(Pitcher Regularity Scores)

(Poetically re-titled “Major League Managers and the Longing for Stability”)

 

            Major league baseball has been substantially re-created and re-invented, over the last hundred years, by the desire of major league managers to create more stable workloads for their pitchers.   This desire has overridden and thus essentially eliminated platooning, has sharply limited pinch hitting, has done away with third catchers, has virtually eliminated complete games, has re-defined normalcy in roster construction, pushing us gradually from 7 pitchers per team to 13, has added dozens of new statistical measurements to the game, and has caused games to get substantially longer and slower, as managers change pitchers every more frequently.

            We have talked a lot, we have written a lot, we have done much research on the issue of pitcher workloads.   The research that we have done—on this issue, perhaps more than any other—has invaded the game itself, and is now relied upon by people in decision-making positions to help guide the ways that pitchers are used.

            But there is a parallel issue, of perhaps equal or similar importance, about which we have done very little research.   That is the issue of the regularity of pitcher usage.

            Suppose that we had a metric to measure not how hard the pitcher had been worked, but how regularly, how consistent and even his workload was over the course of his season or career.   If we had such a metric, think of the other questions that this would open up to us.   Does pitcher performance improve with more regular usage?   Does regularity of usage help to prevent injuries?    Is a workload of 200 innings with regular use perhaps equal to a workload of 150 innings with irregular use, or 120 innings, or 80?

            Is regularity of pitcher use a variable among managers?   If so, is it a relevant variable in terms of manager success, or an irrelevant variable?  Do managers who keep their pitchers on a more regular schedule enjoy more success?

            I have three tasks in this article.   First, I will outline a simple method with which to distinguish between regular and irregular workloads.    Second, having applied this method to more than 1,000 pitcher/seasons—about 3% of all pitcher/seasons in baseball history—I will report the data resulting from those studies.    Third, I will try to explain and interpret the data as best I can, and then I will stick just a toe into those issues that I have outlined above.

 

I.  The Method

 

            It would have been impossible to have done this research—as it would have been impossible to have done any of the research that we do these days—without Retrosheet.   Retrosheet.org in recent months has added daily logs for pitchers dating back to 1920.   This research was done based on those logs.

            My method here is extremely simple.   From the game logs available through Retrosheet, we extract two figures:  the dates on which the pitcher pitched, and the number of batters that he faced.   For Bob Feller, 1946, this is that data:

 

Month	Day	Year	BFP		Month	Day	Year	BFP
4	16	1946	29		7	24	1946	31
4	21	1946	41		7	28	1946	34
4	26	1946	38		7	31	1946	34
4	30	1946	32		8	4	1946	26
5	4	1946	32		8	8	1946	33
5	8	1946	36		8	10	1946	9
5	12	1946	45		8	11	1946	1
5	17	1946	34		8	13	1946	29
5	21	1946	33		8	17	1946	36
5	26	1946	42		8	20	1946	34
5	30	1946	38		8	22	1946	5
6	4	1946	34		8	24	1946	37
6	8	1946	35		8	28	1946	32
6	12	1946	36		9	1	1946	33
6	16	1946	38		9	3	1946	7
6	21	1946	34		9	5	1946	22
6	25	1946	41		9	8	1946	33
6	29	1946	34		9	12	1946	35
7	3	1946	41		9	15	1946	29
7	7	1946	37		9	19	1946	34
7	11	1946	28		9	22	1946	38
7	16	1946	41		9	25	1946	37
7	20	1946	38		9	27	1946	21
7	21	1946	7		9	29	1946	38

 

            Feller pitched 48 times in 1946, facing a total of 1,512 batters.  

            Scanning this log from a modern perspective, we can see a number of quite remarkable things.   Feller pitched a complete game at Washington on July 20, 1946, came back on July 21 and pitched an inning and two-thirds of relief (scoreless relief, striking out four.)   He pitched a complete game at Chicago on August 8^th (a one-hit shutout), came back to pitch in relief in St. Louis on August 10^th (3 innings of perfect relief), pitched in relief again on August 11^th, then started against Detroit on August 13^th.   He gave up three hits, one run in 8 innings against Detroit, lowering his ERA to 1.88, rested three days, started again, rested two days, started again, missed one day, pitched in relief, missed one day, and started again.   In the last five days of the season he pitched a total of 23 innings.

            But our purpose here is not to stare at this in open-mouthed awe, but to score the regularity (or irregularity) of this assignment.   The score that I developed has two parts, and a potential of 25 points per game appearance, after the first appearance.

