Managing pitching staffs in the 1920s and 1930s

I’ve had a fun stretch of light but meaningless columns around here, but it’s time to get back to the more important stuff.

Don’t run away yet – this column isn’t guaranteed to be boring. (I’m not guaranteeing it won’t be, but that’s a whole other issue.) Anyhow, one of my main baseball research obsessions revolves around managers. It’s my biggest one, given how I wrote a book on the subject and all.

In the process of writing my book, I realized that it was guaranteed to be incomplete. I’d have to turn it in for publication, obviously, but as long as I kept thinking about the subject matter, new ideas could come to me, that could no longer go in the book. What’s more, new data would come to light that I didn’t have when writing it.

This column looks at one data dump that I didn’t use in my book that I recently looked into. When I collected info for my book, Retrosheet hadn’t yet begun putting up team splits info for very many seasons before the mid-1950s. (I think they had one or two seasons up, but not enough to draw any meaningful conclusions from.)

That wonderful site has recently finished giving us all the info on the 1920s and 1930s. This allows me to dive in and have some idea how managers ran their staffs way back when. For me, this is loads of fun.

There’s a lot of stuff here, so I’ll just go scavenging for whatever random impressions I come across.

So you know, I’m looking at splits for starters and relievers in regard to quantity and quality. For both bullpen and starters, I use park-adjusted ERA for quality. I use a barely modified form of IP for quantity for both. (It’s IP/GS for starters, and percentage of a team’s innings gobbled up by the bullpen for relievers.)

Obviously, the above areas tell us quite a bit about the talent on hand, but over a period of time, they also tell us about what managers like to do and what sort of teams they prefer to handle. Managers had significantly more control over the rosters in those years than now. GMs normally took care of the farm system and managers figured out which veterans the team should have.

Anyhow, based on all that, here are some of the more interesting things I came across about the game’s more significant managers and how they ran their pitching staffs in these years:

Connie Mack and the bullpen

Connie Mack apparently really put a great deal of emphasis on his bullpen. He relied on it perhaps more than any other prominent manager of this period. His bullpens topped the league in innings pitched four times, were runner-up seven more times, and came in third place another a half-dozen times in his 20 years managing during that period.

Sure, there were years the A’s had some terrible starting rotations that necessitated Mack go to his bullpen early and often, but the A’s actually had a winning record in nine of these 20 years. Mack just liked using his bullpen.

For example, when the A’s won their first world title in 15 years in 1929, their bullpen threw the second most innings of any team in the league, despite the fact that the starting rotation possessed the best park-adjusted ERA of any team in the league.

The next year, when Philly repeated as AL champs, the A’s relievers threw about 50 innings more than the other AL teams averaged, despite their starters having the second best ERA+ in the AL.

This actually fits in with something I noted in my book: Connie Mack was the king of swingmen. He often had a pitcher appear in at least 40 games per year with more than 15 starts and relief appearances each. What’s more, he frequently used his best pitchers as swingmen, as Lefty Grove, Eddie Rommel, and Rube Walberg most frequently filled this role for him. Whereas most managers would want their pitchers to complete what they started, Mack would rather take them out of one game early so they could help out in another game later.

It worked fairly well for him, as he had some terrific bullpens. He also had some pretty bad bullpens (especially in the later 1930s), but a quality bullpen was the hallmark of a quality Connie Mack team. From 1925-31 (when this edition of the Mack A’s peaked), his bullpen had the best or second-best park-adjusted ERA every year except 1927, when they were merely third best.

Altogether, on five separate occasions Mack’s bullpens were first or second in both innings and ERA+. His 1926 squad may be the best bullpen of the era. Despite performing in one of the league’s best hitters’ parks, they posted an ERA of 2.70 in a league with an overall ERA over 4.00. It was only slightly over four, but still – it was. By my back-of-the-envelope calculations, the 1926 A’s had a bullpen ERA+ of 155. Not too shabby.

A Hardball Times Update
Goodbye for now.

I have something called the Tendencies Database that I use for my book to really get a good feel for how managers rate in certain ways. (I’m trying to not use it too much here because it takes too long to explain, but you can read about it here if you’re curious.) When the Retrosheet results for bullpen quantity and quality for 1920-39 are put through it, Mack depended on his relievers more than any other prominent manager of his day.

