Batter's Box Interactive Magazine Batter's Box Interactive Magazine Batter's Box Interactive Magazine
by Leigh Sprague

We have a Pinch Hit today from Leigh Sprague, who will be familiar to you all from his posts on Batter's Box. Welcome Leigh! He writes \\"suffering a horid case of BWS (Baseball Withdrawal Syndrome), I decided to dive headlong into the Manager Stats on Baseball-Reference.com...\\"



Part I


In 2003, Grady Little, Carlos Tosca and Buck Showalter were the only three Major League managers without Major League playing experience. Little was fired early in the offseason, leaving only Tosca and Showalter. Why are there only two? Do former players make the best managers? are managers who were above average major league players better managers than those who were not very good players?

Before beginning the research, I developed a couple of hypothesis:

1. The best managers either (a) played very well, or (b) never played.

2. The worst managers played in the Majors, but very poorly.

I used all managers who (a) began managing any time between 1980 and 2003, and (b) managed at least 100 games. Since 1980, the majority of managers have been former position players. The years 1980-2003 saw the beginning of 96 managerial carers, and of those 96 managers, 81 were former major leaguers. Of those 81 former major leaguers, only four - Larry Dierker, Marcel Lachemann, Phil Regan and Larry Rothschild - were pitchers.

I divided the managers into five groups: (a) 105 or better carer OPS+, (b) 81-104 carer OPS+, (c) 80 or worse carer OPS+, (d) only around for a cup of coffee, and (e) did not play

[All groups sorted by carer winning percentage. Managers with an asterisk were pitchers, and ERA+ substituted for OPS+. Those in the Cup of Coffee group had fewer than 500 carer plate appearances or 100 innings pitched]

Group A - 105 or better carer OPS+: Davey Johnson, Bob Brenly, Dusty Baker, Lou Piniella, Pete Rose, Mike Hargrove, Felipe Alou, Doug Rader, Don Baylor, Tony Perez, Clint Hurdle, Hal McRae, Larry Parrish, Buddy Bell, Davey Lopes, Frank Howard and Alan Trammell.

Group B - 81-104 carer OPS+: Larry Dierker*, Bill Russell, Cito Gaston, Mike Scioscia, Bobby Valentine, John Wathan, Art Howe, Phil Regan*, Cookie Rojas, Marcel Lachemann*, Jim Lefebvre, Jim Essian, Bruce Bochy, Ray Knight, Butch Hobson, Phil Garner, Tony Pena, Steve Boros, Bob Boone, Jerry Coleman, Terry Francona, Lloyd McClendon, Tony Muser, Russ Nixon and Jim Davenport.

Group C - 80 or worse carer OPS+: Bob Melvin, Ron Gardenhire, Tim Johnson, Bud Harrelson, Hal Lanier, Johnny Oates, Bob Lillis, Gene Michael, Buck Rodgers, Larry Bowa, Buck Martinez, Doc Edwards, Jerry Narron, Chuck Cottier, Billy Gardner, Mike Ferraro, Ned Yost, Bill Plummer, Bobby Mattick, Jerry Royster and Luis Pujols.

Group D - Cup of Coffee: Ken Macha, Jim Tracy, Jimy Williams, Joe Morgan (no, not that Joe Morgan), Charlie Manuel, Jim Fanning, Jerry Manuel, Gene LaMont, John Felske, Tom Kelly, Jackie Moore, Tom Runnells, Lee Elia, Rene Lachemann, Jim Snyder, Eric Wedge, Eddie Haas and Larry Rothschild*.

Group E - Did not play: Grady Little, Carlos Tosca, Kevin Kennedy, Jim Frey, Buck Showalter, Terry Collins, Terry Bevington, Greg Riddoch, Tom Trebelhorn, Ray Miller, John Boles, Jim Riggleman, Nick Leyva, Stump Merrill and Cal Ripken Sr.

Here are the winning percentages for each group (they do not add up to an even .500 because I included only managerial carers which began between 1980 and 2003):

Group A .506
Group E .496
Group D .491
Group C .487
Group B .484

Suspicion confirmed, I suppose. The winningest managers are those who never played, played very little, or played very well. Those managers who played, but were not good hitters (here's lookin' at you, Mr. Guillen) had the worst record.


Part II


[Note : The \\"Pythagorean Theorem\\" uses runs scored and runs allowed in order to ascertain a team's probable winning percentage. It is generally accurate. The formula is Runs Scored^1.83/(Runs Scored^1.83 + Runs Allowed^1.83).]