            In Part A, we compare the intervals between appearances. . .counting the number of days off between this appearance and the previous one.   For Feller, 1946, this is as follows (below). The gap between the last start of Feller’s 1945 season and his first in 1946, the off-season being longer then than it is now, was 201 days. . .which actually makes Feller’s 1946 season even more remarkable, when you consider that the season then was about three weeks shorter than it is now: 

 

Month	Day	Year	Gap
4	16	1946	201
4	21	1946	5
4	26	1946	5
4	30	1946	4
5	4	1946	4
5	8	1946	4
5	12	1946	4
5	17	1946	5
5	21	1946	4
5	26	1946	5
5	30	1946	4
6	4	1946	5
6	8	1946	4
6	12	1946	4
6	16	1946	4
6	21	1946	5
6	25	1946	4
6	29	1946	4
7	3	1946	4
7	7	1946	4
7	11	1946	4
7	16	1946	5
7	20	1946	4
7	21	1946	1
7	24	1946	3
7	28	1946	4
7	31	1946	3
8	4	1946	4
8	8	1946	4
8	10	1946	2
8	11	1946	1
8	13	1946	2
8	17	1946	4
8	20	1946	3
8	22	1946	2
8	24	1946	2
8	28	1946	4
9	1	1946	3
9	3	1946	2
9	5	1946	2
9	8	1946	3
9	12	1946	4
9	15	1946	3
9	19	1946	4
9	22	1946	3
9	25	1946	3
9	27	1946	2
9	29	1946	2

 

            Feller’s “gaps”, as you can see, are actually fairly regular for most of the season—five days, five days, four days, four days, four days, etc.   

            I score these “gaps” by asking this question:  Is the gap between these two appearances the same as the gap between the previous two appearances?   If the gap is exactly the same, I score that as “10”, or 10/10.   If the gap is different by one, I score it as 9/10.  If it is different by two, I score it as 6/10.   If it is different by three, I score it as 1/10.   If it is different by four or more, I score it at 0/10—totally irregular work.

            For Feller in 1946, this creates the following scores (S A stands for Score-A):

 

Month	Day	Year	Gap	S A
4	16	1946	201
4	21	1946	5	0
4	26	1946	5	10
4	30	1946	4	9
5	4	1946	4	10
5	8	1946	4	10
5	12	1946	4	10
5	17	1946	5	9
5	21	1946	4	9
5	26	1946	5	9
5	30	1946	4	9
6	4	1946	5	9
6	8	1946	4	9
6	12	1946	4	10
6	16	1946	4	10
6	21	1946	5	9
6	25	1946	4	9
6	29	1946	4	10
7	3	1946	4	10
7	7	1946	4	10
7	11	1946	4	10
7	16	1946	5	9
7	20	1946	4	9
7	21	1946	1	1
7	24	1946	3	6
7	28	1946	4	9
7	31	1946	3	9
8	4	1946	4	9
8	8	1946	4	10
8	10	1946	2	6
8	11	1946	1	9
8	13	1946	2	9
8	17	1946	4	6
8	20	1946	3	9
8	22	1946	2	9
8	24	1946	2	10
8	28	1946	4	6
9	1	1946	3	9
9	3	1946	2	9
9	5	1946	2	10
9	8	1946	3	9
9	12	1946	4	9
9	15	1946	3	9
9	19	1946	4	9
9	22	1946	3	9
9	25	1946	3	10
9	27	1946	2	9
9	29	1946	2	10

 

            Altogether, Feller scores for 1946 at 409 out of a possible 470 points on Part One of our system—86%.

            In that era, the common rule was that the ace of the staff—what we would now call the #1 starter—started on a regular four-day schedule if possible, always started against key opponents, and the rest of the staff had to move around to accommodate him.   If you checked Feller’s teammates, I am certain you would find that their schedules were less regular than Feller’s.   There were many more “travel days” then than now and frequent double-headers, and Feller hop-scotched other pitchers to get more work.   That’s the way it was done at that time.

            The other question that we ask is, “Was the number of batters that this pitcher faced in this outing the same as it was in the previous outing?”

            If the number of batters faced was exactly the same or was within 3 of being the same, we score that at “15” (15/15).    If it was different by 4, that’s 14; different by 5, that’s 13; different by 6, that’s 12.   For each difference in batters faced in excess of three, we subtract one, down to zero.   For Feller, 1946, these are the scores for each outing:

 

Month	Day	Year	Gap	S A		BFP	S B
4	16	1946	201			29
4	21	1946	5	0		41	6
4	26	1946	5	10		38	15
4	30	1946	4	9		32	12
5	4	1946	4	10		32	15
5	8	1946	4	10		36	14
5	12	1946	4	10		45	9
5	17	1946	5	9		34	7
5	21	1946	4	9		33	15
5	26	1946	5	9		42	9
5	30	1946	4	9		38	14
6	4	1946	5	9		34	14
6	8	1946	4	9		35	15
6	12	1946	4	10		36	15
6	16	1946	4	10		38	15
6	21	1946	5	9		34	14
6	25	1946	4	9		41	11
6	29	1946	4	10		34	11
7	3	1946	4	10		41	11
7	7	1946	4	10		37	14
7	11	1946	4	10		28	9
7	16	1946	5	9		41	5
7	20	1946	4	9		38	15
7	21	1946	1	1		7	0
7	24	1946	3	6		31	0
7	28	1946	4	9		34	15
7	31	1946	3	9		34	15
8	4	1946	4	9		26	10
8	8	1946	4	10		33	11
8	10	1946	2	6		9	0
8	11	1946	1	9		1	10
8	13	1946	2	9		29	0
8	17	1946	4	6		36	11
8	20	1946	3	9		34	15
8	22	1946	2	9		5	0
8	24	1946	2	10		37	0
8	28	1946	4	6		32	13
9	1	1946	3	9		33	15
9	3	1946	2	9		7	0
9	5	1946	2	10		22	3
9	8	1946	3	9		33	7
9	12	1946	4	9		35	15
9	15	1946	3	9		29	12
9	19	1946	4	9		34	13
9	22	1946	3	9		38	14
9	25	1946	3	10		37	15
9	27	1946	2	9		21	2
9	29	1946	2	10		38	1

 

 

            Here, as we can see, Feller’s scores are somewhat lower.   Feller’s S B scores for the season total up to 462 out of a possible 705, or 66%.   Adding together S A and S B, Feller’s “Workload Regularity Score” for the 1946 season is 871 out of a possible 1175, or 74%.

