Username:	Password:

Remember me

Forgot your username/password?

Print Email

Home>Articles

A Doc, Two Dons, and a Moose

By Dave Fleming

May 27, 2014

There was a discussion over in Readers Post land about Roy Halladay’s candidacy for the Hall of Fame. In the discussion, two interesting player comparisons came up.

The first was brought to us via izzy24, who I presume is former Red Sox farmhand and sometimes-judo-expert Izzy Alcantara. Izzy24 writes:

"I see Halladay as a better version of Drysdale. That sure sounds like a Hall of Famer to me."

What’s funny about Izzy’s comparison is that Don Drysdale is frequently referecenced as someone who lowers the Hall-of-Fame standards, someone who maybe shouldn’t be in Cooperstown.

A little while later we got this one, from Evanecurb:

"I’d say Mussina was the Don Sutton of his era. Never the best in the game, but a consistent 30 start, 15 win guy for a long time."

Evan quickly recanted this statement, and I don’t want to seem like I’m mocking either of these posts. Judged by their Triple Crown lines, these comparisons make some sense:

Name	Wins	ERA	K
Drysdale	209	2.95	2486
Halladay	203	3.38	2117

Not dissimiliar. Hallady has the higher ERA, but him and Drysdale pitched in different eras. Doc and Big-D share a few other parallels: both are tall right-handers (Drysdale was 6’5", Halladay 6’6"); both were famous pitchers in their primes; both lost their effectiveness in their early thirties.

Mussina and Sutton are a bit alike, too:

Name	Wins	ERA	K
Sutton	324	3.26	3534
Mussina	270	3.68	2813

Sutton has the longer career, but both pitchers won an impressive amount of games. Neither won a Cy Young Award, or came particularly close (Mussina finished second in 1999, but it wasn’t exactly a close vote). They share the same ‘knock’ from baseball traditionalists: they each won 20 games just once in their careers.

But these comparable aren’t that convincing….using wins, strikeouts, and ERA misses an important bit:

Name	Wins	Losses	Win %
Drysdale	209	166	.557
Halladay	203	105	.659

Name	Wins	Losses	Win %
Sutton	324	256	.559
Mussina	270	153	.638

Don Drysdale, pitching on generally good teams in a tough division, won 56% of his starts. Halladay, pitching on slightly less good teams, won 66% of his starts. That’s a significant difference. The same holds true for Sutton and Mussina: while they won an equivalent number of games each year, Mussina was losing fewer games each year than Sutton.

Winning percentage and WAR are in agreement about Mussina versus Sutton…both measures show a sizeable gap in their talent level:

Name	W-L	Win %	rWAR
Mussina	270-153	.638	82.7
Sutton	324-256	.559	68.7

But in the case of Don Drysdale and Roy Halladay, their winning percentage actually does a better job of communicating a difference between the two players than their cumulative WAR does. Which seems a more accurate comparison of Drysdale and Halladay?

This:

Name	rWAR
Drysdale	61.2
Halladay	65.6

Or this:

Name	W-L	Win %
Drysdale	209-166	.557
Halladay	203-205	.659

When I think about these two pitchers, I expect Halladay to rate ahead of Drysdale. That’s not a knock against Drysdale: in his prime, Big D was a fine pitcher. But I don’t know that anyone was rating him ahead of Koufax, Marichal, or Gibson.

I remember all of Halladay career. For a long time, he was in the conversation as the best pitcher in the game. His winning percentage communicates this better than his cumulative WAR.

We’ll come back to these four…

* * *

Okay…down the rabbit hole. Did you know Baseball-Reference lists a split for pitcher’s statistics, based on whether their opponent was better or worse than .500?

This split is really interesting. Let’s start with Jack Morris.

The case for Jack Morris getting elected to Cooperstown is that he won a lot of games. The case against him is that he wasn’t all that good: he was just lucky to have good teammates.

Let’s check the splits:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Jack Morris	254-186	158-66	.705	96-120	.444

Jack Morris was extremely good against bad teams. And he couldn’t beat good teams. He had a losing record against good teams, and it wasn’t too close to .500.

I think, incidentally, that this is a really good way to make your case against Morris. If you’re arguing this in a bar, citing his career WAR isn’t going to convince too many people loyal to older metrics. This split might: most of the time Morris faced a good opponent, he lost that game.

But we’re not trying to win bar arguments here. Morris’s winning percentages against winning teams is .444. I’m not sure what that means, exactly….I don’t know whether that’s a really bad tally or somewhat expected. We need a bit of context to understand the split.