While looking through manager's data during my research for the above, I got the idea of Pythagorean +/- for managers. I looked at the team's actual W-L record relative to the it's Pythagorean W-L. If any one manager managed more than 100 games for that team in that season, then that manager got the credit for the +/-. For example, in 2002, the Blue Jays' actual record was 86-76, whereas their Pythagorean W-L was 87-75. So, for 2003, Carlos Tosca earned a -1.

Obviously, I am not trying to infer anything about Tosca's managerial skills from a -1 +/- in 2003. -1 is certainly small enough to attribute to chance.

What about causation? Well, there could be many reasons for a team either surpassing or falling short of its Pythagorean W-L total. Most notably: luck or lack thereof. But sometimes the correlation is really strong. Is it possible that certain managerial attributes help a team's distribution of runs scored and runs allowed? Conversely, is it possible that certain managerial attributes hurt a team's distribution of runs scored and runs allowed?

I was skeptical at first, but then I found one delightfully strong correlation. No team managed by Jimy Williams has ever reached its Pythagorean W-L: not the Jays of 1986, 1987 or 1988; not the Red Sox of 1997, 1998, 1999 or 2000; not the Astros of 2002 or 2003. That's ten teams finishing with records worse than their run differentials would indicate. The odds of flipping a coin ten times and getting tails each time is 1 in 1,024. Maybe Jimy is unlucky and his teams keep coming up tails, or maybe there really is something there. Whether it is ineptitude, misfortune, or some sort of gypsy curse, I would not want Jimy in my team's clubhouse.

Of those managerial carers that began between 1980 and 2003, here are the top 10 in cumulative Pythagorean +/-:





Manager Team(s) Seasons +/-


Bobby Valentine Tex,NYM 13 +22
Dusty Baker SF,ChC 11 +18
Felipe Alou Mtl,SF 10 +13
Ron Gardenhire Min 2 +13
Pete Rose Cin 5 +12
Jim Frey KC,ChC 3 +11
Mike Hargrove Cle,Bal 12 +10
Art Howe Hou,Oak,NYM 13 +10
Jim Tracy LA 3 +9
Kevin Kennedy Tex,Bos 4 +8


And the worst:


Manager Team(s) Seasons +/-


Jimy Williams Tor,Bos,Hou 10 -27
Jim Riggleman SD,ChC 7 -23
Buddy Bell Det,Col 5 -21
Larry Dierker Hou 5 -16
Lee Elia ChC,Phi 4 -16
Phil Garner Mil,Det 10 -14
Doug Rader Tex,Cal 5 -13
Bud Harrelson NYM 2 -10
Gene LaMont ChW,Pit 7 -10
Tom Trebelhorn Mil,ChC 6 -9



The degree of causation is at least questionable, but some of the numbers do make strong implications. For example: Bobby Valentine, in 13 years of managing (a significant sample size) simply refuses to bring himself closer to the mean. In a thirteen year battle with Pythagoras - who is supposed to win as often as he loses - Valentine is beating him by 22 games. And then there's Jimy.
ARE YOU EXPERIENCED? and PYTHAGOREAN +/- | 25 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.
Coach - Friday, December 12 2003 @ 01:03 PM EST (#83659) #
Nice work, Leigh. The legion of Dusty-bashers may not approve, but this confirms my impression that he somehow squeezes out a few more wins than he \\"should\\" from some marginal talent. I've said before that Gardenhire, though he sure doesn't go by the stathead book, appears to keep his guys loose, happy and productive.

With Valentine, it's more of a clash of personal styles that makes me dislike him than a belief he's a \\"bad\\" manager. Bobby has always reminded me of Billy Martin with more self-control, or less self-destructiveness, however you want to express it. I think he quickly wears out his welcome with a lot of players, but he's a great choice to wake up an apathetic team. The Mike Keenan of baseball.

Jimy Williams comes as no surprise. Where does his mentor, the legendary Mr. Cox, rank vs. Pythagoras?
Leigh - Friday, December 12 2003 @ 01:13 PM EST (#83660) #
Where does his mentor, the legendary Mr. Cox, rank vs. Pythagoras?

Not included in my research because his managerial carer began in 1978 (I used 1980 as the cut-off), Bobby Cox is a +40 in 21 seasons. Only 3 times has a Bobby Cox team fell short of its Pythagorean W-L: Atl 79, Atl 97, Atl 01.
_SportsmanTO - Friday, December 12 2003 @ 01:36 PM EST (#83661) #
That's a pretty impressive study there. It really confirms some of my beliefs about managing today seeing how the most successful are those that had very good carers and those who never played. It confirms my beliefs that managers have to command some kind prescence with their players OR at least gain their respect by managing the personalities more than the actually game.

