August 20, 1999
PAP - A Historical Perspective
How badly were pitchers abused in the 1950's?
Special thanks to Retrosheet
Injuries to pitchers are not a new phenomenon. They date as far back as the rule change that allowed pitchers to throw overhand, and so do the attempts to restrict the workload of pitchers to a safe level.
When Old Hoss Radbourne reminisced about winning 60 games (now reported as 59) in that fine summer of 1884, he talked about how his right arm was so sore he found it almost impossible to brush his hair each morning. He survived that season, but never again approached the dominance he showed that year. Amos Rusie, "The Hoosier Thunderbolt", dominated the high-octane National League in the 1890s like Pedro Martinez does the AL today. Rusie threw a never-to-be-surpassed 482 innings in 1893, the first year that the pitching distance was moved back to 60 feet, six inches. He threw a grand total of 22 innings in the major leagues after his 28th birthday. Ed Walsh, who threw a 20th-century record 464 innings en route to his 40-win campaign in 1908, was all but finished four years later at the age of 32.
There are volumes of similar stories involving young hurlers, although many are less well remembered because many of the subjects' careers ended before they became famous. Monte Ward, who won 69 games before he turned 20, was washed up as a pitcher by the time he was 24. He went on to forge a successful career at shortstop and second base, then took a turn as the founder of the Players' League and the manager of the New York Giants, all before retiring to practice law at the age of 34. Bob Feller, the most successful schoolboy player of all time, was just a shell of himself after he turned 33. Herb Score, whose career was supposedly cut short by the line drive he took off his head, had in fact already injured his arm before the beaning. Score completd 27 of 59 starts and averaged 238 high-pitch innings over his rookie and sophomore seasons.
Identifying the Problems
We can't undo the mistakes of the past, but we can learn from them. One of the advantages of the Pitcher Abuse Points system is that its standards remain roughly constant through different eras. A .280 batting average means very different things in 1930 and 1968, but 120 pitches are 120 pitches in any era.
Well, not quite; there is a lot of evidence to suggest that in the dead-ball era, pitchers would not throw with maximum effort on every pitch. But ever since the home run became a promient factor in the game, it is a fairly safe assumption that pitchers have been putting everything they have into each pitch. If there has been a subtle trend towards maxing out on every throw, it would have been balanced by the increase in resources available to pitchers: improved methods of exercise, weight training and a full-time training staff to counteract the greater strain on their arms.
If, say, we had detailed pitch records for a team from 40 or 50 years ago, we could apply the principles of PAP to measure the workload of pitchers from that era and see if we could learn anything about how their workload affected both their performance and their health.
Well, with great thanks to David Smith and his band of merry men at Retrosheet, we have such detailed records. Most of the play-by-play box scores that Retrosheet has recorded and archived do not, unfortunately, have pitch-by-pitch data. But from late 1947 until mid-1964, Allan Roth, the first full-time statistician in the employ of any team, meticulously kept pitch-by-pitch information on every game the Dodgers played for Branch Rickey. (There's a couple dozen games from 1950 and 1951 for which the data is missing.) And with the help of Retrosheet, we can now present that information to you, and answer some of the questions about pitcher workloads and their impact.
Did starters throw more pitches back then?
That really depends on what you mean by "more pitches". For the 2,690 starts we have on file, Dodger starters threw a total of 253,513 pitches, an average of 94.2 pitches per start. As our modern standard, we took the 1998 NL data, to keep the comparison between non-DH leagues (save a smattering of interleague games in AL parks). In the 2,588 starts made by NL pitchers last year, they threw a total of 247,197 pitches, an average of 95.5 pitches per start. So on average, pitchers throw more pitches today than they did 40 years ago! It's barely a one-pitch-per-start difference, but we were expecting the old-timers to average five or more extra pitches per outing than they do today.
But--and it's a big but--the old-time hurlers amassed an average of 21.6 PAPs per start, while today's pitchers averaged just 8.0. That's an enormous difference; despite throwing fewer pitches overall, yesterday's pitchers averaged 169% more PAPs than today's pitchers do. How can that be?
It all comes down to distribution. We can't stress this point enough: it's why PAP is different than simply looking at average pitch count scores. PAP punishes inconsistency.
