million-dollar engagement rings are more likely to be supermodels. But we don't believe that suites create champions, or that rings create beauty.


9:06 AM Feb 7th
 
Brock Hanke
This has nothing to do with closers (which may be a plus at this point), but you mentioned Bucky Dent and Buddy Biancalana. I started paying loose attention to the idea that World Series might feature the occasional player who didn't do much during the season breaking out during the postseason because, well, if you're pitching in the 1985 Series, you're worrying about how to get out George Brett or Jack Clark, not Biancalana or Tito Landrum, who had to fill in when Vince Coleman got tarped. So you may end up feeding this guy the one sort of pitch he can handle, simply because you don't know that this is the one thing to not do. Biancalana, you know about. Landrum, batting 5th through 7th, hit .429 in the playoffs vs. the Dodgers, and then .360 with a homer and two doubles against the Royals. Landrum was a decent hitter, but nothing like that. I've never formally studied the issue, because I'm not sure how to define "didn't do much in the regular season", but it does seem that the Series get more of these guys than you'd expect. Does this seem reasonable to you? Have you ever looked at the issue?
11:04 PM Feb 6th
 
JackKeefe
Insulting? Really? The first word I used was "Thanks" and I don't believe I used the word "mistake" once in my comment. I just thought it seemed like a nice little study, showing that closers are important, by demonstrating that the vast majority of championship teams have employed a top-notch closer. I don't think the study proves anything more than that, and it wouldn't even be necessary to make the case that closers are important if you've watched baseball at all over the past few decades except that certain otherwise intelligent individuals have been jumping all over themselves trying to denigrate the position. Some of these people can be found in the comments section. You posited that closers were important, I was trying to explain why. Nothing insulting about that.

If you're going to carp on me for being critical, I might as well get my money's worth and offer a criticism. The most important skill in baseball is hitting, which has no set position. You might find more hitters in less defensively demanding positions, but really, hitting can come from anywhere. It's a little misleading to rate the positions in this way. I'm of the opinion that having decent hitters all the way down your lineup is better than having a few superstars mixed in with the scrubs, because you can always pitch around the superstars. As long as you've got enough hitting, you can build the leads that need saving. It doesn't really matter if your right fielder is rated 1 or 2. But that's a quibble for another day.

If you wanted to do some sort of measurement based on my comment---and I'm not asking you to---I suppose it would be something as simple as the last out is 27 degrees more important than the first out, and count backwards from there. I have no idea idea what value to assign one degree beyond that little shorthand. Even if I could, I always get a little queasy assigning specific coordinates to something as fluid and ephemeral as a baseball game. There's just too many variables. I'm happy to judge what I see in the moment.

For instance, one thing that a closer must deal with that can't be adequately measured is the feeling of standing on the mound, all alone, with 40,000 fans going berserk and the opposing team streaming out of the dugout as you've just given up a walkoff hit that has negated the rest of your team's hard work for 8 innings in an instant. And then you have to take the hill the next night and do it again, if you can. Some closers have been destroyed by it. Mitch Williams, Brad Lidge. Donnie Moore committed suicide. One of the hallmarks of Mariano Rivera's greatness is not just his staggering number of successes, but how he handled the most brutal losses. Giving up the game winning hit in Game 7 of the 2001 World Series for New York, just weeks after 9/11? Or how about when the Red Sox nicked him for a run that began their epic comeback down 0-3 in the 2004 Championship Series? The Sox faithful serenaded Mariano on opening day 2005, and what does he do? He tips his cap, and goes on for another 8 years as the best closer in baseball. How do you quantify that? I can't. I'm sure you've tried, but I just didn't get the memo. No offense meant. Two people are allowed to have the same thoughts.

In the meantime, I hope your study has quieted some of the yahoos out there who have likened Mariano Rivera to a punter in football. Now that is insulting.




8:09 PM Feb 6th
 
OldBackstop
By the way, Bill, this is one of the most interesting sabermetrics discussions in quite awhile. Nice work.
8:06 PM Feb 6th
 
OldBackstop
..ermmm....maybe I caught you at a bad moment....