However this brings up a longstanding question that i've always had. What of the myth that catchers seem to be the best managers? Last season Tony Pena got great production out of the KC Royals and he was a fine catcher back in the day. It seemes that for the longest time I heard the theory that former catchers make great managers because they \\"see the whole field\\" and can manage a pitching staff well. Of course that didn't quite work with our pal Buck Martinez.

Anyone care to comment on this?

BTW Leigh thanks for the research, I too am suffering BWS. I just wish Spring Training would be tomorrow :((
Mike Green - Friday, December 12 2003 @ 01:42 PM EST (#83662) #
Very interesting work, Leigh. The long-term record vs. Pythagoras is especially illuminating for managers with 5 or more seasons. It'd be really interesting to have the carer lines for these managers when it comes time for \\"Manager in a Box\\".

I would have guessed that Managers, who as players had long carers and had consistently good strike zone judgment, would be more likely to be successful, as these provide an indication of intelligence and self-discipline that would be useful traits for a manager to have. Another interesting study would be success rate by defensive position-how do former catchers compare with former middle infielder and former outfielders?
_Spicol - Friday, December 12 2003 @ 01:43 PM EST (#83663) #
Suspicion confirmed, I suppose.

That's one conclusion. And your analysis is fun to read and think about. But it's unfounded.

are the winningest managers those who played well because there's some correlation between baseball skill and managerial skill? Or is it more likely that a high profile player won't demean himself by taking the helm of a last place team? Or is it that a front office will have more patience with a high profile player, allowing him to ride out a sub-par period until the team gets better and improves his overall record. Or is it that good players play for longer and for more managers and therefore have more time to be mentored?

Part II is bang on though. Good stuff.
Leigh - Friday, December 12 2003 @ 01:48 PM EST (#83664) #
But it's unfounded

I am pointing out the correlation between quality as a player and winning percentage as a manager. And it is there. I do agree that causation is nebulous at best.
_Spicol - Friday, December 12 2003 @ 01:54 PM EST (#83665) #
I do agree that causation is nebulous at best.

We can chalk it up to interesting but we can't actually DO anything with Part I. In other words, we can't use it to conclude that Ozzie Guillen will be a bad manager. I noticed that you were careful not to make any conclusions but you were headed down that street.
Leigh - Friday, December 12 2003 @ 01:57 PM EST (#83666) #
We can chalk it up to interesting but we can't actually DO anything with Part I

Call it cognative disonance, I guess, but I included the Part I stuff because, well, I researched it. I realized while doing it that it doesn't really say much because there is too much noise.

Part II was discovered serendipitously while researching Part I. I like Part II much better.
_Rej - Friday, December 12 2003 @ 02:46 PM EST (#83667) #
http://www.diamond-mind.com/articles/james_onerun.htm
Just wanted to point out some work Bill James did on managers and one-run games. I hope the homepage link works.

His final conclusion in this piece (I don't know how to do the italics or bold here - help?):

\\"Can one infer anything about a manager from his one-run record?
I would have guessed, going into this study, that the answer to that might be a flat \\"no\\", or, at least, an equivocal \\"no\\" (we can find no evidence within our study that playing well in one-run games is anything but a random occurrence, etc., etc., yada yada yada, snore.) I can't give you that answer, for two reasons:

1) There does seem to be some persistent tendency of teams to play poorly in one-run games, and

2) Teams which play well in one-run games do seem to have some identifiable characteristics, to a small degree.

But I will say this: that I would be careful about drawing any such inferences. Tony Muser is -15 games in one-run decisions. I can't say that this IS just coincindence -- but it certainly could be. It's not an overwhelming number, in and of itself.

Rany began this discussion with a comment about Bobby Cox' relatively poor record in one-run games. Well, from 1990 through 2001 the Atlanta Braves scored 8,836 runs, allowed 7,409. This is a ratio of 1.19 to 1. In one-run games they have gone 297-256, a ratio of 1.16 to 1.

The Braves have missed their expected won-lost record in one-run games, over the ten years, by 2.1 wins. Obviously, no conclusion of any kind can be drawn from such an occurrence. One-run games involve a huge amount of luck. This may be the only safe statement that can be made about them.\\"
Mike D - Friday, December 12 2003 @ 03:14 PM EST (#83668) #
Part II is \\"the show\\" here, as far as I'm concerned. If that could be looked at more in-depth somehow, I think we'd really get some fruitful results.