            In that era, that’s a very high score.   Among the fourteen pitchers for whom I have data covering the 1946 season, Feller’s 74% is easily the highest percentage.    Feller’s workload, in 1946, was more regular than that of any other pitcher that I studied. 

 

II.  The Data

 

            In late 1960, the legendary Casey Stengel was fired by the New York Yankees and replaced with the late Ralph Houk.   Houk’s first phone call, after learning that he had the job, was to Whitey Ford.

            Let’s pause there a second, and read between the lines.   Why did Houk call Ford first?   Was Ford the leader of the team?

            Well, yes, but no more than Berra or Mantle, and there is something else going on here.   You remember Casey’s famous line, “The secret of managing is to keep the five guys who hate you away from the ten guys who are on the fence?”   Ford, by 1960, had (we suspect) become the de facto leader of the Dump Casey faction of the Yankee roster.   He wasn’t ripping Stengel is the newspapers or anything, but it seems pretty clear, reading between the lines a little, that Ford by 1960 had grown disenchanted with Stengel’s management of the team, in large part because of the way that he used his pitchers.   Houk was reaching out to Ford to heal the rift.

            And what did Ford say to Houk, in that phone conversation?   “Just put me on the mound every fourth day and let me pitch.”   (Quote is not exact.) 

            Stengel was a great believer in percentage baseball—that is, in deploying his players at their most effective moments.   He used left-handed hitters against right-handed pitchers, he used right-handers against lefties, and he varied his pitching patterns constantly so that his best pitcher—Ford—would pitch against the strongest opposition and in the parks where he was best suited, while the lesser pitchers would soak up the starts against the teams the Yankees could beat anyway.   Ford never won 20 games for Stengel, largely because he didn’t pitch on a regular four-day rotation like other staff aces.

            It was smart percentage baseball, and that’s the way the game was played by some managers in 1920, in 1930, in 1940, but by 1960 it wasn’t really done that way anymore.   Starting pitcher leveraging, as Chris Jaffe calls it in his new book about managing, was pretty much gone by 1960.   Casey was still doing it, and Whitey didn’t like it.

            These are Whitey Ford’s Regular Use Scores throughout his career:

 

 

 

            Ford’s pitching schedule in 1960 was 53% regular.   In 1961, under Ralph Houk, it was 79% regular—a huge increase.

            In 1960 Ford was 12-9.  In 1961 he was 25-4, and won the Cy Young Award.   Of course, the true facts are always more complicated than the simplest narrative:  It was an expansion year, Ford’s ERA actually went up, etc.   I’m not trying to hide the true facts; I am merely pointing out:

            1)   Ford was used by Stengel in a very irregular pattern,

            2)   Ford was unhappy about this,

            3)   Houk used Ford in a much, much more regular pattern,

            4)   Ford won twice as many games and won the Cy Young Award, and

            5)   Ford was much happier.

            I am using this fairly well-known story to demonstrate that our method works, but there

are actually are a series of story lines from 1961/1962 that are relevant here.    From 1956 to 1961 Lew Burdette had won-lost records of 19-10, 17-9, 20-10, 21-15, 19-13 and 18-11.   Over that six-year span Burdette was second in the major leagues in innings pitched and in games won, behind his running mate Warren Spahn.   He was one of the premier pitchers in the game.

            In 1962 Burdette won just 10 games—10-9, ERA of 4.88, far worse than the league average.   Unhappy, Burdette said the next spring that “the only difference between last year and any other year was that I didn’t get to pitch regularly.”   There’s that word again, regularly.  These are Burdette’s Regular Use Scores:

 

 

 

            The linguist in me wants to point out that what Burdette meant by “regular” was not exactly the same thing that Ford meant by regular.    “Regular” has (at least) three meanings in common usage:

            1)  Consistent,

            2)  Frequent, and

            3)  Standard.

 

            When a dietary supplement says that it gets you regular and keeps you regular, what it means is (1).   When a boyfriend says that he wants more regular action from his girlfriend, what he means is (2).   When a reporter says that the trial was somewhat irregular, what he means is (3).

            When Ford said that he wanted to pitch more regularly, he meant (1), and also possibly (2).   When Burdette complained that he was not being given regular work, he meant (2).   When Billy Martin allowed Rick Langford to pitch 28 complete games in 1980, we could say that that was irregular, meaning (3).   Small point, but. . .we need to think clearly. 