But I have always wondered if strong teams might provide their closers with more "A" level save opps -- three run leads were they can just fire away, that great defense behind them -- while weaker teams might provide one run leads with the kids from the farm on D up in September?

Also, "wins" as stats for closers in Season Scores...are those just scattered pretty evenly, or do strong teams hand their closers some fortuitous stats with their bats (see the last postseason game with Chapman).

I'm POSITIVE you've already thought about this!!! Asking for my own edification....

5:19 PM Feb 6th
 
bjames
Well, obviously Keefe's post is insulting, and Maris's endorsement of it is insulting, because they assume that we are morons who have never thought to deal with these complications that the immensely clever Jack Keefe has thought of, when in reality, we have invested many thousands of hours over a period of decades struggling with these issues. He is free to assume we don't know anything about them because he doesn't actually know a damned thing about the subject. They assume that we have made mistakes that, in reality, we have worked very hard to avoid.

But in your posts, guy and tango, you want to assert that I have made this mistake, overrating saves, when you have no real reason to believe that I have made this mistake. If you study the same issue with a parallel study of your own construction, weighting saves as you want to weight them, I believe you will find that my conclusion is valid given the question that I asked.
2:40 PM Feb 6th
 
Guy123
Well, count me convinced!
2:00 PM Feb 6th
 
bjames
1) I absolutely promise you that Guy123 is COMPLETELY wrong. This finding is NOT due to saves being related to wins. It just absolutely is not.

2) Keefe's point is interesting, but miles behind the sabermetric discussion. The points he makes are true but well known, and have been well known for decades. The problem is quantifying them: this is true, but how large of an effect is it?

Without referring to actual data, you can project the effect of closers being used in game situations to be any size of adjustment that you want it to be. But in ACTUAL analysis, you have to figure out a way to measure it.
1:50 PM Feb 6th
 
jemanji
Following up on the below ... agreed of course that once you have a first cut at triangulating a problem, you should then use the best tools possible to investigate. Now that Mr. James has alerted everybody to the probable impact of closers in the World Series, a lot of followup will ensue. As usual :- )

The point I'm emphasizing is that you can act on data that is not formally valid. Semmelweis discovered hand washing based on simple observation of his patients -- 69 of 120 patients over here got an infection, that type of thing.

An oncologist has used drugs ABC to get 9 months' remission in Joe Shlabotnik. Used DEF and got 3 months' remission. Now must decide (based on tolerance, e.g.) between drugs B+G+H or drugs D+M+N -- as they function in Joe Shlabotnik's system. The doctor will probably use BGH, even at higher cost, based on .... a laughable sample size. Life-or-death decisions are made like this in health care all the time.

A study with normal clinical validity (vis-a-vis Joe Shlablotnik) would be ideal, as you put it. Most of our life decisions don't enjoy that luxury.

Similarly, a GM of a very good major league team has to decide whether to give a Closer his extra year on the contract. He can use Mr. James' information here, or he can ignore it...

Maybe a real GM looks at this and decides, "I don't buy it." Fine. ... Me? Until better information arrives, if I'm that GM, I use the information, weighing it rather heavily. When the next-gen info arrives, great. :- )


10:35 PM Feb 5th
 
jemanji
For sure. Thanks for that.

The selection bias is a threat to *formal* validity until accounted for. In fact I'll raise and re-raise :- ) because the 40-team roster of Series winners *is in fact* biased, and heavily biased.

With respect, I believe it gets a bit philosophical here. In most formal settings, such as FDA approval of a drug, the review panel has arrived at a (diamond-hard) consensus, as to what it deems convincing.

But no matter how universal the consensus, it still has a subjective element. Is 95% a convincing threshold? Is 99.5% convincing? We say "this standard is acceptable, that one is not" based on factors that are, at the end of the day, subjective no matter how many folks have signed off on the standard.

Here, accustomed to a paradigm based on regression and elimination of bias (to our own 'satisfaction'.)