Checking, first, on the recent 300-game winners:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Clemens	354-184	216-73	.747	138-111	.554
R. Johnson	303-166	180-75	.706	123-91	.575
Maddux	355-227	193-94	.672	162-133	.549
Glavine	305-203	159-87	.646	146-116	.557

These four are surprisingly consistent. Jack Morris’s winning percentage against losing teams falls in the middle of this group. But unlike Morris, all four of the 300-game winners posted a positive winning percentage against winning teams.

A couple more recent guys:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Smoltz	213-155	116-69	.627	97-86	.530
Schilling	216-146	122-61	.667	94-85	.525

An obvious pair, and we’re starting to see some ranges. Schilling and Smoltz weren’t quite up to the standards of Clemens, Johnson, and Maddux, against winning or losing teams. They’re closer to Glavine.

Getting older:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Carlton	329-244	175-109	.616	154-135	.533
Seaver	311-205	164-80	.672	147-125	.540

Here’s an early hypothesis about this split: a player’s win-loss record against losing teams tells us a lot about the quality of their team. A player’s win-loss record against winning teams tells us a lot about their ability.

Steve Carlton played on some bad teams…so his winning percentage against losing teams isn’t as impressive as Seaver’s. They were equally effective against good teams.

I have no idea why that would turn out to be true, but it seems to fit. Seaver and Carlton had similar winning percentages against +.500 teams, but different winning percentages against losing teams.

Here’s another one:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Palmer	268-152	159-62	.719	109-90	.548
Jenkins	284-226	123-97	.559	161-129	.555

Palmer played on better teams than Jenkins. Gentleman Jim crushed the losing teams, but his winning percentage against tough competition is dead-even with Fergie Jenkins.

We can calculate, easily enough, the percentage of decisions each pitcher got against winning and losing competition. Palmer had 420 career decisions…199 of those came against winning teams. That’s 47%...most of Palmer’s decisions came against losers.

Fergie Jenkins flips it: 57% of his decisions came against winning teams.

Another theory…pitchers on bad teams tend to have less of a split in their winning percentages than players on good teams.

Two more big-win pitchers:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Niekro	318-274	144-114	.558	174-160	.521
Perry	314-265	150-93	.617	164-172	.488

Niekro and Perry, like Jenkins, got most of their decisions against .500+ teams…56% for Niekro, 58% for Perry. This supports the second theory: Niekro, like Jenkins, has comparable winning percentages against winning and losing teams…just a 37-point difference.

Gaylord Perry is our first ‘surprise’…a ‘great’ pitcher who doesn’t have a .500 winning percentage against winning teams. Bill rated Perry as the 18^th best pitcher in baseball history the last time he crunched numbers, so this is a bit surprising.

Finishing up this generation:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Ryan	324-292	149-104	.589	175-188	.482
Blyleven	287-250	157-101	.609	130-149	.466

Another logical pair, Blyleven and Ryan have very close winning percentage splits. Ryan faced tougher competition: 59% of his decision came against winning teams, compared to 52% for Blyleven. We now have three Hall-of-Famers who have losing records against winning teams.

Alright…two more.

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Tommy John	288-231	149-105	.587	139-126	.525
Jim Kaat	283-237	161-109	.596	122-128	.488

Tommy John, rated higher by rWAR, has a better winning percentage against winning teams than Phil Niekro, Gaylord Perry, Nolan Ryan, Bert Blyleven, and Jim Kaat. That’s interesting, right?

Going further back:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Marichal	243-142	113-65	.669	130-86	.602
Gibson	251-174	129-63	.672	122-111	.524

Another surprise: Juan Marichal had a better record against steep competition than Bob Gibson.

Actually, the Dominican Dandy is ahead of everyone I’ve listed, when it comes to beating winners. Randy Johnson (.575) had the best mark, but Marichal is ahead of him.

Bob Gibson’s splits (.672/.524) are a lot like Curt Schilling’s splits (.667/.525). Two of the most famous postseason pitchers in baseball history show a strong similarity in their winning percentage splits.

We forgot about someone….Sandy. We’ll pair Koufax with another Dominican:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Koufax	165-87	81-42	.659	84-45	.651
Pedro	219-100	124-38	.765	95-62	.605

Joe Posnanski recently wrote that if the Devil gave him one pitcher to play for his soul, he’d take Pedro. If the Devil puts together a .500+ team, I might take Sandy instead. His winning percentage against winning teams is the best of any pitcher I could find.

Koufax might lend credence to my theory about winning percentages against losing teams reflecting a pitcher’s team, whereas winning percentages against winning teams reflects a pitcher’s talent:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Drysdale	209-166	100-55	.645	109-111	.495
Koufax	165-87	81-42	.659	84-45	.651

Drysdale and Koufax won at the same pace against losing teams, but Koufax kept pace against the winning teams. Drysdale didn’t.

