Batter's Box Interactive Magazine Batter's Box Interactive Magazine Batter's Box Interactive Magazine
You've all noticed that the Orioles have played .750 ball (27-9) in one-run games this season?


That happens to be the best winning percentage in one-run games... well, ever. Since we crawled out of the sea and rose up on our hind legs. The 2012 Orioles have played 60-56 in the rest of their games; it's their record in the one-run games that is lifting them, and not Tampa Bay, to the post-season.

What the 2012 Orioles have done is not the most dramatic spread above and beyond normal performance in history. If .500 is the centre, and surely it is, by winning 18 times more than they've lost the Orioles have added 9 games to the Win column above what could reasonably be expected. That's a very, very good thing. But it's not unprecedented. The greatest such single-season pickup in one-run games was posted by another Orioles squad, Earl Weaver's 1970 team. They went an amazing 40-15 in one run games - they played .727 ball and won 12.5 more games than you would reasonably expect . Not that it had a huge impact on the pennant race.  The 1970 Orioles went 68-39 (.635) in the rest of their games and finished 15 games ahead of the second place Yankees. Man, they were good.

By eerie coincidence, it was yet another Baltimore team that had the best winning percentage in one-run games prior to this season. That would have been Earl Weaver's 1981 crew. They went 21-7 in one-run games, which is a pretty nifty .750 winning percentage. It wasn't enough to get them into either of the AL East's post-season rounds, largely because they went 38-39 in the rest of their games

Anyway, all this is a tribute to the genius of Buck Showalter, surely? And Earl Weaver?

Except over the course of his career, Weaver's teams did not win as often in one-run games as they did the rest of the time. Weaver's teams had two of the greatest one-run seasons in history, and over his career his teams played .580 ball in one-run games, which is astonishingly good. I think it's probably the best mark by any manager in history. But it's still not as good as the .589 mark Weaver's Orioles posted in the rest of their games.

And Showalter? Thanks to the amazing performance of the 2012 Orioles, Buck Showalter's teams have now played .494 ball in one run games over the course of his managerial career. Which isn't as good as the .518 record they have in the rest of their games.

Who else? Joe Maddon and Davey Johnson? We all think highly of them. I know I do. Same deal, though. Maddon's teams have played .509 ball in one-run games, .516 the rest of the time. Johnson's teams have played .543 ball in one-run games, and .590 ball in the rest of them.

I gathered together the relevant numbers for 15 managers, more or less at random - well, not quite at random. First I took the top 11 winners of all-time. I kicked out Bucky Harris because... because he's Bucky Harris. I replaced him with Weaver. I then added three interesting active managers (Johnson, Showalter, and Maddon) as well as Cito Gaston just to provide some local flavour.

And here's how these men did in one-run games as opposed to the rest of their games. (NOTE - numbers are not quite complete. I only included complete seasons. If I got into the nuts and bolts of those seasons where the guy managed 64 games or 111 games... I'd be doing this until Halloween. And that's why Billy Martin isn't here, because most of Billy's seasons were like that.)

Anyway....

                 One Run Games                  Other Games
               W       L     PCT           W      L      PCT

Joe McCarthy    520    452    .535        1556    840    .649
John McGraw     749    623    .546        1915    1227    .609
Davey Johnson   342    288    .543        825    573    .590
Earl Weaver     429    311    .580        950    663    .589
Bobby Cox       681    629    .520        1783    1315    .576
Walter Alston   634    537    .541        1408    1078    .566
Tony LaRussa    697    704    .498        1933    1562    .553
Leo Durocher    532    477    .527        1338    1098    .549
Joe Torre       634    553    .534        1599    1315    .549
Sparky Anderson 590    502    .540        1548    1282    .547
Casey Stengel   502    524    .489        1325    1194    .526
Buck Showalter  269    275    .494        769     716    .518
Joe Maddon      163    157    .509         415     390    .516
Cito Gaston     225    235    .489         564     531    .515
Connie Mack    1067   1156    .480        2560    2735    .483


As you can see, everybody's teams played worse in one-run games than they did the rest of time. Everybody's.  I thought Connie Mack might be the exception. Mack was the only one of these men with a losing record over his career, and I thought his teams might have been slightly better in one-run games. But even they were slightly worse.