But sometimes we then turn around and accept 'looser' evidence as the basis for 'beliefs' - that which we find convincing. A doctor suspects Avastin of causing vocal chord problems and stops prescribing it to opera singers, let's say. He may be operating on data less structured than Bill's. The first nurse to start washing her hands between patients, same thing.

Personally I don't need formal validity here to find the data convincing, or at least very suggestive. If you do, I respect that.

Thanks Tom.



10:02 PM Feb 5th
 
tangotiger
I agree that further investigation is warranted. The point is that you can accept the selection bias if you can account for the selection bias. The one position that is most likely to have the selection bias is the one position that shows the most impact. It simply calls for a more nuanced approach.

Bill's season score is a good metric, but using the NOWOLOS part (no wins or losses or saves) would have handled ONE part of the bias.


8:20 PM Feb 5th
 
jemanji
.. by the way Tom, count me among your legions of fans. Admire your work so much.

I like the way you phrase it, "the ideal is" to choose performance before season.

As you know, when we back up and ask about --- > possible inherent differences between playoff baseball and regular-season baseball, we run into the type of selection bias you point out. Just gotta work around the situation, as Mr. James did. He's great at triangulating problems that are awkward to deal with.


3:58 PM Feb 5th
 
jemanji
Selection bias is very undesirable but it is not an absolute consideration to the exclusion of all others, right?

Suppose we have the hypothesis, "The playoffs are no different in any way from regular-season baseball?" How do we test this hypothesis? We're going to have to use the best investigative methods at hand, whether or not they're as conclusive as we'd like. Beats guessing.

Bill asked whether relief pitching was disproportionately impactful in the World Series, and -- being unfriendly towards the affirmative -- tried to prove that it was not, so to speak. However, an RP phenomenon seemed to be reflected in his data.

Further investigation seems warranted to me.

- Jeff


3:49 PM Feb 5th
 
tangotiger
I agree with Guy. Any time you take the AFTER-SEASON performance to estimate the IN-SEASON talent, you will get into a selection bias. And relief pitchers are especially subject to such a bias simply because of the way we evaluate them. This is why the ideal is to choose performance BEFORE-SEASON.

To try to think it through, imagine you wanted to evaluate the rosters of teams for the 2017 season, on March 31, 2017. Well, all you can do is try to "forecast" a player's talent level for the 2017 season based on whatever data you have to that point.

So if you wanted to evaluate the 2016 rosters, you have to base it on knowledge that does NOT include their 2016 performances. That is, you have to "time travel" yourself back to March 31, 2016.


2:35 PM Feb 5th
 
Guy123
It seems highly likely that the main finding here is just an artifact of saves being a major component in the reliever ratings (both Wins Shares and season scores). And teams with a lot of saves tend to be teams with a lot of wins, and teams with a lot of wins are of course more likely to become world champions (both because they are talented teams, and because you need wins to make the postseason).

Looking at franchise totals over the past 5 seasons (combined), there is a .69 correlation between W and S. About 52% of wins generate a save -- that's true for both good and bad teams. If I know how many saves a team had, I can make a pretty fair estimate of their win total. Of course, *some* of this correlation is because a good closer actually helps the team win. But saves captures much more than the closer's individual performance. A team with 45 saves will, on average, win about 5 more games than a 35-save team. But the closer on the first team will obviously not be anything close to 5 wins better than the other closer. So we know that many things beyond the closer's contribution is "leaking" into the save statistic.

jwilt also makes an important observation below: a good closer probably signifies a superior bullpen below the closer level. After all, a closer is, by and large, simply the best of a team's 7 relief pitchers. Mathematically, a team with a weak closer literally cannot have as good a #2 reliever as many of the teams with elite closers. So it would be very surprising if Bill's analysis is not inadvertently measuring the quality of the bullpen more generally.