Going further back:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Feller	266-162	139-52	.728	127-110	.536
Spahn	363-245	205-97	.679	158-148	.516
Roberts	286-245	158-99	.615	128-146	.467

There’s an iceberg of data to play with in this split, and I’ve only approached the tip of it. Looking at Feller and Roberts, you’d conclude that Bob Feller is a ‘great’ pitcher, while Roberts is the lesser pitcher. Feller’s percentages parallel Jim Palmer’s (.719/.548)…Roberts looks a lot like Blyleven (.609/.466).

But here’s a weird extension of those splits:

Name	ERA <.500	ERA .500+
Feller	2.72	3.75
Roberts	3.30	3.50

Feller gave up more runs to good opponents…an additional run for every nine innings pitched. Roberts had less variance against his opponents. I don’t know which is better, frankly…whether it’s more impressive to give up 3.4 runs to everyone, or to beat the tar out of the lesser teams.

Going back further:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Grove	300-141	162-59	.733	138-82	.627
Hubbell	253-154	126-49	.720	127-105	.547

Grove’s winning percentage against sub-.500 teams rates behind just four pitchers that I found (Pedro, Clemens, and two others I haven’t mentioned. His winning percentage against winning teams is second behind Koufax.

If we wanted to group players by families according to their winning percentages split against winning and losing teams, Hubbell belongs with Palmer and Feller:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Hubbell	253-154	126-49	.720	127-105	.547
Palmer	268-152	159-62	.719	109-90	.548
Feller	266-162	139-52	.728	127-110	.536

All three pitchers share a good taste in teammates…Feller’s Indians are perhaps the least known good team, but they were excellent from 1948 to 1956 (Feller’s last season), and missed the pennant by a single game in 1940.

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
W. Johnson	417-279	126-73	.633	140-109	.562
G. Alexander	373-208	171-64	.728	133-106	.556

We only get partial splits for Walter Johnson and Grover Cleveland Alexander, and it’s Old Pete who comes out a hair ahead. Christy and Cy didn’t have enough games to warrant a listing.

* * *

Some special cases now…

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Rick Reuschel	214-191	109-73	.599	105-118	.471
Nolan Ryan	324-292	149-104	.589	175-188	.482
Bert Blyleven	287-250	157-101	.609	130-149	.466
Jim Kaat	283-237	161-109	.596	122-128	.488

WAR-darling Rick Reuschel rates comparably to Nolan Ryan, Blyleven, and Kaat. Not sure it helps his case, really....he doesn’t have the longevity of those other players.

On the other hand, the thinking man’s favorite pitcher of the 1980’s, Dave Stieb, finds a surprising ally:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Jenkins	284-226	123-97	.559	161-129	.555
Stieb	176-137	86-60	.589	90-77	.539

One pitched in Canada, one was Canadian.

Fergie Jenkins has the narrowest ‘gap’ between his winning percentage against sub-.500 teams, and his winning percentage against .500+ teams....here are the top-five:

Name	W-L <.500	Win %	Difference
Jenkins	0.559	0.555	4
Koufax	0.659	0.651	8
Niekro	0.558	0.521	37
Stieb	0.589	0.539	60
Kaat	0.596	0.488	62

And the biggest gaps:

Name	W-L <.500	Win %	Difference
J. Morris	0.770	0.500	261
Clemens	0.705	0.444	193
Feller	0.747	0.554	192
Hubbell	0.720	0.547	173
Palmer	0.720	0.547	171
Alexander	0.719	0.548	171

Jack Morris has the biggest discrepancy in winning percentage between sub-.500 teams and .500+ teams of any retired player I’ve found. There’s one active player ahead of him.

These lists seem…scattered. There’s no obvious pattern that I can see. Moving on.

Toughest competition:

Name	% of Decisions, .500+
Ryan	59%
Drysdale	59%
Perry	58%
Hubbell	57%
Jenkins	57%
Niekro	56%
W. Johnson	56%

The obvious pattern is that these pitchers played on sub-par teams. Drysdale and Hubbell are the exceptions…maybe the Dodgers saved Drysdale for tougher opponents. I expect the Giants did that with ‘Meal Ticket’ Hubbell.

Name	% of Decisions, .500+
R. Johnson	46%
Clemens	46%
Palmer	47%
Mussina	48%
Jim Kaat	48%
Morris	49%
P. Martinez	49%
Schilling	49%

A lot of modern players…the variance between pitching to good/bad teams has declined significantly as baseball schedules have expanded, and the five-man rotation has taken hold. Teams are no longer juggling the rotation to get their best pitchers up against the best teams.