Take a look at the distribution of pitch counts per 1000 starts:
1-60 61-70 71-80 81-90 91-100 101-110 111-120 Old Dodgers 172 55 76 94 124 139 120 1998 NL 50 44 88 172 226 212 124 121-130 131-140 141-150 151-160 161+ Old Dodgers 100 70 31 11 7 1998 NL 65 15 2 0 0
Look at the far right side of the chart. In 0.7% of Dodger starts--19 total--the starter threw more than 160 pitches. Six times a starter threw more than 180 pitches, and three times he cracked 200, led by Joe Hatten's 211-pitch outing in 1948. (Sandy Koufax maxed out at 205 in a 1961 start; Don Drysdale hit 182 in a 1959 outing.) Eighteen starts out of a thousand--just under 2% of the time--an old-time starter would throw more than 150 pitches in a game. Last year, it happened once all season, when Livan Hernandez lived the crazy life to the tune of 153 pitches.
Those high-end pitch counts, 49 games in total, accounted for 11,747 PAPs, just over 20% of the entire total of PAPs in the old-time group. 320 out of 2,690 starts in the first group lasted more than 130 pitches; last year, just 45 out of 2,588 starts lasted that long. That is an enormous difference.
But if you look at the other side of the chart, you can see why, despite having so many high-pitch games, the Dodgers ended up with a lower overall average than the 1998 NL. More than 17% of the starts made by Dodger pitchers lasted fewer than 60 pitches. Last year, just 5% of starters were knocked out that early.
What you're seeing here is a manifestation of a radically different philosophy in bullpen usage. Today, the roles of starter, long man, middle reliever, set-up man and closer are well delineated, maybe overly so. Forty years ago, the boundaries between even such fundamentally different roles as starter and reliever were still indistinct - not as fuzzy as they were in the 20s and 30s, but indistinct nonetheless.
The 1955 Brooklyn Dodgers featured 11 different pitchers who started at least one game (including Tommy Lasorda!), and all 11 appeared in relief at least twice. The top four starters were Don Newcombe (31 starts, three relief appearances), Carl Erskine (29 and two), Johnny Podres (24 and three),and Billy Loes (19 and three). Ed Roebuck, who led the team with 12 (retroactively-figured) saves, appeared in 47 games, all in relief, but Clem Labine, who saved 11 games, started in eight of his 60 appearances, even throwing a complete game.
The result of this interchangeability among pitchers was that managers had no qualms whatsoever about pinch-hitting for a starter in the third inning or pulling him in a jam in the second and bringing in one of his other starters to go for four or five innings. But when the starting pitcher had good stuff, the mentality was that unless he got into a jam or showed obvious signs of fatigue, the game was his.
Keeping in mind that the average pitch count per start was about 95 for both groups, look at the percentage of starts made between 81 and 110 pitches: whereas last year, 61.1% of starters were pulled in that range, just 35.7% of all starts in the other group--barely one-third--fell into that range. To describe the differences between the two eras in a sentence: Pitchers in the 1950s came out of the game when their performance dictated it; pitchers today come out of the game when their workload dictates it.
Cynics might describe this as cookbook managing, making decisions based on conventional wisdom rather than what happens on the field. But on this issue, conventional wisdom has done its job, helping pitchers stay healthier and pitch longer. The idea of pitchers routinely staying at the top of their game into their mid-30s, a concept we take for granted today, was a rarity 40 years ago. The concept of pitchers able to contribute as they approach their 40s was even more remote. Some of that is advances in medicine, to be sure, but a lot of that comes from the diminished workloads at the far right end of the spectrum.
One of the key premises of PAP, in fact, is that conventional wisdom hasn't come far enough, and that not only should the far right of the spectrum--pitch counts of 130 and higher--be abolished for most pitchers, but pitch counts above 110 or so should be out of reach for young starters.
How did different managers vary in their approach to starters' workloads?