While the results of Bill's study are perhaps intriguing, they really give us no reason at all to believe the closer position is more important than any other position. To test that theory, this analysis would have to be done using statistics that isolate a closer's performance from game outcomes.
2:26 PM Feb 5th
 
ksclacktc
@Jack Keefe Nice post! a little logic and common sense never hurt.
7:05 AM Feb 5th
 
ksclacktc
@JackKeefe
7:05 AM Feb 5th
 
MarisFan61
I don't agree that it's 'extremely careless' to wonder strongly about some things based on these findings, because I think the degree of the findings is enough to indicate a likelihood, even considering the not-that-large sample. It seems large enough to wonder strongly.

I also wish the wording of the criticism hadn't been the same as Comey's wording. :-)
12:12 AM Feb 5th
 
sprox
Seems like a 100+ win team (on paper) can afford to blow several leads late in games and still win the division.

Say they convert only 42 of 53 save chances instead of 50 of 53 chances, essentially dropping the team from 103 wins to 95, still more than enough to win the division with room to spare

but you get into the playoffs and suddenly that blown save chance moves you from a 3-1 series lead to a 2-2 tie or from a 4-3 series win to a 4-3 series loss

this seems to fit the narrative but may or may not have any merit
10:10 PM Feb 4th
 
JackKeefe
Thanks for adding your two cents to the proposition that closers are indeed important, when so many in the sabermetric community think of them as practically worthless, despite the years (nay decades) of post-season evidence proving otherwise. So let's move from the question of whether closers are important (an emphatic yes) to why.

Let's start with the obvious, but one that is so obvious that it is never mentioned. The last inning is the only inning where the score counts. If you are up by one in the final inning, you win. That's the only inning you can say that. So it's supremely important to maintain the lead in that inning.

This leads to a more controversial point. A solo home run that gives a team the lead in the bottom of the ninth wins them the game, with no chance for the other team to respond. A leadoff home run in the top of the first gives the opponent 27 outs to respond. So a run scored late in the game has more value than a run scored early in the game, proportional to the number of outs the opponent has left to match that run. That's why you see managers hellbent on keeping other teams from scoring from the 7th inning on. Not all runs are created equal.

Lastly, the reason so many stat-heads dismiss closers is that they pitch fewer innings than starters. But I think that misses the point. Closers may pitch fewer innings, but they typically make twice the number of appearances than starters, so there are more games they have an impact on.

In fact, if you've got a lockdown closer, just the threat of him lurking in the bullpen puts pressure on the opponent to take an early lead. You are essentially playing an 8 inning game. Depending on the depth of your bullpen, you can whittle that number down to 7 innings, or even 6. That was the KC Royals formula, and the Rivera Yanks. Stay close early, win late. It puts enormous pressure on the other team every single game. Starters just can't match that threat more than once a week.


4:23 PM Feb 4th
 
wovenstrap
Turns out, there IS a clip reel of Rivers in Game 6 of the 1977 World Series, which was a very famous game. It's enough to give an impression of what Rivers was like as a player (signature move: bat spin after swing & miss). Posting here because ABC showed a graphic during one of his at-bats, to the effect that in the regular season, Rivers had the highest batting average of any Yankee after the 6th inning -- precisely the kind of information that is NOT associated with the late 1970s. But someone was on the case.
5:58 AM Feb 4th
 
wovenstrap
I grew up in the NYC area in the late 70s, a Yankees fan, and I just wanted to thank you for putting the name "Mickey Rivers" into my head this morning. I haven't thought about him for 20 years. As you said, he was a good player, also a very distinctive and unusual player, fun to watch and I'm pretty sure good with a quote in the clubhouse. I'd give anything to watch a 10-minute clip reel on YouTube about him.
5:48 AM Feb 4th
 
shthar
This kinda fits in with a thought I had when the Yankees signed Chapman for 80 Million dollars.

If you're not going to the world series, it's INSANE to pay that much for him.

If you ARE in the world series, he's worth a BILLION dollars.
12:47 AM Feb 4th
 
jeffsol
Not drawing any conclusions but thinking about why it might be true that WS Champions have such high quality closers which would not indicate they are generally more important than previous analysis has broadly indicated, there are two ideas that come to mind for me:

1) As stated earlier, perhaps closers are more important, as many believe, for winning in the postseason. Certainly the postseason schedule allows a top closer to pitch a much greater share of innings than they ever could in the regular season, which could emphasize their importance in winning the WS rather than getting to the postseason.

