Do you know why strikeouts, over time, only go up?
Strikeouts, over time, always increase, for this reason. Strikeout pitchers are more effective than pitchers who don’t get strikeouts, therefore teams are always looking for pitchers who can get more strikeouts, and also looking to deploy those pitchers they have in such a way that they will get the most strikeouts. This effect would be offset by the tendency of teams to look for hitters who don’t strike out, if hitters who did not strike out were also better hitters. However, hitters who strike out are generally not less effective than hitters who do not strike out; hitters who strike out are generally just as effective as or more effective than hitters who don’t strike out. Thus, there is no pressure to find hitters who don’t strike out. This asymmetry pushes strikeout totals higher over time.
It occurred to me in January that I should be able to measure this asymmetrical effect, and thus measure the upward pressure on strikeouts at any point in baseball history. This can be measured in the following way:
1) Measure the extent to which the high-strikeout pitchers are more effective than the low-strikeout pitchers.
2) Measure the extent to which the high-strikeout hitters are less effective than the low-strikeout hitters (if they are), or measure whatever effect there is there.
3) Combine these two measurements into one.
I’m going to jump right to the conclusion here. I started this study in an optimistic frame of mind, thinking that the study might show us that we have finally reached an equilibrium, where there was no more upward pressure on strikeouts, or a near-equilibrium, where the upward pressure on strikeouts was minimal.
Wrong.
I don’t know if you know. …strikeouts, which generally only increase over time, have increased enormously in the last eight years. In 2003 the major league average was 6.4 strikeouts per nine innings pitched. In 2011 it was 7.3 strikeouts per nine innings. Rarely or never before have strikeouts increased at such a rapid pace.
I thought I had seen signs that this era of rapidly increasing strikeouts might have reached an end. Unfortunately, if my study is correct, then. . .no such luck.
Let’s go back to the beginning. My study covered the seasons 1916 to 2011, a 96-year period. Let’s take the pitchers in the first three years of that period, the years 1916 to 1918. In those years there were 373 batters who had 300 or more plate appearances (373 batter/seasons.) Suppose we divide those into four groups: the top 25% in strikeouts per at bat, the middle 50%, and the bottom 25%.
In that era (1916-1918) the hitters who struck out the most were the least effective hitters. There was little power in the game; the hitters who struck out the most could not compensate for that by hitting home runs, because there just weren’t very many home runs. In that era the high-strikeout hitters created 4.20 runs per 27 outs, on average. The mid-range strikeout hitters created 4.23 runs per 27 outs, whereas the low strikeout hitters—who included Ty Cobb, Tris Speaker, Joe Jackson, Eddie Collins and George Sisler—created 4.72 runs per 27 outs. The "strikeout push" effect of this can be measured at negative .52 (4.20 minus 4.72).
On the other hand, the high-strikeout pitchers of that era—also studying only those who faced 300 or more batters in a season—the high-strikeout pitchers of the 1916 to 1918 era allowed an average of 3.09 runs per nine innings. That, again, is the top 25% in strikeouts per inning. The mid-range 50% allowed 3.57 runs per nine innings, and the bottom 25% in strikeout rates allowed 3.73 runs per nine innings.
The runs allowed rates for pitchers are lower than those for hitters because we are only studying the regulars and near-regulars. If we included the batters with 100 plate appearances and the pitchers who pitched 25 innings, the runs allowed rate for the pitchers should be the same as the runs created rate for the batters, but that’s not important right now.
Anyway, the high-strikeout pitchers had a runs-allowed rate of 3.09; the low-strikeout pitchers had a runs-allowed rate of 3.73. That’s a "strikeout push" effect of 0.64 (3.73 minus 3.09). If we add together the strikeout push effect of the batters (negative 0.52) and the strikeout push effect of the pitchers (0.64), the total is 0.12. The "Strikeout Push Effect" for the years 1916 to 1918 is 0.12.
That’s a very low figure; I’ll get ahead of myself and tell you that that’s a very, very low figure; it creates only minimal upward pressure on strikeouts. If we included more factors and if we had included the part-time players and the failed pitchers, a Strikeout Push Effect of only 0.12 might actually equate to a DOWNWARD pressure on strikeouts. I believe that it would.