* * *

Looking at two active players approaching 200 wins:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Bartolo Colon	192-133	110-60	.647	82-73	.529
Bob Gibson	251-174	129-63	.672	122-111	.524
Mark Buehrle	193-143	104-57	.646	89-86	.509

No one really talks about Colon or Mark Buehrle as Hall-of-Fame bound. It’s interesting that, at least by this split, they’re not dissimiliar to Bob Gibson.

And the two active players with 200+ victories:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Sabathia	208-119	127-38	0.770	81-81	0.500
Hudson	209-113	121-48	0.716	88-65	0.575

Sabathia, surprisingly, is the active pitcher making a charge at Jack Morris….he has the biggest difference in split winning percentages of any pitcher I looked at.

As for Tim Hudson....his kin is surprising:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Hudson	209-113	121-48	0.716	88-65	0.575
R. Johnson	303-166	180-75	0.706	123-91	0.575

Fun to come at the end of an exercise with Tim Hudson and Randy Johnson looking practically identical.

* * *

We started with two comparables…coming back to them one at a time:

Pitcher	Career W-L	W-L <.500	%	W-L .500+	%
Drysdale	209-166	100-55	.645	109-111	.495
Halladay	203-105	108-37	.745	95-68	.583

Halladay has an 100 point edge in winning percentage in both splits, setting him out of Drysdale’s company. Halladay’s closest compatriot is actually Pedro Martinez ( .765/.605)….Pedro is ahead on both counts, but it’s close.

Sutton and Mussina doesn’t fit, either:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Sutton	324-256	172-105	.621	152-151	.502
Mussina	270-153	151-71	.680	119-82	.592

Don Sutton’s closest compatriot is Gaylord Perry. He beat the bad teams, but was about 50-50 against the good ones.

And Mike Mussina, apitcher almost never mentions as a big-game pitcher, turns out to have the fifth best winning percentage against winning teams of all the pitchers listed. Against tough competition, only Koufax, Grove, Pedro and Juan Marichal won games at a better clip than Mike Mussina.

So our readers made the wrong comparables. Don Sutton isn’t comparable to Mike Mussina…Don Sutton’s better comparable is his one-time teammate Don Drysdale:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Drysdale	209-166	100-55	.645	109-111	.495
Sutton	324-256	172-105	.621	152-151	.502

And Roy Halladay can reasonably be compared to Mike Mussina:

Pitcher	Career W-L	W-L <.500	Win %	W-L .500+	Win %
Mussina	270-153	151-71	.680	119-82	.592
Halladay	203-105	108-37	.745	95-68	.583

Alright…out of that rabbit hole.

David Fleming is a writer living in Wellington, New Zealand. He welcomes comments, questions, and suggestions here and at dfleming1986@yahoo.com.

COMMENTS (28 Comments, most recent shown first)

shinsplint
ok, sure...thanks.
8:03 PM May 29th

tangotiger
I suggest going to my blog instead.

tangotiger.com/index.php/site/comments/redefinethewin
7:16 PM May 29th

shinsplint
Oh yeah, you're right. When i translated it to an equation, I guess I forgot an element of my own formula. Thanks.

Anyway, maybe we shouldn't hijack Dave's post here, as I'm sure he'd like comments related to his observations. Perhaps we could continue comments in the "Pitcher Points" thread?
5:34 PM May 29th

tangotiger
I thought I read your thing that after you got the "4" score, you multiplied it by the number of his innings. Hence, it's what I said.
5:04 PM May 29th

shinsplint
tangotiger, I may be missing what you're doing, but I see my equation as being R - ((9*RA)/IP), assuming a 9 inning game. So if a pitcher's offense scores 10 runs in a game and he gives up 4 in 6 innings, then he gets a "4". And that "4" is compared to the other pitcher's totals.

I was using Win Points and Loss Points to allocate 100 points for pitchers on the winning and losing teams, but I think it works well to determine the winner or loser as well, which would just be the pitcher with the most points.

In regards to varying levels of offensive support, you'll see that "pitching to the score" is valued using my formula. Therefore a starter can give up 5 runs when his team scored 10 and he still gets the lion's share of the points relative to the reliever(s) based on his innings pitched advantage. However, if his team scores just a run or two, then he's got to pitch real well to get the most points. In that case, many times a reliever will have the most points. Or at least have a high number of points. So there's a reward for all pitchers for taking advantage of "leverage" situations. If a reliever pitches two scoreless innings in a 5-4 win, he gets many more points than two scoreless innings in a 10-4 win.
3:51 PM May 29th

tangotiger
0.25 outs - RA is the same thing as outs - 4 RA. Right now, I'm using outs - 5 RA.