2) As many (most?) closers have been, at some level, failed starters, it seems like it may be easier to find a great closer than a great starter. Perhaps better organizations are just more efficient at finding a closer in one way or another, and being a better organization, are more likely to win the WS.

Just my thoughts, I could be way off base but was what came to my mind as I read the article....
12:43 AM Feb 4th
 
wdr1946
What about the second- fifth starters? Shouldn't they be brought into the picture? If your No. One starter is Lefty Grove in 1931, and your fourth starter is Steve Blass in 1973, they cancel out. Also, I'd like to see this ranking with pre-1977 teams. I suspect that the 1927 Yankees would be No. One.​
12:16 AM Feb 4th
 
evanecurb
Rivera was on five of the 40 world championship teams. Is it possible that he might skew the results a little?
9:52 PM Feb 3rd
 
CharlesSaeger
More substantively, what would the pitcher and closer rankings be if you tossed out saves, wins, and losses?
9:16 PM Feb 3rd
 
CharlesSaeger
"So the irony is that the Yankees’ weak spot in ’78 was shortstop Bucky F. Dent"

Let me hazard a guess that this article was written by someone who miiiight be a Red Sox fan.
9:11 PM Feb 3rd
 
jgf704
I would echo jrickert's question: What does a similar analysis for the 4 teams that make it to the championship series give?
8:37 PM Feb 3rd
 
hotstatrat
I wonder what the average reliever ranking is if you skip the Yankee championships during Mariano's years. I figure he alone improves reliever's standing in this test. He is such an outlier. Sure, the Yankees were helped in their championships thanks to Mo, so it's not fair to ignore him, but as you note 40 teams is not a huge sample size and having the all-time greatest reliever on a team that was in the post season an unprecedented 17 times in 18 years - all but the first two with Mariano as their closer - he's got to have a huge impact on this study.
7:09 PM Feb 3rd
 
337
More arguing: Tresh's "very unimpressive" year in AAA in 1961 had him second in the league in hits, 6th in BA. Regulars in the International League in 1961, btw, included Boog Powell, Donn Clendenon, Joe Morgan, and a 45-year-old Luke Easter.
6:02 PM Feb 3rd
 
bjames
Responding to JRickert's query:

I suppose the next questions would be: What does this process show for the teams that lost the World Series?
What does this process show for the teams that lost the League Championship Series? I'm not sure 40 years is enough to show a difference using these numbers but I am curious what the data show.



Well, it would take a long time to re-run the entire process, but I checked the rankings of the CLOSERS on teams which lost the Weird Sorrels. The average rank of the closer on the losing team in the Series is 10.8. . . .much higher than the winning teams, but still quite low compared to the known norms. Only 22 of the 40 teams had top 10 closers (55%), which is on the low end. Six Series-Losing teams had #1 relievers, which is very high. The only Losing World Series teams to trade for their closer in mid-season were the 2016 Indians and the 1991-1992 Braves.

I suppose I should have emphasized in the article that one needed to be careful in drawing conclusions based on a study of 40 units, but I took that to be obvious. Several responders were EXTREMELY careless in the inferences they drew, so I should have included the caution.
5:33 PM Feb 3rd
 
337
Well, if you're not supposed to argue with me, it's probably bad form for me to argue back--BUT Tommy Davis started 32 games in RF in his entire career. He was the Dodgers regular LFer from 1962 through 1966 (except for the year that he missed most of the season to injury), and Tresh did move around more than I remembered (filling in for Kubek at SS, Boyer at 3B, Mantle in CF when they were hurt) but he played many more innings in LF from 1963 though 1967 than any other Yankee. You're right that it was a throwaway line, however, in that they ended an amazingly long and successful run for both without really ever having an established regular at LF for either team during the 1950s and early 1960s.​
5:29 PM Feb 3rd
 
bjames
apaster--But you couldn't answer those questions by this approach. All you would have is the discrepancies between one set of averages and the other, and those discrepancies couldn't possibly be large enough to sustain any conclusions.
4:51 PM Feb 3rd
 
apaster
To what extent is the importance of a closer tied to post-season performance rather than performance overall? In other words, how would these results differ if we looked at the teams with the best record in the major each year, rather than the world series champion.