            Anyway, the regularity of Burdette’s usage did decline in 1960, no matter how we define “regular”, but there is another side to that story.   The manager of the Milwaukee Braves in 1959, when Lew Burdette was 21-15 and getting really regular use (meaning frequent), was Fred Haney.  It is the view of the author that Haney in 1959 may have had the worst season by any manager in major league history.   Haney took a team that. . .I am not exaggerating for emphasis; this is exactly what I think. . .Haney took a team that should have won the National League by 20 games, and managed to lose the race.   If you compare the roster of the 1959 Braves to the 1959 Dodgers, it is difficult to explain how this is possible.

            One of the 28 stupid things that Haney did was, he worked Spahn and Burdette to death.   Spahn was a great pitcher and Burdette was pretty good, but Haney over-worked them both, they pitched .500 ball over the last two months, and they both finished 21-15, pitching 582 innings.   This might be understandable if Haney was short of other pitching options, but in fact he had a very good third starter (Bob Buhl) and a quite exceptional number of young pitchers, whom he refused to pitch.   He had Carlton Willey, who in 1958 had made 19 starts, led the league in shutouts with 4, and posted a 2.70 ERA, plus Joey Jay (97 innings in 1958 with a 2.13 ERA) and Juan Pizarro (97 innings in 1958 with a 2.69 ERA).    Haney didn’t believe any of these young pitchers was trustworthy, and tried instead to make Burdette and Spahn win the pennant for him.

            So Haney got fired, and the team was turned over to Charlie Dressen, but Dressen was an older manager who had been around the block several times, and he liked older pitchers as well.   Joey Jay didn’t get to pitch. . .well, REGULARLY, until 1961, when he was traded to Cincinnati:

 

 

            In 1959, pitching on a very irregular schedule for Fred Haney (48% regular), Jay was 6-11 with a 4.10 ERA.  In 1960, with his regular use score increasing to 63%, Jay was 9-8 with a 3.25 ERA.   In 1961, traded to Cincinnati, Jay’s Regular Use Score went up to 77%.   His record improved to 21-10 in 34 starts.  (His ERA went up, but that’s an illusion created by the park and league.)

            It wasn’t just Jay, either.   Juan Pizarro was a lefty with one of the best fastballs of that era.  In 1960, fighting for regular work, he went 6-7 with an ERA a run over the league norm, park adjusted.   Traded to the White Sox that winter, he also got regular work, and he also immediately became effective.   I don’t have regularity scores for Pizarro, but it seems like a reasonable assumption that his regularity of use also increased sharply with the trade.

            You remember what I said earlier about Bob Feller—that in this era, most teams tried to keep their #1 starter on a four-day schedule, and everybody else had to work around that.   What Burdette was complaining about, really, was that from 1956 to 1961 he was one of the big dogs, that everybody else had to operate around.    In 1962 he was downgraded to being one of the little dogs, who had to work around the big dogs.   He didn’t like it.

            But really, it should have happened sooner; the Braves should not have been jacking around Pizarro, Jay and six other young pitchers to keep Lew Burdette happy.

            There is a very powerful trend, across history, of pitcher usage becoming more regular and more predictable.   What we are talking about, really, are (at least) five different changes in the game, all operating toward the common purpose of regularizing a pitcher’s work schedule.   Those five changes are:

            1)  The disappearance of double-headers,

            2)  The division of pitching staffs into starters and relievers,

            3)  The abandonment of the practice of starter leveraging,

            4)  The expansion of the bullpens, and

            5)  The adoption of pitch limits.

            The key point, really, is (4).   The expansion of the bullpens, I will argue, has been primarily driven by this exact reason:  The desire to regularize workloads.   We have written volumes about the expansion of the bullpens.   But when you think about it, what are the extra pitchers really doing?   They’re stabilizing the staff.   They’re making it possible to keep workloads regular.   That’s the main purpose of the larger bullpens; to regularize workloads.   We have missed this fact, I think, because we focus on what we can measure.   Not being able to measure the regularity of usage, we have essentially looked right through the real cause of the expansion of bullpens, which is the widespread belief that pitchers are most effective when they have regular work schedules.   But perhaps I am getting ahead of myself.

            Since 1920, the regularity of use of major league pitchers has advanced relentlessly.  While my studies cover only about 3% of major league pitchers and not even a representative 3%, the study nonetheless demonstrates a massive change in the regularity of usage by pitchers over the last nine decades.  For those pitchers for whom I have figured data, these are the overall Regular Use Scores for each of the last nine decades:

 

 

            While our studies are limited, I would predict that a comprehensive study would change these percentages by only tiny amounts, and I think we may safely assume that the conclusion that the regularity of pitcher usage has shown steady increases over most of baseball history will not be undermined by additional research.