Part I doesn't make much intuitive sense, in my opinion. First of all, OPS+ is an imperfect indicator of overall baseball ability -- and an irrelevant indicator of on-field intelligence and instincts.

How many players have a high OPS+, despite playing dumb baseball and minimizing their immense natural ability? Similarly, how many players overcome a lack of talent to etch out a big-league carer with smarts? Plus, OPS+ weights power rather heavily and ignores defence. Was Ozzie Smith a \\"poor\\" player that therefore wouldn't be a good managerial mind? Should Manny Ramirez or Jim Thome start working on their managerial resumes?

Finally, wins don't mean much. Lou Piniella did a terrific job in keeping the D-Rays to under 100 losses; Jerry Manuel led a talented White Sox club straight into the ground with a mediocre, albeit \\"winning,\\" record. Part II is much more valuable, and Bobby Valentine, from the dreaded \\"Group B,\\" has an excellent record vs. Pythagoras.

In any event, I don't think the correlation is that significant.

Part II, on the other hand, raises a variety of interesting questions. Can we somehow get an even better proxy for in-game managerial ability? I think this is pretty telling. Correct me if I'm wrong, Leigh, but the practical application of an excellent record vs. Pythagoras seems to be an ability to win close games at a rate beyond that which luck would predict.

Meanwhile, on Coach's point about Dusty maximizing talent: It would be interesting to see how managers' clubs performed relative to their PECOTA Pythagorean record (if such a calculation even exists). That would speak more to a manager's ability to get the best possible performances out of lightly-regarded talent (think '91 Twins).

In other words, how did managers do with teams relative to their expected record ex ante (before the season)? Coach, I'm not sure that Dusty's Pythagorean record supports his reputation (which I think he deserves) of getting players to play above their heads. Part II doesn't analyze whether teams played beyond their expected limitations; instead, it seems to simply talk about how efficiently managers converted the performances they actually got, in retrospect, into wins. Hence Jimy Williams in last place.

One more thing: Leigh, this is a tremendous research effort! Very interesting.
Leigh - Friday, December 12 2003 @ 03:28 PM EST (#83669) #
Exactly, Mike D.

Part I is fluff, like I said, I probably included to justify having done the research.

Part II, however, is good, I think. It's not about the quantity of playing time (as that is indicated in the run differential), it is about the quality of playing time. Not simply playing the right guys, but playing the right guys at the right time.
Pistol - Friday, December 12 2003 @ 09:51 PM EST (#83670) #
I think I may have ranked the managers by wins/year over or under PY.

Looking at those managers that exceed the PY, with the exception of Pete Rose at 2.4 (who would have bet that....), no manager with more than 3 years in that study is averaging more than 2 wins/season greater than PY.

On the other hand there were several managers that were 4 and 5 games each year worse than PY.

So I might conclude that no matter how good of a manager you are you won't exceed your PY by anything significant, but if you're a bad manager you can really cost your teams wins.
_Nick G - Friday, December 12 2003 @ 10:14 PM EST (#83671) #
IMHO, comparing pythag W/L to W/L can tell us very little about how good a manger is. I think there is the mistaken assumption by some that a pythag W/L which is greater than an actual W/L indicates that a manager is playing the better players over the inferior ones. This isn't the case. If Dusty Baker throws Estes out there everyday, that results in a high runs allowed, which is reflected both in real W/L and pythag W/L.

I think a case could be made that Dusty Baker outperforms his pythag precisely because of his weakness for playing inferior players over superior players. Thus, in high leverage situations (precisely the point at which one can increase the difference between one's pythag W/L and real W/L), he has good players like Mark Bellhorn on the bench to pinch hit (or course, the argument against this is that Dusty's benches have often contained nothing but refuse like Dunston.)

Anyway, a better indication of a manger's skill I think would be to take something like W/L based on PECOTA forecasts and compare them to final W/L.
_Nick G - Friday, December 12 2003 @ 10:16 PM EST (#83672) #
Ah, just noticed Mike D already said pretty much what I wanted to say in his last couple of paragraphs.
Leigh - Friday, December 12 2003 @ 10:24 PM EST (#83673) #
I think I may have ranked the managers by wins/year over or under PY.

That was my original plan, but I decided to go instead with the cumulative numbers because they better represent the extent to which the manager is good at beating Pythagorus.

For example, here are Bobby Cox' 21 seasons:
+4,-4,+4,0,+2,+1,+2,0,+2,+4,0,+1,+6,+2,-2,0,+5,+5,-2,+5,+5 = +40

And here are Hal Lanier's 3 seasons (1986-1988 Hou):
+5,-2,+3 = +6

If I use the wins/year over or under PY, I get Lanier at 2.0 and Cox at 1.9.