This was the O'Malleys' team, remember, so we only have three managers to look at over the 17 years of the study. Well, four, actually; Leo Durocher ran the team for half a season in 1948 before Burt Shotton took over. Shotton was replaced by Chuck Dressen in 1951, and after the 1953 season, the one in which the Dodgers went 105-49 and Roger Kahn would wax poetic about 20 years later, Dressen opened his mouth too much and got replaced by Walter Alston. Here's the data:
Year Manager GS PAP PAP/S 115+ 130+ 145+ Max 1947 Burt Shotton 105 1555 14.8 24 12 1 145 1948 Leo Durocher 72 2082 28.9 23 11 5 170 1948 Burt Shotton 81 2168 26.8 23 9 3 211 1949 Burt Shotton 156 3804 24.4 46 24 7 180 1950 Burt Shotton 141 3405 24.1 50 20 6 154 1951 Chuck Dressen 136 3129 23.0 48 13 8 157 1952 Chuck Dressen 155 3005 19.4 43 21 4 156 1953 Chuck Dressen 154 3400 22.1 47 23 6 150 1954 Walter Alston 154 2970 19.3 43 18 4 154 1955 Walter Alston 154 2733 17.7 43 17 4 159 1956 Walter Alston 154 1830 11.9 32 9 1 149 1957 Walter Alston 154 2551 16.6 35 14 4 169 1958 Walter Alston 153 2864 18.7 42 18 4 159 1959 Walter Alston 156 4310 27.6 49 22 9 182 1960 Walter Alston 154 4696 30.5 55 27 8 193 1961 Walter Alston 154 4405 28.6 47 24 6 207 1962 Walter Alston 165 3688 22.4 54 23 5 163 1963 Walter Alston 163 2935 18.0 40 16 3 165 1964 Walter Alston 127 2591 20.4 35 14 5 152
Durocher, in his half-season with the Dodgers, averaged almost 29 PAPs per start, the second highest of any season in the study. This is in agreement with his managerial reputation as someone to whom winning now was the highest priority. As Bill James said of him, "Durocher had no long-term plan. Next year was next year; two years down the road was something he'd worry about two years down the road."
Dressen took relatively good care of his starters; in three years under him, no Dodger threw more than 157 pitches in a start. In 1950, before Dressen took over, the Bums ranked sixth of eight teams in runs allowed (remember, this is 12 years and 3,000 miles from Dodger Stadium). In the three years he managed, they ranked fifth, second and third in the NL in runs allowed, and would stay in the top half of the league after Walt Alston took over.
Alston may be the most interesting study here. In his first five seasons at the helm, Alston protected his starters as much as any manager before him, including Dressen. Only twice in those five seasons did a pitcher throw 160 pitches, and in 1956 his performance would not have looked out of place in today's game. By way of comparison, Jim Leyland's Marlins led the NL last year with 2,433 PAPs, an average of 15.0 per start. But in 1959, the Dodger workloads suddenly jumped nearly 50% to their highest total since 1948, and the following year they averaged over 30 PAPs per start.
Why? Well, in 1959 Don Drysdale suddenly came of age, making the All-Star team for the first time and leading the league in strikeouts with 242. And Alston put him to work: after picking up just 365 PAPs in 28 starts the year before, he was worked to the tune of 1,855 PAPs in 36 starts in 1959. Beginning in 1947, the highest previous workload was just 1194 PAPs by Carl Erskine in 1953. Drysdale passed that by nearly 900.
The increase in Drysdale's workload alone--1,490 more PAPs in eight more starts--by itself explains the jump of 1,446 PAPs between 1958 and 1959. Take Drysdale's starts out and the average PAP/S wmoved from 20.0 to 20.5, a negligible difference. Consider that in 1959, a Dodger starter threw more than 145 pitches on seven occasions: four by Drysdale and three by Sandy Koufax. Drysdale would never again record that many PAPs in a season, but in 1960 Koufax, still inching his way towards superstardom, went from 947 to 1,647 PAPs, and his workload--not surprisingly--would come down only when he retired.
And then, after three years of abuse, Alston suddenly relaxed the reins a little in 1962, bringing his starters' workloads down. Not to the same level as when he took over the club, but in line with the manner of the managers before him. This, despite the fact that Koufax was just entering his prime and Drysdale was still going strong.
It's really not all that difficult to explain. Here, we'll do it in two words: Dodger Stadium.