At any rate, strikeouts did go down. In 1916 there were 9,525 strikeouts in the major leagues; in 1922, in basically the same number of games, there were 6,915 strikeouts in the majors (as best we know. The standard of record-keeping in that era is so poor that the league strikeout totals for hitters don’t jibe with the strikeout records for pitchers, often by several hundred strikeouts.) Our method shows that in that era there was little upward push on strikeouts, and strikeouts were going down.
OK, we’re going to say that the strikeout push effect for 1917 was 0.12, and I’m going to run through this again to make sure you understand what that means:
|
1916
|
to
|
1918
|
High Strikeout Batters
|
4.20
|
|
|
|
1916
|
to
|
1918
|
Low Strikeout Batters
|
4.72
|
|
|
|
1916
|
to
|
1918
|
Batter Strikeout Push
|
|
-0.52
|
|
|
|
|
|
|
|
|
|
|
1916
|
to
|
1918
|
High Strikeout Pitchers
|
3.09
|
|
|
|
1916
|
to
|
1918
|
Low Strikeout Pitchers
|
3.73
|
|
|
|
1916
|
to
|
1918
|
Pitcher Strikeout Push
|
|
0.64
|
|
|
|
|
|
|
|
|
|
|
1916
|
to
|
1918
|
Combined Strikeout Push Effect
|
|
|
+0.12
|
This number began to creep upward. In the years 1922 to 1924 the Strikeout Push Effect was +0.34—still very low. In the late 1920s, however, this number exploded. Following the example of Babe Ruth, more and more hitters began to swing hard, risking strikeouts to get some home runs. In 1926 (1925-1927) the Combined Strikeout Push Effect was 0.52. In 1929 (1928 to 1930) it was 1.57.
We could have predicted, then, that strikeouts would increase, and they did. In 1930 there were 7,931 strikeouts in the major leagues. In 1940 there were 9,056—still less than there had been in the Walter Johnson era.
In 1918 Ty Cobb, Joe Jackson and Tris Speaker were in the low-strikeout group, tending to push strikeouts down. In 1930 Babe Ruth, Hack Wilson and Jimmie Foxx were in the high-strikeout group, tending to push strikeouts up. If the best hitters are striking out a lot, you don’t worry about strikeouts, and when you don’t worry about strikeouts, strikeouts go up.
After the 1930 era, the upward pressure on strikeouts began to abate gradually. By 1935 the Push Effect was down to 1.13 (from its high of 1.57); in 1941 it was down to .84, and in 1947 it was back down to .44. The strikeout curve flattened out. In 1940 there had been 9,056 strikeouts in the major leagues; in 1949 there were 8,956. Still less than in the Walter Johnson era.
The three great hitters of the 1940s—Musial, DiMaggio and Ted Williams—were guys who DIDN’T strike out much. If you look back to the 1920s and early 1930s, the league leaders in batters strikeouts were Babe Ruth, Hack Wilson and Jimmie Foxx. In the mid-1940s they were Vince DiMaggio, Pat Seerey and Chet Laabs.
Soon, however, another generation of muscle men/strikeout guys emerged. Ralph Kiner led the National League in strikeouts in 1946, Hank Sauer in 1948, Duke Snider in 1949, Gil Hodges in 1951, Eddie Mathews in 1952. Among the American League leaders were Larry Doby and Mickey Mantle. The good hitters were striking out again.
Breaking it down further, in 1935 the Strikeout Push Effect was +1.22 for pitchers, but negative .09 for hitters, thus +1.13 total. In 1947 Push Effect was +.89 for pitchers, but negative .45 for hitters, thus +0.44 combined. By 1956 it was +.70 for pitchers, negative .09 for hitters, the net effect edging back up to +0.61. In 1959 it was +0.71; in 1962 it was +1.04. There was strong upward pressure on strikeouts once again.
The redefinition of the strike zone in 1963 came in the context of a game in which strikeouts were already increasing, and there was already strong upward pressure on strikeouts.