I think that Lanier's success could be a fluke, whereas Cox' appears to be more substantive. This is reflected in the cumulative totals, which is why I chose to use them.
Leigh - Friday, December 12 2003 @ 10:28 PM EST (#83674) #
I think there is the mistaken assumption by some that a pythag W/L which is greater than an actual W/L indicates that a manager is playing the better players over the inferior ones.

I am well aware of this. And so is my report. I am saying that getting beaten by Pythagorus, consistently, might indicate that the wrong players are being played at the wrong time. It's not about the quantity of playing time, but the timeliness of it.
Leigh - Friday, December 12 2003 @ 10:28 PM EST (#83675) #
I think there is the mistaken assumption by some that a pythag W/L which is greater than an actual W/L indicates that a manager is playing the better players over the inferior ones.

I am well aware of this. And so is my report. I am saying that getting beaten by Pythagorus, consistently, might indicate that the wrong players are being played at the wrong time. It's not about the quantity of playing time, but the timeliness of it.
_Xander - Friday, December 12 2003 @ 10:36 PM EST (#83676) #
Damn...you guys and your stats...it might be a cause for concern...

But I do appreciate your love for the game.

Cheers
_Mick - Saturday, December 13 2003 @ 04:01 AM EST (#83677) #
Leigh, I think you should send this URL and a pointer specifically to Part II to Neyer, who would love it. Then you should write it up (or ask Neyer or some other \\"name\\" to write it up using your data and give you co-authorship) for a SABR convention presentation or journal article.

What might be even more interesting is to expand your data to 1977-2001 (25 years) and do the same Part II study on the preceding 25 years (1952-1976) and see if it's changed.

The split is not random -- it's at the advent of free agency when salary structures really started to warp -- and the quarter-centuries would give you a significant sample size.
_Mick - Saturday, December 13 2003 @ 04:04 AM EST (#83678) #
On a separate note, I'm just gonna point this out with very little comment and let it sit here as possibly a second article.

Baseball's traditional oldwhiteguy cronyism begs the point that there are only two minorities in the two lists of 10 -- Baker and Alou. And they both are in the Top 3 of the good list.
_Brian - Saturday, December 13 2003 @ 05:26 AM EST (#83679) #
This really piqued my interest, as I had done a few things with Pythagorean Win% about a year ago. So I hit the Lahman DB, updated it with the 2003 numbers and started trying to replicate your results - which I was able to do, except for...

No team managed by Jimy Williams has ever reached its Pythagorean W-L

The 1999 Red Sox were 94-68, while Pythagoras pegged them at 92-70.

His total should be -23, which puts him in a dead heat with Riggleman. I went back to 1920 with the data and (using the same criteria you did) they are tied for 3rd worst since then, behind Bucky Harris (-35) and Jimmie \\"Ace\\" Wilson (-33).

It's probably worth noting that the +40 posted by the Imitable Mr. Cox is the highest cumulative total of any manager since 1920.
Leigh - Saturday, December 13 2003 @ 10:10 AM EST (#83680) #
you are right about Williams, Brian. I did it manually (no spreadsheet); I guess I missed that one. Except that you say he should now be -23. I think he should now be -25.
_Brian - Saturday, December 13 2003 @ 01:07 PM EST (#83681) #
For Williams I have:
1986 TOR -2
1987 TOR -4
1988 TOR -3
1989 TOR 0 (only managed 36 games)
1997 BOS -2
1998 BOS -2
1999 BOS 2
2000 BOS -1
2001 BOS -1
2002 HOU -3
2003 HOU -7
_Nick G - Saturday, December 13 2003 @ 01:41 PM EST (#83682) #
The thing which makes Jimy Williams' stunning lack of success even more stunning is that on most of the teams he's managed he's had a very good bullpen. With Jays he had either an Eichorn/Henke or Ward/Henke combination. With the Sox he had Gordon and Lowe. And with the Astros Lidge and Wagner.
Leigh - Saturday, December 13 2003 @ 02:11 PM EST (#83683) #
For Williams I have:
1986 TOR -2
1987 TOR -4
1988 TOR -3
1989 TOR 0 (only managed 36 games)
1997 BOS -2
1998 BOS -2
1999 BOS 2
2000 BOS -1
2001 BOS -1
2002 HOU -3
2003 HOU -7


That's what I have too. The total is -25.
ARE YOU EXPERIENCED? and PYTHAGOREAN +/- | 25 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.