The House That Sandy Christened
Sandy Koufax, in 1961, led the NL with 269 strikeouts, but his season wasn't all that extraordinary: 18-13 with a 3.52 ERA. In 1962 he made only 26 starts, but went 14-7 with a league-leading 2.54 ERA; he would lead the league in ERA in each of the next four years, never once getting above 2.04. In 1961, Don Drysdale was 13-10 with a 3.69 ERA. In 1962 he went 25-9, had a 2.83 ERA and won the Cy Young award. His ERA would get above 2.83 only once in the next six seasons.
But if you look at their ERAs on the road, there is no sharp improvement in 1962:
Year Koufax's ERA Drysdale's ERA 1958 3.75 5.05 1959 5.05 4.03 1960 3.00 2.51 1961 2.78 4.54 1962 3.53 3.64 1963 2.31 2.65 1964 2.93 2.39 1965 2.72 3.09 1966 1.96 4.65
There is a fairly sizeable improvement starting in 1963, the year that Major League Baseball re-defined the strike zone to include, roughly, the area between the hitter's shoelaces and his chin. The National League ERA, 3.94 in 1962, dropped to 3.29 in 1963, and would stay in the low 3.00s (dropping to 2.99 in 1968) until the owners looked around, noticed that their stadiums were half-empty, and changed the rules back in 1969.
In this pitchers' era, nowhere was it more friendly to pitch than Dodger Stadium. The foul area was generous, the dimensions ample, the elevation negligible, the hitting background difficult and the pitcher's mound, which had been legislated up to 15 inches in 1963, was tweaked so that it was only slightly lower than the upper deck. From 1958 until 1961, the Dodgers played their home games in the Los Angeles Coliseum, which was a good hitter's park, especially for right-handed hitters. When they moved to Chavez Ravine in 1962...here are the home ERAs for the Dodger Duo:
Year Koufax's ERA Drysdale's ERA 1958 5.60 3.40 1959 3.14 2.97 1960 5.27 3.14 1961 4.29 2.83 1962 1.75 2.11 1963 1.38 2.60 1964 0.85 1.97 1965 1.38 2.45 1966 1.52 2.34
If you can't draw a dark black line between "1961" and "1962", you must be out of ink.
The result of all this is that while the Dodger starters may have been throwing the same number of innings, they were not facing the same number of batters. From 1960 to 1963, the number of complete games thrown by the Dodgers were 46, 40, 44, and 51. Accounting for the slightly-longer schedule beginning in 1962, there was no significant difference overall. But in 1961, Drysdale averaged 4.29 batters per inning; in 1962 that dropped to 4.10. For Koufax, it went from 4.18 to 4.04 to 3.89 (!) in 1963.
Drysdale won the Cy Young in 1962, and pitching as well as he was, averaged 7.5 innings a start, compared to just 6.4 the year before. His average pitch count jumped from 98.1 to 111.2, and his PAP score jumped from 753 to 1373. But in 1963, despite throwing 315 innings over 42 starts--a 7.5 average--he picked up the amazing total of just 500 PAPs, or less than Orlando Hernandez earned in half as many starts last year.
Koufax was still making over a half-dozen relief appearances a year in the early 60s, so it's difficult to figure out how many innings he threw per start. But after averaging the astonishing total of 57.2 PAPs per start in 1961, that figure dropped to 47.7, 45.7 and 38.4 in the following three years, even as he averaged 7.8 and 7.9 innings per start in 1963 and 1964.
The message is clear: a good pitcher's park, and a good pitcher's environment in general, helps to keep the workload of starters down. The message is not as true today as it was in the 60s, when great starters were completing half their starts, because today a good starter in a good pitchers' park might throw seven or eight innings a start, while his brethren in less hospitable venues might be taken out after six. The "theoretical maximum" number of pitches that would be thrown in a complete game is rarely a consideration anymore. Rather than extending pitch counts to meet an innings quota, the inning totals are shrinking to accomodate a pitch limit.
But this is also why, in an era as offense-driven as today's is, pitchers are frequently injured despite the fact that the average starter goes far fewer innings than he did even 10 years ago. The innings may have come down, but as anyone who has sat and watched Russ Ortiz or Freddy Garcia or some other young hurler labor through a five-inning, 125-pitch outing, the pitch counts haven't.
Eventually, we hope to present a season-by-season breakdown of all the significant hurlers from these Dodgers, to find whatever nuggets of wisdom are buried in the data.
Rany Jazayerli is an author of Baseball Prospectus.