Or, to approximate the answer without having to run the study for more teams, were the closer-heavy champions better or worse teams overall?

The question is whether having a good closer is an advantage in general for teams, or whether, as popular wisdom suggests, more important in the postseason.
3:58 PM Feb 3rd
 
bjames
The Cubs this year had the best record in baseball; they had 53 Save Opportunities. The Padres, Diamondbacks and Braves had records of 68-94, 69-93 and 68-94, not necessarily in that order--and had 53, 53 and 58 Save Opportunities.

In the American League the A's went 69-93, and had 65 Save Opportunities; the Indians went 94-68 and had 48. In the aggregate Save Opportunities do increase slightly the performance of the team, but there are a lot of other variables.
3:43 PM Feb 3rd
 
jimmybart
Bill: not a trade, but the 2006 Cardinals did switch from Isringhausen to Wainwright late in the season due to an injury to Izzy, as I recall. Which probably was a blessing in disguise because he was faltering dramatically (along with the rest of the team) in the second half, only to right the ship just in time. So the '06 team had the worst W-L record to ever win a World Series (83-80), right after 104 and 105 win seasons. Go figure.
3:20 PM Feb 3rd
 
FrankD
What is the result of ranking the teams by median player rank? Or ranking the teams by standard deviation of individual rankings? Are teams better off having all players rank 15 or half of the team rank 1 and other half 29 ?
3:01 PM Feb 3rd
 
FrankD
I like the study. But it does assume a linearity in the 1-30 assignments. Is having a 1 and a 28 on your team the same as having a 14 and a 15 on another team? I would prefer having the 1 and the 28 but I have not done analysis to show this to be valid. Perhaps a slight weighting is warranted to better capture the effect of the spread between top/bottom players, but any weighting scheme is also subjective. I do like the simplicity of presented method.

Also - isn't the effect of a closer magnified in a short series vs regular season in that a closer can appear in a higher percentage of the games of a series than a season? So having a top closer is a bigger advantage when going against a team with a lessor closer in the playoffs.
2:56 PM Feb 3rd
 
jrickert
I suppose the next questions would be: What does this process show for the teams that lost the World Series?
What does this process show for the teams that lost the League Championship Series? I'm not sure 40 years is enough to show a difference using these numbers but I am curious what the data show.


2:26 PM Feb 3rd
 
jwilt
Previously I wondered out loud if having a good closer was indicative of having a good bullpen overall, and if a bad closer indicated you probably had a bad bullpen.

So I did a tentative start of a query to figure that out. Please poke holes in this, I did it in 10 minutes. Using bb-ref I found the worst and best ERA+ of everyone with 25 or more saves since 1976. I cut the list off after 19 of each, basically because that was where the bad list hit an ERA+ of 90 after you throw out the guys who'd been traded in midseason.

On bb-ref's team pages they list the top five relievers in games pitched in between the starters and the other guys on the staff who pithed less. What I did was using those five pitchers (really four after you take out the closer in question) I counted how many had ERA+'s of 100 or better.

The 19 bad closers teams had an average of 2.68 of the four with league average or better ERA+'s. Three of the 19 teams had all four relievers behind the closer with 100 or better ERA+'s.

The 19 great closers had an average of 3.21 teammates with league average or better ERA+'s. 11 of the 19 teams had all five of their relievers with the most games pitched better than average.