***

Redoing the same expansion idea, but using 2.25 R for runs scored by your team in a loss, and we get: outs - 12 RA. I'm using outs - 10 RA.

***

So, your metric passes the initial sniff test. The question therefore is how does it work when the team scores 1,2,3,4,5 runs in a win, or 3,4,5,6,7 runs in a loss. Does making the metric flexible by using the number of runs a team scores give better results than the fixed equations that I use does?

In my case, I'm basically using 5.4 offense runs in a win and 2.7 offense runs in a loss.

Thanks for giving me something to think about. Your method might actually work out better.

2:34 PM May 29th

tangotiger
Your metric is this:
R * IP minus RA * 9

With R being the number of runs his offense scores. In an average win, the offense will score around 6.75 runs. Just for illustration, let me use that.

6.75 * IP minus RA * 9

IP is also outs/3, so we have:

2.25 * outs minus RA * 9

Which we can further reduce to:
9 * (0.25*outs - RA)

The "9" is extraneous. That leaves us with:

0.25*outs - RA

Which is exactly what my original proposal was.

Now, the big difference is whereas I do NOT consider the number of runs the offense scores, YOU do. Well, I consider it in terms of whether the team wins or not, so that I know whether to give a pitcher a W or L. But, what I like about what you did is that it kind of accounts for the game conditions, the park, the season, whatever.

So, I like the basic idea of using the team offense. I don't know that the construction you have is the best, but the framework seems reasonable.

***

As for "sharing" responsibility: the idea is that if you want something basic, make it basic. Once you get into sharing, etc, something that implies precision, you can't rely on something crude.

2:28 PM May 29th

shinsplint
Re: tangotiger..."removing the timing aspect of the performance relative to his own team's offense". Interesting. That's exactly the philosophy I used in my Pitcher Points link. Meanwhile, I'll take a look at your link..

boards.billjamesonline.com/showthread.php?4000-Pitcher-Points
12:06 PM May 29th

tangotiger
For those interested, I expand on the idea of Redefining a Win

tangotiger.com/index.php/site/article/redefinethewin
11:56 AM May 29th

tangotiger
The bonus points was discarded, and the idea is basically rolled into the system as constructed.

The partial points: in another thread, I argued that if we stick to the basic principles of what a W/L record, then the only thing I am changing is removing the "timing" aspect of the performance (relative to his own team's offense), and focusing it only on the actual results of the pitcher (and his fielders).

If we do ANYTHING other than that, then we are not "redefining" the W/L, but actually creating something new. That is, in my proposal, I'd be "competing" with the existing official W/L rule, and so, only one can survive. Hence, the redefinition.

Under your proposal (and others have suggested something like that, pluys other things too), it would allow the existing definition to survive. There's nothing wrong with your proposal. It just won't wipe away the existing rule.

11:28 AM May 29th

chuck
tangotiger, not quite on the topic of this article, but related...
I was reading the discussion thread on your site about re-defining the Win (and Loss). For those that haven't seen it:
tangotiger.com/index.php/site/comments/fixing-the-assigned-w-l-record

2 questions. If I understand it right, the current state of the Assigned Win process would be (for wins):
give 1 point for each out recorded,
subtract 5 points for each run allowed.
The pitcher with the most points gets the W.
In case of a tie, give it to the pitcher that gave up fewer runs.
(Then, to the pitcher with the most outs, and then to the pitcher who appeared first in the game.)

For assigning the L:
give 1 point for each out recorded,
-10 points for each run allowed.
Pitcher with the biggest negative score gets the L.
For ties, the L goes to the pitcher that allowed more runs.
After that, to the pitcher with fewer outs, and then to the pitcher who appeared first.

In the initial post in the thread, you were suggesting bonus points:
1 bonus point for each out beyond the 1st 9. I take that to mean for all pitchers, a bonus point for every out after the 3rd inning of the game. So, question 1), is the bonus point still part of the process or was it discarded? I could see how it would award more win points for both starters that stayed in the game long (as in Bill's game score), as well as relievers in higher leverage situations.