            These are the top 50 and the bottom 50 pitcher/seasons in our study, in terms of regularity of use.   I included pitchers with less than 20 game appearances, but didn’t count them:

 

 

            The earliest pitcher on the list of those with the most regular workloads was Drysdale in 1963.   These are the bottom 50:

 

Rank	Pitcher	Year	Points	Possible	Pct
	Bucky Walters	1934	0	25	0%
	Joey Jay	1953	0	50	0%
	Tom Zachary	1936	32	175	18%
	Pedro Martinez	1992	5	25	20%
	Bucky Walters	1948	31	150	21%
	Randy Gumpert	1938	18	75	24%
	Steve Carlton	1988	20	75	27%
	Bobo Newsom	1929	15	50	30%
	Whitlow Wyatt	1931	25	75	33%
1	Ellis Kinder	1952	189	550	34%
	Whitlow Wyatt	1944	69	200	34%
	Ernie Bonham	1949	150	425	35%
	Freddie Fitzsimmons	1941	106	300	35%
	Tommy Byrne	1953	100	275	36%
2	Tom Zachary	1929	228	625	36%
	Bill Henry	1955	146	400	36%
3	Bill Henry	1953	183	500	37%
4	Whitlow Wyatt	1930	184	500	37%
	Tommy Byrne	1943	92	250	37%
5	Freddie Fitzsimmons	1939	240	650	37%
6	Elam Vangilder	1920	214	575	37%
	Mel Parnell	1955	112	300	37%
	Whitlow Wyatt	1929	28	75	37%
7	Tom Zachary	1928	243	650	37%
	Firpo Marberry	1935	38	100	38%
8	Bob Feller	1956	171	450	38%
	Larry Gura	1985	68	175	39%
9	Dolph Luque	1930	297	750	40%
10	Lefty Grove	1927	495	1250	40%
11	Tommy Byrne	1951	268	675	40%
	Freddie Fitzsimmons	1935	169	425	40%
	Tommy Byrne	1946	30	75	40%
	Lindy McDaniel	1955	30	75	40%
12	Freddie Fitzsimmons	1940	191	475	40%
13	Dolph Luque	1931	182	450	40%
14	Dazzy Vance	1933	275	675	41%
15	Eddie Rommel	1929	318	775	41%
16	Alex Ferguson	1928	339	825	41%
17	Bob Feller	1955	247	600	41%
18	Bucky Walters	1935	237	575	41%
	Whitlow Wyatt	1939	157	375	42%
19	Tom Zachary	1930	273	650	42%
20	Joe Page	1945	200	475	42%
21	Jim Konstanty	1944	200	475	42%
22	Dizzy Trout	1943	455	1075	42%
	Bobo Newsom	1953	170	400	42%
23	Firpo Marberry	1934	394	925	43%
24	Randy Gumpert	1951	341	800	43%
25	Earl Whitehill	1939	248	575	43%
	Firpo Marberry	1936	54	125	43%
26	Dizzy Trout	1939	346	800	43%
27	Dazzy Vance	1934	260	600	43%
28	Larry Gura	1976	206	475	43%
	Eddie Rommel	1932	174	400	43%
29	Early Wynn	1944	348	800	43%
	Bobby Ojeda	1994	11	25	44%
30	Herb Pennock	1930	264	600	44%
31	Herb Pennock	1922	341	775	44%
	Early Wynn	1946	177	400	44%
32	Eddie Rommel	1921	499	1125	44%
33	Bobo Newsom	1943	411	925	44%
	Steve Carlton	1966	89	200	44%
34	Tom Zachary	1922	345	775	45%
35	Dizzy Trout	1947	348	775	45%
	Freddie Fitzsimmons	1943	90	200	45%
36	Bobby Shantz	1955	248	550	45%
37	Herb Pennock	1925	519	1150	45%
38	Alex Ferguson	1922	430	950	45%
39	Dizzy Trout	1952	397	875	45%
40	Eddie Rommel	1922	568	1250	45%
41	Bobo Newsom	1946	377	825	46%
42	Elam Vangilder	1921	343	750	46%
43	Randy Gumpert	1950	446	975	46%
44	Elam Vangilder	1926	469	1025	46%
45	Ernie Bonham	1941	254	550	46%
46	Mel Parnell	1951	406	875	46%
47	Eddie Rommel	1930	395	850	46%
48	Earl Whitehill	1937	373	800	47%
49	Herb Pennock	1932	245	525	47%
50	Herb Pennock	1933	257	550	47%

 

            Among those with 20 or more game appearances, the most recent pitcher was Larry Gura in 1976—a very interesting season, by the way.   There is almost no chronological overlap between the lowest and highest scores.  Since Johan Santana, 2006, had the most regular work regimen and Ellis Kinder, 1952, had the least regular, let’s compare and contrast their work patterns:

 

 

Johan Santana, 2006						Ellis Kinder, 1952
Month	Day	Year	BFP			Month	Day	Year	BFP
4	4	2006	28			4	16	1952	12
4	9	2006	23			4	18	1952	15
4	15	2006	28			4	23	1952	37
4	21	2006	29			5	3	1952	37
4	27	2006	30			5	5	1952	2
5	2	2006	28			5	16	1952	2
5	7	2006	27			5	18	1952	6
5	12	2006	26			5	19	1952	11
5	17	2006	27			5	24	1952	33
5	23	2006	30			5	30	1952	40
5	28	2006	29			6	1	1952	9
6	2	2006	26			6	5	1952	33
6	8	2006	22			6	7	1952	5
6	13	2006	28			8	14	1952	6
6	18	2006	25			8	24	1952	29
6	23	2006	30			9	1	1952	15
6	28	2006	26			9	5	1952	25
7	3	2006	25			9	7	1952	4
7	9	2006	29			9	9	1952	1
7	15	2006	26			9	13	1952	29
7	20	2006	26			9	16	1952	8
7	25	2006	27			9	24	1952	32
7	30	2006	28			9	28	1952	9
8	4	2006	25
8	9	2006	27
8	15	2006	27
8	20	2006	28
8	26	2006	29
8	31	2006	27
9	5	2006	27
9	10	2006	23
9	15	2006	31
9	21	2006	25
9	26	2006	31

 

 

            Santana in 2006 never faced fewer than 23 batters in an outing, nor more than 31.    His largest difference between batters faced in any two consecutive appearances was eight, and only ten times was there a difference larger than three.