To break into our story with a note of caution, I have been assuming here that the Strikeout Push Effect is predictive. I haven’t actually proven that it is predictive. It is possible that the effects I am measuring for 1962 are not predicting what will happen in the game in 1963-1972, but reflecting what has happened in 1953-1962. It’s possible. I am assuming that it is predictive because
a) I designed it to be predictive in theory, and
b) It is obvious that the increases in the Strikeout Push Effect do in fact track with the changes in the strikeout rate, that when the Strikeout Push Effect goes up, strikeouts go up.
But the chicken-and-egg question is mathematically a much harder question, and I haven’t actually gotten into that. Just don’t want to mislead you on that issue.
OK, in 1962 the Strikeout Push Effect was at 1.04, the highest it had been since 1938. In 1965 it was 1.23. Despite the lowering of the mound in 1968—or perhaps because of it—the Strikeout Push Effect continued to ascend, up to 1.30 in 1971 and 1.41 in 1977, the highest it had been since 1930.
From the mid-1970s until the early 1990s, the Strikeout Push Effect declined, but remained high by historical standards. By 1991 the Strikeout Push Effect was down to +0.88—but +0.88 is not a low figure. +0.88 indicates that there is still significant upward pressure on strikeouts.
Strikeouts per game increased only from 5.2 per game in 1977 to 5.6 in 1992, relatively modest increases. As baseball entered the steroid era, however, strikeouts began to increase more rapidly. By 2001 we were up to 6.7 strikeouts per game.
In 2004 major league baseball had 6.6 strikeouts per nine innings—but a Strikeout Push Effect of +1.64. The Strikeout Push Effect of 2004 was at an all-time high.
And, as would be predicted by this theory, strikeouts did in fact explode after 2004. The last eight years have had not only historic numbers of strikeouts, but historic increases in the rate of strikeouts.
Where are we now?
Well. . .it ain’t pretty. Since 2004 the Strikeout Push Effect has dropped slightly but steadily, down to a present figure of +1.47. In the years 2009 to 2011, high-strikeout hitters created 5.04 runs per 27 outs, while low-strikeout hitters created only 4.93 runs per 27 outs. On the other hand, high-strikeout pitchers allowed only 3.65 runs per 9 innings, while low-strikeout pitchers allowed 5.00.
I would have much preferred to find that the upward trend in strikeouts was leveling off. Unfortunately, that does not appear to be the case. High-strikeout pitchers in today’s game are dramatically more effective than low-strikeout pitchers, while high-strikeout batters are also somewhat more effective than low-strikeout batters. We are where we have always been, only worse. Strikeouts, in my opinion, will continue to go up.
Trayvon III
I was thinking about something else related to the Trayvon Martin case. Remember that guy who showed up in the middle of the circus, placing a "bounty" on George Zimmerman? Why wasn’t that guy arrested the next day?
It’s obviously a crime, right. . ..making a terroristic threat. If I offered a bounty for your murder, I would assume that I would be immediately arrested, and if you offered a bounty for my murder, I would hope that you would be immediately arrested. Why wasn’t this guy arrested?
Two reasons, I think. First, the authorities didn’t want to expand the scope of their problems. They’ve already got a shitstorm on their hands; they don’t need a second one.
But that’s not a good reason, when you think about it, because of the example it sets to ignore the threat. We’re going to have other controversies in the future, in which people should have been arrested but aren’t. It’s pretty obvious that the precedent set by ignoring this kind of thing is more dangerous than the problem created by expanding the parameters of the immediate shitstorm. I don’t think that’s the main thing.
The main thing, I think, is they didn’t want to make that guy any bigger than he was. The guy was obviously trying to use the media and the publicity associated with the Martin/Zimmerman case to make a name for himself. If you arrest him, then he becomes the issue.
If you can arrest him and put him away for ten years, OK, you’d probably go ahead and do that. But. . .again, I’m guessing here. . .but I’m guessing that if he is arrested for making a terroristic threat, it’s probably theoretically possible to put him away for ten years but it’s probably not going to happen in the real world. In the real world if you arrest him, he probably gets $25 million worth of publicity in exchange for three months in jail, negotiates some sort of a plea, and he’s back on the streets and a bigger problem now than he was before. I’m guessing that’s why it wasn’t done.