Question 2) would you see any merit to dividing the win between pitchers each game? For instance, a pitcher goes 6 innings, allowing 3 runs. Reliever A goes 2 innings, gives up 1 run, and the Closer allows no runs (and the team wins).
Starter gets 18 for outs, -15 for runs (+ 9 bonus, if using those).
Reliever A gets 6 for outs, -5 for the run (+ 6 bonus)
Closer gets 3 for outs (+3 bonus)

Starter: 12 pts, Reliever A: 7 pts, Closer: 6 pts.
Under the system right now, the starter gets the W.
I'm suggesting he get 12/(12+7+6), or .48 of the W.
Reliever A gets 7/25, or .28, and the Closer the remaining .24.

Your system has a much greater chance of being accepted by people, being a simple one to do. But I like the dividing, as it would assign more wins and losses to all relievers (possibly eliminating the hold and save) and more equitably distribute the credit for the W's and debit for the L's.
10:27 AM May 29th

tangotiger
Baseball Reference lets you list years, so, just a few clicks away...
9:47 AM May 29th

studes
I was intrigued about your Koufax finding, given that he was known to absolutely dominate the Mets and Astros in his prime. Here is his W/L % by year during his five great years (l.t. .500 teams listed first):

1962: .800/.625
1963: .833/.833
1964: 1.000/.667
1965: .824/.706
1966: .789/.706

Other than 1963, the differential in these splits look about normal to me. I wonder if the small difference in his career splits is due to the types of teams he pitched against in the first half of his career, vs. those in the second half.
7:56 AM May 29th

tangotiger
Baseball Reference to the rescue:

Min 1000 IP, against >.500 teams
bbref.com/pi/shareit/Ed4Pe

and <.500 teams
bbref.com/pi/shareit/ddB6l
9:36 PM May 28th

chuck
I found a pitcher with a higher percentage of his decisions against below-.500 teams than David Cone: Don Newcombe, at 59%. (Cone was 57%).

Newcombe's splits:
96-44 vs <.500 (.686)
53-46 vs .500+ (.535)

7:21 PM May 28th

llozada
Dave, great article. I believe your "biggest gap" pitchers table is wrong. It looks like you flipped Morris's and Clemens's numbers.

I think it should be:

Name W-L <.500 Win % Difference
Sabathia 0.770 0.500 270
Morris 0.705 0.444 261
Clemens 0.747 0.554 193
Hubbell 0.720 0.547 173
Palmer 0.719 0.548 171
Alexander 0.728 0.556 172

Feller's are actually .728 .536 192 right between Clemens and Hubbel.

C.C. and not Jack actually has the largest gap at 270.

Sorry for being the one picking on your work.
6:33 PM May 28th

WarrenJohnson
Fine article, and a lot of good reader comments as well. But the list
that follows "And the biggest gaps" has some errors: Morris is listed
with Sabathia's percentages, Clemens with Morris's, and Feller with
Clemens's. Hubbell's are correct, but Palmer is listed with Hubbell's
percentages, and Alexander with Palmer's.
2:52 PM May 28th

chuck
On Guidry/Ford... pardon the length, but it's interesting, that difference between them.

Guidry's imbalance looks to be the circumstances in which he pitched. There were only 4 seasons where he had an imbalance between strong and weak teams larger than 3 games. The starts against stronger teams are listed first:
1979: 23 / 7
1981: 17 / 4
1982: 19 / 14
1983: 11 / 20
In 1979, 9 of the 13 other teams had +.500 records. There were a few really bad teams, and/or a lot of good-to-decent ones.

In 1981 the strike disrupted the balance of the schedule. NY ended up playing many fewer games against the weaker opponents.

In 1982, it's only 5 games, and it appears random that he was balanced towards the strong teams, just as it swings back the other way in 1983. In '83 8 of the 13 other teams were below .500, so it was natural for him to split that way, too.

Ford's circumstances were that for almost his whole career his team was one of the top teams. It was natural for him to be facing sub-.500 teams more often than not.
It’s very apparent that from 1953-1960 he was strategically worked more often against the better clubs than the schedule would imply.

1950: rookie year. 8 more starts against weaker clubs, so Stengel wasn’t throwing him in the deep end right away, til he knew what he had.

1953. There were 4 other .500+ teams. Ford started 20 times against them, and 10 vs. the weaker teams, more than would be implied by 4/7.

1954. There were just 2 other .500+ teams. Almost impossible not to have an imbalance towards the weaker teams, and yet Ford started 13 of his 28 against the .500+ guys.

1955. This year appears more balanced. 1 more start against .500+, but there were 4 of those teams out of the 7.

1956. Again, Ford started 20 times vs. the .500+ teams, 10 against the others. There were 4 out of the 7 at .500+, but that’s still a noticeably higher rate.

1957. 10 vs strong; 7 vs weak. But again 4 out of 7 were .500+.
1958. 14 vs strong; 15 vs weak. 4 of 7 were .500+. This year does not appear to show a concerted effort to pitch him against the better clubs. That might be true of ’57 also.