            The most batters that Santana faced in any outing was 31.   Kinder in 1952 exceeded that six times in ten starts.   The fewest batters Santa faced was 23; Kinder had less than that 14 times.   Kinder’s work assignment for the day was outside the range of Santana’s extremes 20 times in 23 game appearances.

            Santana worked on four or five days rest literally in every start all season, except for the first start.   He never once worked off-schedule, or even slightly off-schedule.

            Kinder worked on consecutive days once, and worked on one day’s rest seven times—but also had three gaps of ten days or more without taking the mound.

 

III.   Analysis and Comment

 

            As I said, this steady increase in the regularity of pitcher’s work is actually a result of five distinct factors, which are:

            1)  The disappearance of double-headers,

            2)  The division of pitching staffs into starters and relievers,

            3)  The abandonment of the practice of starter leveraging,

            4)  The expansion of the bullpens, and

            5)  The adoption of pitch limits.

 

            1)  The disappearance of double-headers.

            The pitchers of the 1940s made more than 30% of their game appearances in double headers.   Bobo Newsom was a very colorful pitcher who had a long career (1929-1953) with a lot of different teams.   In his career he made exactly 600 game appearances, of which 208—34.7%--were in double headers.   The percentages for other pitchers of his era were similar:  Bucky Walters, 32%; Joe Page, 33%; Bob Feller, 29%; Whitlow Wyatt, 31%; Hugh Casey, 31%; Spud Chandler, 36%.    Teams would not infrequently play back-to-back double-headers, and many, many times would have Wednesday and Thursday off for travel, then would play a double-header on Friday and another on Sunday.

            Over time, more than 90% of these double headers have disappeared.   Modern pitchers typically make 3% or less of their Game Appearances in double headers (Bartolo Colon, 2.7%; Mark Buehrle, 2.1%; John Smoltz, 3.0%; Adam Wainwright, 0.7%.)

            Obviously, the disappearance of double-headers has helped to regularize pitcher work schedules.

 

           

            2)  The division of pitching staffs into starters and relievers.

            There were 305 major league pitchers who made 100 or more game appearances between 1920 and 1939, every one of whom appeared both as a starter and as a reliever.   All of them appeared in at least 5% of their games as starters, and at least 5% as relievers.   Only a small fraction had starter percentages or reliever percentages over 80%.

            Carl Hubbell, Dazzy Vance, Ted Lyons, Herb Pennock, Lefty Grove and Dizzy Dean all made 20 to 30% of their game appearances in relief (during the 1920s and 1930s.   Some of them may have gone under 20% if you include the 1910s or the 1940s.)

            By the late 1930s, managers were beginning to divide their pitching staffs into starters and relievers.   I believe that it was Joe McCarthy who first divided his staff into pure starters and pure relievers, and I believe that he did this in the mid-1930s.   Further research might better inform us about that issue, but in any case, by World War II “pure relievers” or “dedicated relievers” had emerged, and starting pitchers were generally being excused from the responsibility of finishing up for weakening teammates.

            Let’s note that points (1) and (2) were totally unrelated, and that the division of pitching staffs into starters and relievers had nothing at all to do with the disappearance of double-headers.  In fact, the division of staffs into starters and relievers came at a time when the number of double-headers was increasing very rapidly, due to the addition of lights to the parks.   Owners of the 1920s and 1930s generally liked double-headers, which drew large crowds, but double-headers created scheduling issues because you generally had to start them no later than 3:00 in order to be sure you got both games in before darkness.  The addition of lights enabled owners to start Friday-night double-headers at 5:00 or later, which caused the number of scheduled double-headers to spike upward during the 1940s.

 

            3)  The abandonment of the practice of starter leveraging.

            Chris Jaffe, in Evaluating Baseball’s Managers (McFarland, 2010), writes a lot about “starter leveraging”, by which he means jiggling the starting schedule so that your best pitchers are matched up against the best teams, and left-handers are used most often against teams that may be  vulnerable to left-handers and in parks that may favor left-handers.   For example, he writes that (Al) “Lopez’s career helps trace the course and extinction of starting pitcher leveraging.   In Cleveland, he oversaw one of the best collections of starting pitching in history (and did not engage in). . .an especially impressive degree of leveraging, but it is noteworthy that any existed given the overall quality and depth of Lopez’s starting rotation. . .When Lopez came to Chicago, he maintained this approach, using ace Billy Pierce frequently against top rivals.   As noted (earlier), Lopez’s decision not to start Pierce against the Dodgers in the 1959 World Series helped bring about the end of starter leveraging (but Lopez continued to do some of it until 1963.)  In 1964, Lopez changed.   The reform that had blown across the baseball world over the previous few years finally came to Chicago.”