So that’s how he did it: Charlie Grimm and starting pitchers

Charlie Grimm is something of an enigma as a manager. In his book on managers, Bill James notes that Grimm’s teams were very successful throughout his entire time even though Grimm himself was never considered to be a genius. That said, he consistently did as well as one could hope, which is the highest compliment you can pay a manager. In my book, I was very happy with the section I wrote on him (some of my proofreaders told me it was one of their favorite parts), but I still had no really clear idea how he did it.

The new Retrosheet data helps. Grimm built his teams around superior starting pitching. In the six campaigns he ran the Cubs for the full season in this period, their starters had the best ERA+ three times, and were always in the top half of the league. What’s more, they also threw their share of innings as well. Twice they topped the Senior Circuit in most IP per start and did well in that category in other years as well.

In 1933, his first full season as skipper, his Cubs were first in starter innings and second in ERA+. Two years later, when they won the pennant under him, the results were reversed with the best ERA+ and second-highest innings total. In 1936, they led the league in both categories.

According to the Tendencies Database, Grimm got about as much out of his starting pitchers as any manager in baseball history, including Bobby Cox. That isn’t entirely a fair statement, as Cox has almost 30 years managing while Grimm only has 10 years with this info (including four years with the Braves in the 1950s), but there is an overall theme: His teams consistently won with starting pitchers.

I found the above Grimm info so stunning I decided to do some more digging on his pitchers and found out something interesting. Here’s one interesting list I have laying around. Among all managers who lasted at least 10 seasons as their team’s primary skipper, the following had the highest percentage of ERA-qualifying starters equal or better the league strikeout rate. In other words, it lists the managers who most relied on fireballers. (Listed are the managers, the number of pitchers who qualified for the ERA title under them, the strikeout better-ers, and percentage):

Manager	      Qual	Better	%
Charlie Grimm	60	44	73.3%
Wilbert Robinson  80	58	72.5%
Davey Johnson	48	34	70.8%
Walter Alston	81	57	70.4%
Joe McCarthy	91	63	69.2%

No one tops Grimm. Not in all baseball history. That’s rather impressive. And it fits in perfectly with the Retrosheet info. The man won with dominating starting pitchers.

Billy Martin in historical perspective

The new info can help present information about one of the most well known examples of how a manager used his pitching staff: Billy Martin and the 1980 Oakland A’s. Martin rather infamously pushed the entire staff harder than made sense and all the young hurlers broke down shortly afterward.

When I wrote my book I had team splits info for every year from 1956-2008. Not only did the 1980 Oakland starters have the most IP by any team in that period, but the gap between first and second was equal to the gap between 75th place. It isn’t as bad when you look at IP/GS, because Martin’s 1981 A’s came in second place.

Well, how do the 1980 A’s fair when compared to the teams from long ago? Well, incredibly, Oakland’s 1,261.1 innings logged by their starters stills tops the field. Ye, gods.

That isn’t really fair because the season is longer now. Besides, I already noted IP/GS is a better way than raw IP. When you look at innings per start, a handful of teams do nose out Martin’s bunch, as the chart below reveals:

Year	Team	IP/GS
1923	NYY	8.03
1922	NYY	7.99
1920	CWS	7.96
1920	BRVS	7.90
1920	PIT	7.81
1932	NYY	7.81
1920	CIN	7.81
1920	BRK	7.81
1980	OAK	7.78

Notice something there? They are almost entirely made up of teams from the early 1920s. That’s interesting. There have been three periods in baseball history when workloads for starting pitchers declined noticeably: 1) In the 1890s when the pitchers were pushed back to 60 feet, 6 inches; 2) In the early 1920s when the lively ball came out; and 3) In the 1990s when pitch counts became all the rage. In each instance, the game changed in a few years, causing managers to adapt to how they used pitchers.

So how did Martin run the 1980 A’s? Like someone who hadn’t fully realized the deadball era had ended.

Joe McCarthy’s hard-pressed starting pitchers

The above list brings up one other point I think is worth making: how Joe McCarthy handled his starting pitchers. After all, the only non-Martin team in the above list was McCarthy’s 1932 Yanks.

While that season was extreme for him, he did push his starters as hard as he could on a regular basis. In fact, every year that he ran the Yankees that I have information for, the team’s starters were first or second in the league in innings per start. That’s impressive.