The most impressive performance, by a mile is Weaver's. But I think the work of Sparky Anderson, Walter Alston, and Joe Torre in one-run games probably deserves your respect as well.
Managers and One Run Games | 15 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.
ayjackson - Monday, September 24 2012 @ 08:12 PM EDT (#263780) #
So, Maggie, baby, doll, does winning percentage in one-run games for all managers form a normal distribution around .500?
Magpie - Monday, September 24 2012 @ 09:03 PM EDT (#263783) #
does winning percentage in one-run games for all managers form a normal distribution around .500?

Of course. If you're looking at all managers, it's always .500 for all games. Doesn't matter whether it's one-run games or blowouts. (Not sure what you meant there.)

What is predictable is one-run games dragging everyone towards .500. I f you take twenty teams that play .300 ball in the rest of their games, they will play about .350 ball in one run-games; twenty teams that play .400 ball the rest of the time will play something like .435 in one run games; teams that play .450 ball the rest of the time will play something like .470 ball in one-run games. Note that the spread between one-run games and the rest narrows as you get closer to .500 - and of course the exact same process occurs in reverse for teams that are better than .500.
robertdudek - Tuesday, September 25 2012 @ 07:12 AM EDT (#263787) #
Of course. If you're looking at all managers, it's always .500 for all games. Doesn't matter whether it's one-run games or blowouts. (Not sure what you meant there.)

I think what ay is asking is - is manager winning percentage normally distributed (i.e. the bell curve). That is - are there an equal number  of bad managers one standard deviation below .500 as there are one standard deviation above? Are there fat tails (e.g. a larger than expected cluster of very good managers or a large cluster of very bad managers)?

Without looking at the data, my own prediction is that the data will not be "normally" distributed, as managers who are very bad tend to lose their jobs and therefore tend not to be included in the data.

robertdudek - Tuesday, September 25 2012 @ 07:16 AM EDT (#263788) #
I thought Connie Mack might be the exception. Mack was the only one of these men with a losing record over his career, and I thought his teams might have been slightly better in one-run games. But even they were slightly worse.

In a sense Mack IS the exception on the list - as his teams are below average, we would expect them to be better at one-run games than other games.
rtcaino - Tuesday, September 25 2012 @ 11:46 AM EDT (#263805) #

Interesting piece, Magpie.

 

Earl Weaver did well, it seems, to keep his record in one run games so close to his high winning percentage in other games.  Of course, everyone else probably know that already.

Magpie - Tuesday, September 25 2012 @ 03:59 PM EDT (#263841) #
as managers who are very bad tend to lose their jobs and therefore tend not to be included in the data.

That's definitely true! (Although mostly I just didn't understand the question.) There are 125 men who have managed at least 1000 games in the major leagues. Of that group, just 48 have records below .400 (the top 48 are all above .520). And the group that was below .500 includes Mack, Tom Kelly, Bill Carrigan, Bucky Harris - who all won multiple championships - as well as Gil Hodges, Chuck Tanner, Frank Robinson, who all had some success. What I expect you'll find with managers (I've been too busy the last couple of days to really look, I'm afraid) is a large pool of men who had losing records and didn't make it to 1000 games. Which actually is the normal bell curve distribution of things in baseball - only the extreme high end of the curve is in play, and the largest group consists of the least accomplished.

We might consider Gene Mauch a little. Someone who is not regarded as a bad manager, at all, but his teams did lose more than they won. And as it turns out, Mauch's teams played better in one-run games (.509) than they did in the rest of them (.481.)

Connie Mack's career in Philadelphia was so long that there are several careers, which cover the great dynasties he built (and dismantled), and the old, old man presiding over hopeleslly bad teams at the end. Mack's 15 best teams played .630 ball overall - but in one-run games they played .566 ball and in the rest of the games they played .656 ball; and his 15 worst teams (.325 overall) predictably follow the opposite pattern. Those teams played .388 ball in one-run games, and .298 the rest of the time. The twenty teams in between were slighlty below average (.484) and they did slightly better in one-run games (.487 to .483)
ayjackson - Wednesday, September 26 2012 @ 12:05 AM EDT (#263859) #

I guess it's not really about normal distributions, but I'm wondering if Weaver is good or weather his record could be random.  I don't know if distance from the mean (in standard deviations) would tell me that or not.