So... the first back-of-the-napkin quick-and-dirty look seems to say good closers tend to have better bullpen teammates than bad closers. But certainly not always. You have teams like the '04 Marlins where Armando Benitez' 1.29 ERA and 47 saves were dramatically better than anyone else in the pen, none of their next four relievers had an ERA+ of 100 or better. That was the only team like that of the 38.
2:06 PM Feb 3rd
 
bjames
Regarding Tresh and Tommy Davis solidifying Yankees and Dodgers in left field. . .I realize it is a throwaway line and not really relevant to your point, but not true. Tresh (who as I recall had a very unimpressive AAA season in 1961) would probably have been in the minors in 1962, but Tony Kubek had to do military service. The Yankees pressed Tresh into service not as a left fielder, but as a shortstop, then he hit well and they moved him to left. But then Mantle got hurt in '63, so Tresh actually played CENTER in '63. . .and bounced back and forth into center I think the rest of his career.

Tommie Davis sort of the same thing; he actually had runs with the Dodgers in right field and at third base. They were trying hard in '61 and '62 to make him a third baseman. So I don't think either of these players really solidified the left field position.

Well, I'm not supposed to argue with the posters, so I'll go back in my hole now. . .
1:17 PM Feb 3rd
 
jemanji
The star-studded 76 Reds and 78 Yankees ... thanks for that. The Big Red Machine was my boyhood team, Johnny Bench my first sports hero. Enjoyed it.

I always felt like it was a shame we lost that era; it seemed to me that the great teams of the 1970's (including the 75 Red Sox, the Finley A's, etc) were more "satisfying" and that we've lost something since. (Still with the 85 Royals, etc., we were seeing the same thing.)

But some of those 1970s teams had Real Madrid lineups, stars at every position, who were homegrown in a "fair" manner. This study clarifies that in a concrete way.

Fun stuff.


1:02 PM Feb 3rd
 
jemanji
Great study. You wonder whether this light bulb will impact baseball a bit like the Saves stat did.

A cool aspect of it, is that it reminds us that playoff baseball is 'valid,' as it were. It's fashionable with saberdudes (elsewhere) to consider a championship nothing but luck. To tut-tut about celebrating the 2015 Royals any better than their +83 run differential.

A great relief stud is like a "Get Out of Inning Free" card, and in elite baseball games it turns out that card weighs quite heavily in a strategic and practical sense.

For me, a clarity about this point makes baseball a more coherent strategy game and more fun to watch. Plus, we got Edwin Diaz ;- )


12:58 PM Feb 3rd
 
bjames
Regarding contending teams trading for a closer at the deadline/going on to win the World Series. . . .the 2016 Cubs are actually the ONLY instance of that happening. It has probably happened more often at every other position than closer; I don't know. The only other vaguely similar case was the 2003 Marlins. The Marlins traded for Ugueth Urbina in mid-season, but that is the exact opposite of the syndrome described by Hal10000, the team trading the closer who is having a great year. He was expensive, and he was having a lousy year for Texas, so they wanted out of his contract. The Marlins took him on, NOT to be their closer but as bullpen depth; Braden Looper was their closer at that time, and Urbina did not have a save opportunity in his first six weeks with the Marlins. But then Urbina started pitching well and Looper was not pitching great, so the Marlins flipped them in September.


The argument about good teams having more save opportunities is also not true and a waste of time, but. . . .
12:57 PM Feb 3rd
 
ksclacktc
Nice article Bill!

You used the tools of sabermetrics to answer a question you wanted answered. And, then answered it to your level of satisfaction. You have always operated in this manner.

It is refreshing, because as a fan since 1982 when I was a freshman in college, I have seen many changes in how baseball is evaluated and understood. Many of them have been magnificent and revolutionary. Way too often now, I see the love for an analytical tool and the end of any further challenge to the question. If the numbers say this, that's the end. No logic, common sense, or open mind. Just hubris.

Good work revising your view on this issue.




12:54 PM Feb 3rd
 
jwilt
This being the case, to what extent are World Champion teams more likely to produce a top closer simply due to extra opportunities?

That's also one thought I had. I've read that bad teams and good teams don't differ a lot in numbers of close games, that good teams have more wins that aren't close. But this would be an interesting study.

Also... maybe having a good closer is tied to having a good bullpen overall. At least sometimes. If you have a closer with a 4.00 ERA that probably means you have a setup guy who's not great, and a 3rd guy with a 4.75. And a good 6-man bullpen probably does have a fairly high correlation to winning. You're not usually going to the Series if your best reliever of six or eight has a 4.00 ERA.