            Jaffe demonstrates by a number of such narratives, documented by sabermetric analysis,  that virtually all major league managers did juggle their rotations to leverage their best starting pitchers prior to 1955, but that this policy died in the late 1950s and early 1960s.

            In connection with this change, I want to focus on this part of the chart that I gave you earlier:

 

 

 

            The regularity with which pitchers have been used has increased in every decade, yes, and has probably increased in every five-year period, if we had enough data to measure that reliably.    But the time when it really increased was between 1960 and 1979.   From the 1920s through the 1950s, the regularity of pitcher use increased by a total of 6%.   In the 1960s it increased by 7% in one decade, and then by another 6% in the 1970s.   That 20-year period accounts for basically one-half the increase over the nine decades.

            What was really driving that change?  It wasn’t the disappearance of the double headers; that happened twenty years later.

            I go back to the Ralph Houk story that I told before, and to the other anecdotes from that same era, having to do with Lew Burdette, Joey Jay and others.   What was driving this change was that a very strong belief entered baseball, right about that moment, that pitchers would benefit from regularity of use.   Workloads for pitchers didn’t go down at that time.  If anything, they went up.    This wasn’t a belief in avoiding overworking pitchers; that belief became important about twenty years later.   It was a separate and distinct belief, in regular and predictable workloads.

            Look, for example, at the Regular Use Scores of Don Drysdale, from 1959 to 1963:

 

 

            Why?  Why did Drysdale’s regularity of use increase from 56% in 1959 to 80% in 1962, 88% in 1963?   Drysdale was the Dodger’s number one pitcher in 1959; he was their number one pitcher in 1962.   The Dodgers were in a tight pennant race, leading to a playoff, in 1959; they were in a tight pennant race, leading to a playoff, in 1962.   Drysdale was managed by Walter Alston in 1959, and he was managed by Walter Alston in 1963.

            What changed was simple:  between 1959 and 1963, the notion entered baseball, in a really vigorous way, that pitchers would perform their best when working on a regular schedule.   The Jim Bunning thing in 1964, when Mauch pulled Bunning off-rotation and the Phillies crashed and burned, cemented that link in the public mind.   The reality is, managers in the heat of the pennant race had done exactly what Mauch did in 1964 quite routinely for at least 50 years before then.   Sometimes it worked, sometimes it didn’t work.   Either way, it was never a big deal.

            The difference was, Mauch did it after the public had come to believe that pitchers performed their best given a regular work schedule.   That’s all; that’s the only difference between Mauch doing that in 1964 and Eddie Sawyer and Burt Shotton both doing it in the last week of the season in 1950.   In 1950 it was standard practice because there was no strong belief in the value of regular work schedules for pitchers; in 1964 it was a blunder because there was. 

The Longing for regular use patterns had entered baseball in a big way.

            In 1959 the league ERAs were 3.95 and 3.86.   In 1968 the league ERAs were 2.98 and 2.99.

            It is possible, in my opinion, that one thing that caused the pitcher domination of the 1960s was the regularization of pitcher schedules.    The re-definition of the strike zone in 1962/63 certainly contributed, yes—but the trend downward continued after 1963.   Most of it was after 1963.   People speculate on what caused this to happen.   One thing that may have caused this to happen was that pitchers in the 1960s were working in more regular patterns.

 

            4)  The expansion of the bullpens.

            Of course we all know how bullpens have expanded in the last 30 years, and many of us have written about this.   Many of us have written critically about this, and I don’t think I need to review that.

            The particular contribution of this research, however, is

1)  To establish a method to document differences between pitchers in the regularity of use, and

2)  To suggest the possibility that the desire to create regular work patterns for all pitchers is in fact the major driving force behind the expansion of bullpens.

 

            What multiple layers of relievers do for a pitching staff is to act as a shock absorber against the normal irregularity of the schedule.    When you have multiple relievers with subtle distinctions in their job assignments, it becomes more practical to expand or adjust those assignments to minimize stress points for starting pitchers and the critical two relievers.   Not only is this done, but that’s why those pitchers are there; that is the benefit that they give to a team.

            About eight/ten years ago, there was a debate going about the appropriate use of the closer.   In the middle of this debate I was hired by the Red Sox, and thus lost most of my ability to participate in that debate, because whatever I said could be played off the Red Sox in ways that were not helpful to the team.

            But what I really did not understand, until about 2005, was this.  If you look at Mariano Rivera, at Trevor Hoffman, Papelbon and others, these pitchers reach and sustain levels of effectiveness, relative to context, that are historically unprecedented.   No starter in history—not Koufax, not Randy, not Walter, not Grove, not anybody—was as effective, inning per inning and adjusting for context, as Mariano Rivera, and I’m not just talking about ERA.   Strikeouts, walks, home runs allowed. ..you name it, and the other guys are right there with him.  Why is that?

            It’s because the modern system of using Closers creates for these pitchers a role that is both very limited and very regular, extremely predictable.  Those guys throw maybe three dozen pitches a week—and do so at moments that can be predicted with a high degree of reliability.   They know when they are going to be in the game.   They can loosen up, stretch, push themselves to a peak, and they take the mound only when they are at that peak.   That enables them to reach fantastic levels of sustained effectiveness.