robertdudek - Wednesday, September 26 2012 @ 06:53 AM EDT (#263862) #
I'll let someone else run the numbers, but given that we should expect about a .540 WPCT for Weaver in one-run games, I think it's a fair bet that there is something going on here other than random fluctuation.
robertdudek - Wednesday, September 26 2012 @ 06:58 AM EDT (#263863) #
By the way, Bill James wrote about one-run games and he claimed that a good estimate for record in one run games expectation was RS/(RS+RA), or  simply, the Pythagorean WPCT with the exponent set to 1.
Beyonder - Wednesday, September 26 2012 @ 09:32 AM EDT (#263866) #
"There are 125 men who have managed at least 1000 games in the major leagues. Of that group, just 48 have records below .400 (the top 48 are all above .520)."

Judging from the next sentence, I think you mean .500 right Magpie?

Maybe this is just the flip side of the same point, but it should not be surprising that there are few "very bad" managers who have managed 1000+ games. This follows not just from the fact that bad managers get weeded out, but also from regression to the mean. The more games you manage, the centrifugal pull of .500 gets stronger.
Magpie - Wednesday, September 26 2012 @ 03:42 PM EDT (#263902) #
Judging from the next sentence, I think you mean .500 right Magpie?

Nope, the top 48 are above .520 - and as there are 48 men below .500 it leaves 29 in the middle, all between .500 and .520.
Magpie - Wednesday, September 26 2012 @ 04:11 PM EDT (#263904) #
How about we set the bar in different places and see how the groups sort themselves out?

If you set the bar at 250 games or more, the median winning percentage (number 178 of a group of 336) is .493, and the groups break down as follows:.

.550 or better: 41 (12.2.%)
.500 to .549: 124 (36.9%)
.450 to .499: 98 (29.1%)
.449 or lower: 73 (21.7%)

If you set the bar at 500 games, the median winning percentage (number 117 of a group of 233) is .504, and the groups break down as follows:

.550 or better: 33 (14.2.%)
.500 to .549: 95 (40.8%)
.450 to .499: 68 (29.2%)
.449 or lower: 37 (15.9%)


If you set the bar at 1000 games, the median winning percentage (number 63 of a group of 125) is .508, and the groups break down as follows:

.550 or better: 22 (17.6.%)
.500 to .549: 55 (44.0%)
.450 to .499: 40 (32.0%)
.449 or lower: 8 (6.4%)


If you set the bar at 1500 games, the median winning percentage (number 39 of a group of 77) is .517, and the groups break down as follows:

.550 or better: 17 (22.1.%)
.500 to .549: 41 (53.2%)
.450 to .499: 18 (23.4%)
.449 or lower: 1 (1.3%)

That's pretty much what you'd expect, no?
Beyonder - Wednesday, September 26 2012 @ 04:17 PM EDT (#263905) #
Sorry Magpie. Either I wasn't clear or I don't understand you. There are no managers among the 125 you mention with win-loss records below .400. The closest is Jimmy Wilson at .401? Are we talking about something different?
Magpie - Wednesday, September 26 2012 @ 05:26 PM EDT (#263907) #
There are no managers among the 125 you mention with win-loss records below .400. The closest is Jimmy Wilson at .401?

That's right - Wilson has the lowest winning percentage of anyone's who managed at least 1000 career games. He's one of eight below .450

No one with a winning percentage below .400 has been allowed to manage 1000 games; very few have even been allowed to manage 500 games. Of the sub .400 managers, Preston Gomez leads the way with 875 games. The others above 500 games are Zack Taylor, Art Fletcher, Fred Tenney, and John McCloskey. (Roy Hartsfield just missed!) With the exception of Gomez, who was San Diego's original manager and last managed in 1980, all of those guys worked a long time ago. The most recent of the others managed the St. Louis Browns, which should give you an idea....
Magpie - Wednesday, September 26 2012 @ 05:29 PM EDT (#263908) #
The more games you manage, the centrifugal pull of .500 gets stronger.

That would probably be true, except that it's the guys who are best able to resist that pull who get to keep working.
Managers and One Run Games | 15 comments | Create New Account
The following comments are owned by whomever posted them. This site is not responsible for what they say.