1959. Only 2 of the other 7 teams were .500+. Ford started 11 vs. those two teams and 20 vs. the other 5. It’s a bit higher percentage than 2/7, so that looks somewhat purposeful.

1960. Again, just 2 of the 7 were .500+, so it was hard not to pitch more against the weaker teams. Ford started 12 vs strong, 17 vs weak, again a higher pct. than if he’d been used equally.

No question that Stengel in these years threw him up against the better teams. His splits were 119 starts vs the .500+ and 120 vs the sub-.500’s. Taking out his rookie year, it was 117 to 110.

But pitching more regularly under Houk, over the 1961-63 seasons Ford's splits were:
44 vs. .500+, 69 vs. the others.
This was natural, as only 3 of the 9 other teams were over .500 in ’61 and ’63, and 4 out of 9 in ’62.
In all the non-Stengel seasons, the split was 84 vs strong, 117 vs weak, and I think was a function of the divide in the league and of using him regularly.
12:07 PM May 28th

tangotiger
I have an Individualized Won-Loss system (The Indis) that represents a player's career in terms of W-L (similar idea as Win Shares and Loss Shares).

For Halladay, he's 101-29
For Drysdale, he's 107-50

The difference is obviously 6 wins and 21 losses. Now, one CAN make the argument that if Doc had 27 more "games" that he'd only "win" 6 more games (in his depleted condition).

If we focus on their peak (roughly 100 "games"), we get:
92-12 Doc
78-26 Drysdale

The problem with WAR, WAA and all the other ubermetrics is when they are represented as single dimensions. By doing that, you MUST accept the baseline comparison level. If you can't do that, then you should look at the data in two dimensions.

9:12 AM May 28th

shinsplint
I was going to say a similar thing to what tango said. The implication appeared to be that Halladay didn't have that superior WAR to Drysdale, but somehow transcended his WAR to have a much better W/L pct. However, much of Drysdale's WAR was because of pitching about 25% more innings, apparently much of it at a below average, but above replacement level. Therefore it would be expected that Drysdale's W/L pct would be significantly less without assuming Halladay had more of a knack for winning games.
8:06 AM May 28th

garywmaloney
Chuck writes: "A big difference: 55% of Guidry's decisions were against +.500 teams, while just 45% of Ford's were. "

Excuse me -- doesn't that figure totally contradict all the research (e.g. Chris Jaffe, Bill) suggesting that Ford was heavily leveraged by Stengel over the first 10 of his seasons, pitching disproportionately against the BEST teams in the AL?

Just doesn't sound right -- Ford wasn't used on a regular rotation until Houk in 1961 (whereupon he went 25-4 and won the Cy Young). Am I missing something here?
6:41 AM May 28th

tangotiger
WAR has a longevity component to it. Comparing Doc and Drysdale with similar WARs, but giving you an unsatisfying answer means that you prefer Wins Above Average (WAA) rather than WAR.

In that case, Doc is +41 and Drysdale is +29.

We have different metrics to convey different perspectives. If you don't think WAR does it for you, then maybe WAA does.
6:21 AM May 28th

chuck
The biggest bully I found was a 6', 190-pound guy. None other than the pitcher Bill listed as the best Big Game Pitcher: Roy Oswalt.
.732 vs sub-.500
.466 vs +.500.
If one divides the 1st pct by the 2nd to get a ratio, Oswalt has the 4th highest ratio (I've found among the big winners) of beating up on the bad teams. Ahead of him are:
Jack Morris .705/.444,
Early Wynn .684/.434, and
Wes Ferrell .732/.465.
However, 56% of Oswalt's decisions were against the weaker teams.
For Wynn that number was just 47% and for Morris and Ferrell 51%.

I found a guy who beat up on the bad teams much worse than Sabathia has, though over fewer games:
Spud Chandler went 71-12 against sub-.500 teams (.855).
Versus good ones, just 38-31.
Like Ford, terrific support.

Pitchers who had a lot of decisions against below-.500 clubs:
57% David Cone
56% Oswalt, Kevin Brown, Andy Pettitte.
55% Ford, Tim Wakefield, and Chandler.

Ford and Guidry. Both small left-handed Yankee stars.
Both had excellent run support and good defensive teams.
Their winning pct splits:
Ford .759/.606
Guidry .709/.604
A big difference: 55% of Guidry's decisions were against +.500 teams, while just 45% of Ford's were.
Guidry and Marichal are similar in that respect. Great run support, and 56% of decisions against the better teams for Marichal.