            I never understood this, until about 2005, and frankly, neither did any of you.   I know that you didn’t because of the things that you wrote.   People write about using your closer in the 7^th inning when the high-leverage situation arises in the seventh inning.   If you think you can do that, you don’t understand.   You can’t reach that peak of perfection on an irregular work schedule.   You’ve got to KNOW when you’re going to be needed.    If you used Mariano in the 7^th inning half the time, he would NOT be Mariano.

            And then there’s the press guys, who think that some guys can pitch in the 9^th inning, while other guys can only pitch in the 7^th. . .which totally misses the point, but then, we all miss half the points three-fourths of the time.

            What all of those other relievers are doing, really, is allowing the critical relievers to pitch only at those moments of peak effectiveness.   They’re load-levelers.  We’ve missed this; up to now.  We’ve left it out of the discussion.   Now it’s in.   That’s what this article is about.   I am arguing that you can’t really evaluate the worth of a deep bullpen unless you can deal with the advantage of creating stable workloads for other pitchers, because that’s really why it’s done.

           

            5)  The adoption of pitch limits.

            About which, again, enough has already been said—but again, I would argue that the real point of pitch limits is not to reduce the workload of starting pitchers, but to stabilize it.

 

            OK, but are we dealing here with five different phenomena, which combine to create a consistent evolution in the game, or are we dealing with a single phenomenon which has five different manifestations?   Or, looking again at this chart:

 

 

 

            Have we gone as far as we can go?   

Because if it is five different phenomenon, then it may be that these five distinct changes that were driving the Regular Use Percentages up have all played themselves out, and thus it may be that we are at the end of the line in this massive, 90-year shift.  

If, on the other hand, it is not five different phenomenon but five different manifestations of the same phenomenon—the Longing to create ever more stability in pitcher workloads—then we have to assume that this nine-decade trend may continue for several more decades until the Regular Use Percentages reach 90% or 92% or 94%.   And if that’s the case, then we have to ask ourselves, since these other manifestations of the shift have pretty much played themselves out, how this desire to create regular use patterns will change baseball in the future.

The disappearance of the double headers—that’s clearly a different thing.    That was not caused by the desire to create regular workloads; we can be pretty sure that it wasn’t.   But I would argue that the other four factors are, in reality, all simply manifestations of the same over-riding and over-whelming desire, on the part of managers and pitching coaches, to create stable workloads—thus

1)  that this desire has created massive changes in the game of baseball in the past, and

2)  we should expect that it will create additional significant changes in the game of baseball in the future. 

I can cite for you another change, caused by this same Longing, which is already occurring.    More aggressive use of the Triple-A roster as an extension of the major league roster.

I have worked for the Red Sox since 2002—eight years.  Over that time, I have seen a very substantial increase in the willingness, both of the Red Sox and other teams, to make a roster move to avoid stress points for pitcher usage.    Eight years ago, you could pretty much assume that the pitchers the other team had on their roster in the morning would be the pitchers they would have on their roster at game time.   Now, you can’t.   Same thing—front offices move pitchers up and down to protect the regular use patterns of the key pitchers on the team.

 

In working through this process, we skipped a critical question.    Is this a legitimate benefit?    Do pitchers in fact pitch better and/or get injured less when they have regular, predictable work schedules than they do when their workloads are less regular?    Or are we chasing a phantom here, bending ourselves out of shape and wrenching the game of baseball out of its traditional shape merely to pursue an “advantage” which is no real advantage at all?

Well, I don’t doubt that pitchers pitch better when they have a regular work schedule.   This is the universal assumption of all baseball men, very deeply ingrained; one can’t challenge that assumption based on a mere absence of evidence.   We don’t have any evidence or any reason to believe that this is not true, and, intuitively, I am strongly inclined to believe that it is true.

            I do have some serious questions, however, about whether the benefits derived from the regular usage of pitchers are substantial enough to justify the things that have been done in that effort.   The deeply rooted and constantly growing belief in the benefits of regular work for pitchers has changed baseball like a growing tree crushing and throwing aside a meddlesome sidewalk or even an old street.

That’s why I wanted to post this method:  To put us in a position to ask those questions.   Let’s assume that some managers use their pitchers in more regular patterns than other managers—if not now, then certainly at some points in the past.   Can we validate by sabermetric research that managers who are ahead of the game in this fashion get better results from their pitchers than managers who are behind the game?    Can we show that they have fewer injuries?  Can we show that they have better ERAs?  Can we show that they win more games?

Can we demonstrate that pitchers who are used on a regular basis are more effective than those who are not?  Can we demonstrate that they are injured less often?

If we cannot demonstrate that these benefits exist, then should we call into question the things that are done to pursue this benefit?   That’s getting ahead of ourselves; I guess we shouldn’t ask that question until we see what we have.

It may be that the method that I have outlined here will be inadequate to the tasks set before us.   Better methods may be needed.   These issues will be entwined with issues of fatigue and workload, and it will be in some cases impossible to tease them apart, so that some of these questions will prove to be un-answerable.   It just seems to me that this is too large a subject to ignore, that we have ignored it for too long, and I’m trying to move the discussion off the dime.