I also like Hal's idea that closers often are the first thing thrown overboard when you're 40-70. The old "who needs a great closer when we're down 8-3 in the 9th every day?" bit.
12:50 PM Feb 3rd
 
rtayatay
How much do team wins influence closer saves, and how much do closer saves influence the closers' rankings? It seems like those might be related.
12:49 PM Feb 3rd
 
337
Re: the lack of excellent LFers on World's Championship teams, I've long noticed that the two dominating dynasties of 1950s and early 1950s, the Yankees and Dodgers, were always weak in LF. Sometimes they had a decent patch out there (Andy Pafko) but LF was usually a spot they played someone for a year or two and then improvised their next, until Tommy Davis and Tom Tresh came along at the same time.​
12:24 PM Feb 3rd
 
bjames


Interesting results, but I wonder if the best teams tend to seek out the best closers once they realize that they are World-Series-worthy, and thus the World Series winner has a top closer.


I think that has only happened two or three times. People are just focusing on it because the 2016 teams did that. I'm not sure there is ANY other case where it has happened. I'll go check.
12:23 PM Feb 3rd
 
Hal10000
nteresting results, but I wonder if the best teams tend to seek out the best closers once they realize that they are World-Series-worthy, and thus the World Series winner has a top closer.

This was my thought exactly. If you have a great second baseman and you're not contending, you'll probably hold on to him, hoping to contend next year. But if you have a great closer and you're not contending, he's probably the first thing you'll trade. It would be interesting to see how many of these top-10 closers were acquired during the season (or how many were just Mariano Rivera) compared to other positions.

My feel on the "strength up the middle" cliche was it reflected the idea that it's harder to get a good up-the-middle player than a good corner hitter. Ranking them this way can take that bias out, comparing them to their peers. But it always felt more like "you can find a hard-hitting first baseman anywhere" type deal.
11:50 AM Feb 3rd
 
rucksack
"Playing time counts; a player who hits .280 in 300 at bats is not equal to a player who hits .280 in 550 at bats. A closer who has a 1.80 ERA in 45 games and 27 saves is not equal to a closer who has a 1.80 ERA in 70 games and 42 saves."

This being the case, to what extent are World Champion teams more likely to produce a top closer simply due to extra opportunities?
11:41 AM Feb 3rd
 
shinsplint
Interesting results, but I wonder if the best teams tend to seek out the best closers once they realize that they are World-Series-worthy, and thus the World Series winner has a top closer. This could mean that there is a correlation between World Series winners and great closers without great closers necessarily being that much of a cause of the championship.
11:30 AM Feb 3rd
 
FrankD
Able to overcome having poor closer ('87) and no left-handed pitching ('91) maybe we should give Tom Kelly more credit as a manager ...... with a nod to HHH stadium
10:45 AM Feb 3rd
 
MarisFan61
Does that mean that not only has the field of sabermetrics tended to underrate the importance of closers, but MLB has too?

Even though out loud everybody else besides sabermetrics has seemed to value closers more than sabermetrics has, closers have never been paid as much as starters. Of course some of them have gotten mega bucks, but in general, it seems the money offered to top closers has only been something like what #4 starters get. Maybe #3 starters, but not higher.

A rationale for this could be that relievers are less predictable than players of other positions, including starters. Maybe the thinking is, yeah, you need a top closer, but you can't count on it very well that yesterday's top closer will still be good tomorrow. Certainly this is truer for closers than for position players, but is it true for closers vs. starters?

Of course another rationale is that you need to make the post-season, and presumably the top closer isn't as crucial for that. But there's always at least a team or two that's eager to pay 'whatever it takes' to maximize their chance to win it all NOW. I think this means MLB has indeed also been somewhat undervaluing top closers.
10:25 AM Feb 3rd
 
 
©2017 Be Jolly, Inc. All Rights Reserved.|Web site design and development by Americaneagle.com|Terms & Conditions|Privacy Policy