Sure, they had good arms that made it likely that anyone would get a lot of innings out of them, but it’s a bit more than that. The best year exemplifying McCarthy’s willingness to push his starters hard actually came in 1933, the year after he pushed them so incredibly hard.

In 1933, when the rest of the league averaged about 6.9 innings per start, the Yankees starters topped 7.3. This was especially noteworthy because they were actually pretty bad that year. You’d never expect that from a Yankees staff, but it’s true: Despite pitching in the best pitchers’ park in baseball, the Yankees starters had an ERA barely better than league average (4.18 to 4.28). In fact, when their ERA is adjusted for park, they featured the worst staff in the league. Yet they still led the league in innings pitched.

That’s the only time from 1920-39 the team whose starting pitchers had the worst overall ERA+ led the league in innings. Heck, in the more than 150 leagues I have splits for, there are only two other times that’s happened (the 1952 Phillies and 2000 Orioles). Suffice it to say, I think this tells us something about how McCarthy liked to use his starting pitchers.

Now riddle me this: In 1932 McCarthy pushed his staff as hard as any manager in the last 85 years. The next year, they flopped on the whole. Coincidence? Could be, but I dunno. I think McCarthy is the greatest manager of all-time, but everyone makes mistakes, and this might’ve cost the Yanks their shot at the 1933 pennant.


11 Comments
Oldest
Newest Most Voted
Inline Feedbacks
View all comments
Tom Farmer
14 years ago

Have you done any work on how Billy’s pitchers did in terms of (1) Pitches thrown and (2) did you look at them in terms of Bill James’ index of pitcher destruction – i.e. big innings with 30,40 pitches, that sort of thing?

Though it has been often said that high workload destroys pitchers, I’m not entirely sure at all that it has been “proven” in any conclusive form.  It seems to be something a lot of people believe, but is it like how people used to focus on batting average, not walks?  Is it an old writers tale, or is it legitimate, that lots of work ruins pitchers?

Mike Emeigh
14 years ago

The main differences between the 1932 and 1933 Yankees were that (1) the younger replacements for Pennock and Pipgras (Russ Van Atta and Don Brennan) didn’t work out; (2) Allen was hurt and less effective; and (3) Ruffing reverted to previous form – his 1933 season is not a whole lot different from 1930 and 1931, and 1932 sticks out as the outlyer.

Chris J.
14 years ago

Tom,

Not sure what you’re looking for.  Pitching puts stress on the arm, and too much stress can cause an arm to break down.  I’m a bit skeptical of reductionism in which having pitchers go over Limit X (whatever Limit X might be) is inherently evil, but the general principle that pitching leads to pitching injuries seems sound to me.  Martin’s 1980 A’s were the hardest worked pitchers in the the 80 years.

Mike,

Don’t forget about Danny MacFayden, who was a flaming ball of poo in 1933.  I know he was in the bullpen almost all the year, but my own sense (based on nothing more than that, admittedly) is that there’s a link between 1932 and his going to the bullpen in 1933.  He’d been a solid starting pitcher for years in the past, and would be one again in the future.  So why push him out of that role in 1933?  My hunch is that his arm was sore.  And it would explain why he was so dreadful that year.

It also explains why the Yanks had to break in a pair of new starters as well.  Van Atta and Brennan were in their late 20s without MLB experience by that time.  Pipgras was able to start 21 games anyway that year, almost as many as the year before.  Allen and MacFayden were the key men from 1932 far less effective in 1933. 

Interesting point about Ruffing.  What a weird career he had.  Based on how well he pitched in 1932 and much later, I wonder if he was one of those who had chronic dead arm when younger but outgrew/survived it?  Not only was his ERA+ much better in 1932, but he even led the league in strikeouts.  In 1933, he wasn’t blowing them away, yet he still had enough talent to be great when he was old.

Gilbert
14 years ago

Might be interesting to correlate old time bullpen usage with quality of PH on the team, and whether starters were relieved mid-inning or PH’d for.  If you only have one decent bench bat you might be less likely to deploy him in the 6/7th in a close game if it might leave you in the bottom of the 9th down a run with runners in scoring position having to put in your Mendoza to PH for the pitcher.

gbewing
14 years ago

love your work, I continue to learn more about history and clarify my own questions on managers and eras.