I've only found one pitcher who had a better record vs the +.500 teams than against the sub-.500 teams.
Bret Saberhagen:
83-54 vs above-.500 teams (.606)
84-63 vs below-.500 teams (.571)
2:51 AM May 28th

chuck
Fun and interesting article, Dave.
Who's who?
The 3 big Oriole starters against +.500 teams:
.558, .550, and .548.

The 3 top guns on the Braves, vs +.500 teams:
.557, .549 and .530.
Answers in a bit.

The elephant in the room is team support, mostly on offense.
Here's Nolan Ryan's record:
.589 vs sub-.500 teams
.482 vs +.500 teams.

But here are Ryan's OPS splits:
.608 vs sub-.500 teams
.604 vs +.500 teams
Neither could hit him, he was slightly better against the good teams, and yet he was 107 percentage points worse against them. My guess is that his teams were pretty bad at scoring against the good teams (pitchers).

Not only that, but 59% of his decisions were against +.500 teams. Others with a high percentage:
63% Dazzy Vance
60% Larry Jackson, Burleigh Grimes
59% Ryan, Bucky Walters, Ned Garver, Dutch Leonard, Eddie Lopat, and Don Drysdale.

Drysdale, as mentioned, likely was spot-started against tougher opponents. Walters was on good teams in the middle of his career, and not-so-good on either end of it. Lopat was on 4 mediocre White Sox teams before joining the Yankee powerhouse. It looks like he, too, may have been saved for the tougher teams. But others on that list played on bad teams and just naturally had a majority of their decisions against tough teams.

Perhaps the most impressive pitcher in this regard is Dazzy Vance. He had 63% of his decisions against +.500 teams, and yet went
122-90 against them (.575) and .600 against the weaker teams.

Like Ryan, Whitey Ford also had just about even OPS splits, too:
.639 vs sub-.500 teams
.642 vs +.500 teams.
He also had a big spread between percentages. But he kicked the sub-teams' asses and the good teams also. Winning percentages:
.759 vs sub-.500 teams,
.606 vs +.500 teams.
He got tremendous run support and defensive support.

Ok, the Orioles and their record against the good teams:
86-68 (.558) Cuellar
83-68 (.550) McNally
109-90 (.548) Palmer
Against the below-.500 teams, the order is reversed.

The Braves' guys against good teams:
146-116 (.557) Glavine
162-133 (.549) Maddux
097-086 (.530) Smoltz
Against sub-.500 teams, it's .672, .646, and .627, with Glavine and Maddux switching places.

2:19 AM May 28th

evanecurb
Interesting work, Dave. Thanks. I think the sub .500 / .500+ splits are great to have when arguing HOF candidacy. But there's a simple principle at work that is always there when dealing with W-L records: the quality of your team. You mention that Carlton pitched on some bad teams, but that was brief. The Cardinals were good in the late sixties, and the Phillies were good in the mid to late seventies and early eighties. It was the early seventies when Carlton pitched on bad Phillies' teams, maybe 20% of his career? Truth be told, the overwhelming majority of 200 game winners and 200 game winners with high winning percentages pitched on good teams. Who are the exceptions? Roberts and Niekro, for sure...Blyleven and Ryan, maybe. Larry Jackson, Bob Friend... I'm sure there were others.

The tendency, when trying to place won-lost records in context, is either to talk about whether or not the pitcher was on a 'good' team (i.e. a team that wins a lot of games), and/or whether or not the pitcher had a lot of run support. Two other factors that matter are the quality of opponents, which you cover very well, and the quality of the defense. The 1970s Orioles, A's, and Reds pitchers all benefited from being on teams that both scored a lot of runs and had a strong defense. So we expect the pitchers on these teams to have good W-L records.
12:25 AM May 28th

izzy24
I remember when Izzy Alcantra came up for a cup of coffee with the Red Sox and I was very excited that we shared the same nickname. I was much less excited after he tried to decapitate a catcher (although, I was very impressed with how quickly he committed to doing it).
9:19 PM May 27th

ajmilner
As for Marichal's .602 record against winning teams -- Whitey Ford was .606 (94-61) against winners.
9:14 PM May 27th

those
Short career, but Don Gullett was 53-27 (.663) against .500+ teams.

Also worth noting: Joe Niekro was 106-130 (.449) with a 4.05 ERA against .500+ teams, but did not allow a run in 20 postseason innings.
8:37 PM May 27th

A Doc, Two Dons, and a Moose

COMMENTS (28 Comments, most recent shown first)

Leave a comment

Report inappropriate comment


Type of Abuse:
Comments: