The Four Man Rotation Redux

On my recent vacation in Florida I had a chance to pick up The Neyer/James Guide to Pitchers. Wow! In addition to providing an entertaining (if brief) history of pitching followed by histories of each pitch and biographies of some underrated pitchers Rob Neyer and Bill James provide an encyclopedia of what pitchers threw. So, for example if you want to know what Fergie Jenkins threw you can flip to his entry and see the following:

"Pitches: 1. Four-Seam Fastball 2. Two-Seam Fastball 3. Slider 4. Curve 5. Forkball (as change)"

They then document the source of this info along with quotes from other sources about the style or way in which the pitcher threw. Absolutely fascinating - a must have for those interested in baseball history and particularly for those who love pitching. . BTW, I see that Neyer has now published some errata for the book.

Pitcher Abuse?
The only controversial aspect of the book and topic that has been hot in the sabermetric community as of late is an essay by James called "Abuse and Durability" and a rebuttal by Rany Jazayerli and Keith Woolner from Baseball Prospectus titled "A Response in Defense of PAP" (I blogged briefly about PAP (pitcher abuse points) recently). These essays reflect James' criticism of the methodology used by Jazayerli and Woolner and their response. In short, James uses a set of comparison studies to compare pitchers marked as abused by PAP against other pitchers and tries to determine how the value of each set of pitchers changes over time. He concludes that the "pitchers that Jazayerli and Woolner identify as 'abused' declined in value by 10% in the following season. That can't possibly be an unnatural decline. This system simply does not identify at-risk pitchers, period." [emphasis in the original]

This idea that pitchers could be abused and the modern emphasis on pitch counts can be traced to the work of one-time sabermetrician for the Texas Rangers Craig Wright in his book The Diamond Appraised as well as to James himself in the early Baseball Abstracts (in 1982 he discusses Billy Martin's pattern of abuse). In fact, James quotes Wright's book in criticizing PAP. I pulled out my copy (no sabermetric library should be without one) and recalled that Wright did his study based on what he called Estimated Batters Faced Per Start (BFS) calculated as:

BFS = ((IP * 3) + H + BB) / ((GS + ((G-GS) * .5)))

Wright then identified those pitchers with high BFS numbers at young ages to see how their careers progressed versus other pitchers with low BFS values at similar ages (typically 18-24). After doing his comparison study on pitchers with high BFS numbers in the teens versus those who were not worked as hard Wright lays out the following guidelines for pitcher use (p211):

"1. As a teenager, a pitcher should not be allowed to throw two-hundred innings seasons or have a BFS over 28.5 in any significant span (150-plus innings)."
2. A teenage pitcher should not start on three days rest..."
3. For ages twenty to twenty-two, they should average no more than 105 pitches per start for the season (105 pitches is the rough equivalent of a 30.0 BFS). A single game ceiling should be set at 130 pitches.
4. For age twenty-three to twenty-four, the restraints can be eased up, but their season average should stay under 110 pitches in most cases. The single game ceiling can be jumped to 140 as long as the pitcher is still strong."

To Wright it is clear that the age of the pitchers is the crucial element coupled with the average number of pitches over a prolonged period as well as the maximum number of pitches in a start. It is this crucial fact that James points out when he says "Craig was concerned mostly with the effect of hard usage on young pitchers." [emphasis in the original]. Although James calls the omission of a strong age component a "marginal concern" in the PAP system it appears to me to be the central reason why James doesn't detect abuse. His studies simply look at the list of pitchers selected as most abused by the PAP system which consist of pitchers of all ages, many of whom since they are mature, aren't really impacted by high pitch counts as discussed in this interesting article on mlb.com by Will Carroll.

Adjustments for age have been made to PAP since 2001 as pointed out in the rebuttal but James downplays their importance in his criticism. Woolner and Jazayerli, however, in defending their work say that "Bill is almost certainly right when he states that almost all of the risk inherent in throwing too many pitches occurs in a pitcher's formative years. The connection between overuse and catastrophic injury seems to drop off quickly past age 25 or so." The PAP system, however, still includes all pitchers and so likely "labels" some pitchers as abused who are mature and can handle the work load.

Checking For Abuse
Because Wright performed his study in the late 1980s I wanted to check and see if his recommendations had been followed so I ran some queries to test points 1,3, and 4.

To test point (1) I ran a query showing those teenage pitchers with 200 inning seasons or a BFS over 28.5 in 150 innings or more since 1960 (of course now that batters faced is known I did not need to rely on Wright's formula). The only two pitchers that match these criteria since 1960 were a pair of 19 year olds - Gary Nolan for the Reds in 1967 (226 IP and BFS of 29.1) and Wally Bunker in 1964 for the Orioles (214 IP, 29.3 BFS). Nolan's arm injury is well known although Bunker also developed a sore arm in 1965 and was never the same winning only 41 more big league games. The sample size is too small to say anything definitive of course.

To test point (3) I ran a query showing those pitchers who threw more than 105 pitches per start using Tangotiger's basic pitch count estimator (3.3*((IP * 3) + H + BB))+(1.5*SO)+(2.2*BB) but replacing (IP*3)+H+BB with actual batters faced) at ages 20 through 22 (as of July 1) in over 30 starts since 1960. There were 51 seasons that qualified.

Name            Year   Age         GS     IP           BFS     P/GS

--------------- ------ ----------- ------ ------------ ------- -------

Jose Rijo       1986   21          26     193.7        32.9    127.9

Mark Fidrych    1976   21          29     250.3        34.3    122.4

Bert Blyleven   1973   22          40     325.0        33.0    122.3

Frank Tanana    1975   21          33     257.3        31.2    120.0

Frank Tanana    1974   20          35     268.7        32.2    118.8

Vida Blue       1971   21          39     312.0        30.9    118.7

Dwight Gooden   1986   21          33     250.0        30.9    116.4

Don Gullett     1973   22          30     228.3        31.4    116.3

Dwight Gooden   1985   20          35     276.7        30.4    116.2

Jon Matlack     1972   22          32     244.0        31.3    116.2

Fernando Valenz 1982   21          37     285.0        31.2    116.1

Mike Nagy       1969   21          28     196.7        31.1    115.4

Denny McLain    1965   21          29     220.3        30.4    114.9

Larry Dierker   1968   21          32     233.7        30.7    114.8

Ray Sadecki     1961   20          31     222.7        30.6    113.8

Bert Blyleven   1972   21          38     287.3        30.5    113.6

Greg Maddux     1988   22          34     249.0        30.8    113.0

Catfish Hunter  1967   21          35     259.7        30.1    113.0

Jerry Garvin    1977   21          34     244.7        30.8    112.6

Dean Chance     1963   22          35     248.0        30.2    112.6

Dave Rozema     1977   20          28     218.3        31.8    112.5

Dave Boswell    1967   22          32     222.7        28.8    112.1

Floyd Youmans   1986   22          32     219.0        28.3    110.9

Ismael Valdes   1995   21          27     197.7        29.8    110.8

Britt Burns     1980   21          32     238.0        30.3    110.6

John Smoltz     1989   22          29     208.0        29.2    110.5

Bert Blyleven   1971   20          38     278.3        29.6    110.0

Bill Butler     1969   22          29     193.7        28.8    109.9

Burt Hooton     1972   22          31     218.3        29.5    109.6

Ramon Martinez  1990   22          33     234.3        28.8    109.6

Roger Erickson  1978   21          37     265.7        30.3    109.5

Jim Palmer      1966   20          30     208.3        28.9    109.4

Kerry Wood      1998   21          26     166.7        26.9    109.4

Joe Coleman     1969   22          36     247.7        28.9    109.1

Ed Correa       1986   20          32     202.3        27.7    108.9

Ray Culp        1963   21          30     203.3        28.0    108.7

Mark Lemongello 1977   21          30     214.7        30.3    108.1

Carlos Zambrano 2003   22          32     214.0        28.3    107.9

Denny McLain    1966   22          38     264.3        28.4    107.4

Steve Dunning   1971   22          29     184.0        28.0    107.4

Gary Nolan      1970   22          37     250.7        28.4    106.9

Don Robinson    1978   21          32     228.3        29.3    106.9

Clay Kirby      1970   22          34     214.7        27.9    106.7

Wayne Simpson   1970   21          26     176.0        28.1    106.4

Dave Rozema     1978   21          28     209.3        30.3    106.2

Bret Saberhagen 1985   21          32     235.3        29.1    106.0

Catfish Hunter  1968   22          34     234.0        28.5    106.0

Dennis Eckersle 1976   21          30     199.3        27.4    106.0

Storm Davis     1983   21          29     200.3        28.7    105.9

Les Cain        1970   22          29     180.7        27.3    105.7

In these 50 seasons are 42 unique pitchers. It is telling that only four pitchers have made the list since 1990 which is possibly an indication that organizations are more careful and indeed are heeding the advice of Wright at least indirectly. Of course, the things that jumps out to me are the two recent additions to this list, Kerry Wood and Carlos Zambrano of the Cubs. Wood did incur a serious injury that wiped out the end of 1998 and all of 1999 and suffered a sore tricep this season. Zambrano has yet to be injured but his precense on the list and his continually high pitch counts make me a little uneasy. (As an aside it should be noted that the pitch count estimator estimated 107.9 pitches per start for Zambrano in 2003 when he actually had 106.4).

To test point (4) I ran a query of those pitchers aged 23 or 24 who threw more than 115 pitches per start in 25 starts or more. This produced a list of 36 seasons with 30 unique pitchers.

Name            Year   Age         GS     IP           BFS     P/GS 

--------------- ------ ----------- ------ ------------ ------- -------

Rick Sutcliffe  1979   23          30     242.0        33.9    124.7

Jim Bouton      1963   24          30     249.3        33.5    124.2

Fernando Valenz 1984   23          34     261.0        31.7    122.1

Jim Hughes      1975   23          34     249.7        32.4    120.7

Dick Drago      1969   24          26     200.7        32.6    119.4

Bert Blyleven   1975   24          35     275.7        31.5    119.4

Joe Coleman     1970   23          29     218.7        31.7    119.2

Jack Hamilton   1962   23          26     182.0        31.5    119.0

Denny Lemaster  1963   24          31     237.0        31.5    119.0

Jim Maloney     1963   23          33     250.3        30.6    118.9

Tommy Greene    1991   24          27     207.7        31.7    118.7

Lynn McGlothen  1974   24          31     237.3        31.9    118.4

Jim Merritt     1967   23          28     227.7        32.5    118.4

Livan Hernandez 1998   23          33     234.3        31.5    118.3

Jon Matlack     1974   24          34     265.3        31.6    118.0

Frank Tanana    1977   23          31     241.3        31.4    117.8

Fernando Valenz 1983   22          35     257.0        31.3    117.5

Mel Stottlemyre 1965   23          37     291.0        32.1    117.5

Clay Kirby      1971   23          36     267.3        30.8    117.4

Don Sutton      1968   23          27     207.7        31.4    117.3

Denny McLain    1968   24          41     336.0        31.4    117.3

Dean Chance     1964   23          35     278.3        31.2    117.3

Bert Blyleven   1974   23          37     281.0        31.1    117.2

Dennis Eckersle 1978   23          35     268.3        32.0    117.1

Andy Messersmit 1969   23          33     250.0        30.5    117.1

Jim Kaat        1962   23          35     269.0        31.8    117.0

Joe Coleman     1971   24          38     286.0        30.9    116.8

Larry Dierker   1970   23          36     269.7        31.4    116.7

Dave Righetti   1982   23          27     183.0        29.8    116.1

Steve Barber    1961   23          34     248.3        30.6    116.0

Bret Saberhagen 1987   23          33     257.0        31.8    115.7

Chuck Estrada   1961   23          31     212.0        29.8    115.5

Dick Ellsworth  1963   23          37     290.7        31.4    115.4

Stan Bahnsen    1968   23          34     267.3        31.5    115.4

Sam McDowell    1966   23          28     194.3        28.8    115.1

Blue Moon Odom  1968   23          31     231.3        30.6    115.0

Once again very recent pitchers don't really show up on the list with the exception of Livan Hernandez in 1998.

In looking at these lists it is apparent that something has changed in the workloads of young pitchers since the late 60s and 70s. I believe there are three primary reasons:

The focus on pitch counts is the most obvious explanation. Organizations seem to be protecting their young pitchers better by not working them so hard when they first reach the majors. A case in point is Royals rookie Zack Grienke. In 14 starts Grienke has yet to throw more than 109 pitches or pitch more than 7 innings. I believe the Royals are doing the right thing with Grienke because they are not allowing him to pitch when he is tired. It's clear from watching his starts that he tires around 90 pitches and so is likely more prone to injury as his mechanics degrade past the point of fatigue.

A secondary explanation is the use not only of closers which came into vogue in the late 1970s but the addition of "setup men" in the 1990s. A starter on a team with a solid bullpen rarely needs to go more than 7 innings these days which puts downward pressure on their pitch counts down. These two factors together are easily demonstrated by looking at the number of compete games in the major leagues that has gone from 20% in 1980 to just 4% in 2003.

A third factor is the increased run environment that pitchers have toiled in since 1993. More runs means managers taking pitchers out of games sooner thereby limiting innings and pitch counts.

These last two points are illustrated by the fact that if you look at the top 100 seasons since 1960 ranking them in descending order by IP per start you won't find any seasons after 1989. In fact, it isn't until 147th place that Roger Clemens' 1997 season appears when he averaged 7.8 innings in 34 starts.

But the real question is whether this altered pattern has actually decreased the number of injuries to starting pitchers. Jazayerli and Woolner suggest that the jury is still out but are hopeful in their rebuttal stating that "A revolution in the management of starting pitchers is underway, and the early signs suggest that the revolution may well lead to fewer injuries." They don't, however, list what these signs are. This question is also difficult to analyze since teams use the disabled list more frequently today than in the past in part because diagnostic technology has improved and therefore precautionary DL stints are more common, and partly because of the large investment that players represent that encourages a better-safe-than-sorry approach.

I think all parties agree that its clear that abuse along the lines documented by Wright indicates that young pitchers can and were overworked fairly routinely in the past which often led to serious injuries. However, if evidence of reduced injuries to young starters in the present era fails to materialize, it could well be that James is correct in surmising that most injuries to pitchers both young and old are catastrophic and not the result of overuse. The truth probably lies in the middle somewhere - young pitchers are prone to both overuse and catastrophic injuries while mature pitchers typically incur catastrophic injuries. Understanding the relative frequency of each is the key to formulating studies and systems that might detect it.

A second explanation as to why recent pitchers don't show up on these lists assuming that injury frequency has not lessened, might be the alteration in workload from a very young age. Since young pitchers have been so closely monitored throughout their careers from little league on up, their arms are not as strong as pitchers of the past. In that case the prevelance of arm injuries might not change but would simply be triggered at lower and lower pitch counts. For example, in the 1970s a game of 140 pitches might be considered the borderline for a mature starter whereas today it might be 120 pitches and tomorrow 100 pitches.

Four Man Rotation
And finally, this leads us to the title of this post. Could a return to the four man rotation actually help the situation? Wright found in his study that pitchers are not less effective on three days rest than on four when looking at data from 1986-87 and advocated a return to the four-man rotation. This was backed up by a study done by Jazayerli using data from 1978 to 1995. Of course, far fewer pitchers ever pitch on three days rest today. The last manager to be a strong proponent was Earl Weaver in the early 1980s and the last experiment with it I can remember is the Reds Bob Boone's in 2003 which ended by mid season despite support from pitching coach Don Gullet and assistant GM Brad Kullman.

It seems to me (and others I've read) that the focus on reducing pitch counts has become married to the assumption that more rest between starts is also good for young pitchers. You could easily see where this might be the case since both reduce the number of pitches a pitcher throws over the course of a seaon. But is it just possible that for most pitchers throwing fewer pitches (meaning a maximum of 105 for young pitchers and 120 or so for mature pitchers for example) more frequently (every four days instead of five) will actually build arm strength, endurance, and lead to fewer injuries in the long run? Of course it will also have the side benefit of allowing better pitchers to throw more innings thereby producing more wins for the team that employs the four-man rotation. I'd love to see some team try it in 2005 but now that the pendulum has swung in the opposite direction it will likely take a while for it to swing back.



Other reading on pitch counts:
Kicking the crutches out
What Pitch Counts Hath Wrought
What Pitch Counts Hath Wrought: Part Deux
Swinging from the Heels: Painting a Fake Tunnel on a Blind Alley