Rob Bonter
14 years ago

“The next year (1930) when Philly repeated as AL champs…”  should read:  “The next year, when Philly repeated as WORLD CHAMPS…”

Mike Emeigh
14 years ago

Gilbert:

Old-time bullpen usage was overwhelmingly the result of a starter leaving with his team trailing, or getting bombed. The ace reliever model didn’t really come into vogue until the mid-50s – you had the occasional Marberry or Murphy, to be sure, but using relievers to protect leads was not part of the strategy until the Dodgers and Yankees started doing it regularly in the late 40s and early 50s.

Chris J.
14 years ago

Rob,

Good catch.  I confused who won the 1930 and 1931 World Series.

Mark Ruckhaus
14 years ago

I’m wondering if this is all a self-fulfilling prophecy kind of thing.

Judging by 1932, three of the four best teams in the league were tops in CG. And all four were pretty decent thanks to there being three “have-nots” in the AL that year.

The top three teams in the league in CG had totals in the mid 90s. Three teams were in the mid 60s and the two worst teams in the league, Chicago and Boston, came in at 50 and 42, respectively.

Now, alleged common knowledge says that the five-man rotation was something which came about in the past 40 years or so. But, judging by the number of starts by each team’s top starters, it indicates that most managers were using something close to a five-man staff back in ‘32. OK, those starters pitched a bit in between (which, in theory, could also be done today as a starter might throw a couple dozen hard pitches in the bullpen two or three days after his previous start).

As the Yankees won the league by 13 games and outscored the opposition by nearly two runs a game, maybe the Yankees’ starters weren’t really busting it out there. After all, Gomez’ year isn’t indicative of a 24-7 pitcher.

The team I find amusing is not the Yankees but, rather, Washington. They were slightly below average in CG but won 93 games and won the league the next season.

Walter Johnson worked Alvin Crowder like a horse. But so did Joe Cronin in ‘33. By ‘34, Crowder was shot. With the exception (barely, because he also relieved 13 times) of Monte Weaver, Johnson ran a revolving door.

As much as it’s a self-fulfilling prophecy, maybe it’s a philosophy thing too. McCarthy had three regulars he could parade out to the mound on something which looked like a five-man rotation with three other guys who took at least a semi-regular turn while Johnson had one guy he really depended on, another who took his regular turn (Monte Weaver) while, maybe with the exception of Marberry, saying a prayer for the rest of his staff.

Chris J.
14 years ago

Mark,

I’d agree that it’s a self-fulfilling prophecy that the 1932 Yanks would have a lot of complete games, but the highest IP/GS of any team in the last 87 years?  That’s something a bit different.

Interesting comment about Johnson.  He managed like he learned under Clark Griffith.  When he managed in MLB for 20 years, Griffith’s teams were usually near the bottom of the league or at least bottom half in CG.  He used his bullpen more than anyone else in the first decade of the twentieth century.

But he always rode he main ace hard.  In 1918, Walter Johnson became the next-to-last man to qualify for an ERA title while completing all his starts.  The team as a whole finished fifth in the league in CG.

That’s nothing.  His 1904 New York squad completed 123 games – easily the fewest in the league (everyone else had at least 134) despite the presence of Jack Chesbo, who completed 48 CG – the most by any pitcher since 1893.  (On a purely self-serving note, let me state all this can be found in my book, “Evaluating Baseball’s Managers, 1876-2008”). 

At any rate, there really was no such thing as a pitching rotation in the 1930s.  Even when GS totals were around the level of a rotation, it doesn’t fit.  Part of a rotation isn’t workload, but usage patterns: guys rotate in a set ABCDE order.  That never happened back then.  Among other things, there was no such thing as a clear starter/reliever split, as all starters could be called on to work out of the bullpen. 

Nice comment.  Appreciated it.

bob sawyer
14 years ago

The big difference between the 20s and 30s and today is the frequency of double headers. Whenever two pitchers work on the same day it makes sense that the manager will start the higher quality pitcher again before the lower quality. At the same time, double headers create a requirement to use a number 5, number 6, or even a number 7 starting pitcher because every one else was tired.
Joe McCarthy’t reaction to the the poor performance of his starfing pitcher in 1933 was to separate the starting and reliever roles. The 1935 staff had 5 guys who rarely relieve and 4 guys who rarely start. Bill James believes this may be the first such staff in Major league history.