History may prove that that was a bad decision. This guy isn’t going to go away; the next time there’s an opportunity, he’s going to be back in the middle of it, making himself an issue, making the problem a little worse. It may prove, in the long run, that it would have been better to arrest him and start the clock rolling on the long-term prison sentence that lurks at the end of the block. We’ll see.
The Perfect Age Study
Back to baseball. I do studies all the time that I don’t get time to write up. I did these two studies in January, never got time to write about either of them until now. I have 200 unpublished studies, will never get time to catch up.
Let us say that the perfect age for a baseball player is 27 or 28. (By the way, in February I had a meeting with a guy from England who I gather is the world’s foremost soccer researcher. One of the questions I asked him was what was the peak age for a soccer player. "27", he said immediately, and then explained that it varies with the position; goal tenders peak later because they don’t have to run as much, and certain types of players peak earlier because all they do is run around frantically, but basically. ….27.)
Anyway, let’s say that the peak age for a baseball player is 27 or 28, and let us say that a player has an "age score" which is:
100 if he is 27 or 28,
4 points less than 100 for each year that he is younger than 27, and
2 points less than 100 for each year that he is older than 28.
A player at 25 or 32, then, has an "age score" of 92; a player at age 20 or 42 has an "age score" of 72, while a player who is 17 or 48 has an age score of 60. Jamie Moyer has an age score of 56, meaning that he is not really at the perfect age for a baseball player.
In 1879 Monte Ward, who was 19 years old, won 47 games in the National League. We can take this as an indicator of the quality of the league. When you have 19-year-olds dominating the league, that indicates that the quality of the league could be a little weak. It is among the common indictments of baseball in World War II that the game was played by teenagers and old men, and it is among the complaints about expansion that the 1963 Houston Colts had ten teenagers who played for them (two of whom were Rusty Staub and Joe Morgan, and the other eight of whom were not.) When the quality of baseball goes down, the average age score goes down.
This applies also to levels of baseball; if you strung out the minor leagues, the perfect age score would be higher in AAA than in AA, higher in AA than in A ball, higher in High A than in Low A, higher in Low A than in rookie ball, etc. The Perfect Age score would be higher in college baseball than in high school.
We can figure the perfect age score for each season in baseball history by simply multiplying each player’s plate appearances by his age score, and each pitcher’s batters faced by his age score, and finding the weighted average. We not only could; I actually have. Maybe I’ll start by presenting the data:

All of those numbers look a lot alike, don’t they? I set up the system that way for a reason. It is part of a system to measure the quality of a league at every level, not simply the relative quality of the majors in 2011 as opposed to 1950. We could make a perfect age score for a league that would place major league baseball on the same scale with T-Ball. Assuming that T-Ball kids are five, their "age score" would be 12. A downside of doing it that way, however, is that the numbers all look alike. We can "correct" for this by setting the third-lowest score on this chart equal to zero, and the third-highest score equal to a hundred. That makes the patterns easier to see, and I’ll try to help that by marking "up" numbers in blue and "down" numbers in green:
1) The "Perfect Age Score" started out very low, and remained relatively low throughout the 19th century, although increasing steadily.
2) The number surged forward in 1900, when the National League disbanded four teams, cutting from a 12-team league to an 8-team league, and then dropped back down in 1901, when the American League opened for business.
3) After a high in 1904 the numbers declined substantially, for reasons that I don’t understand.
4) The perfect age score reached its all-time peak in 1919-1921. The players in that era were clustered more closely around ages 27-28 than at any other time in baseball history—still. I don’t know why that happened, but it was a very dynamic situation, with the coming of Babe Ruth, the banning of the spitball, the interruption of the game due to World War I and the expulsion of the Black Sox and the other corrupt players.
5) After 1920 the numbers declined steadily due, I believe, to economic forces. Booming attendance drove salaries up, so that aging players could make more money as 35-year-olds than they had made in their prime seasons. This kept a lot of older players in the game.
6) This is perhaps the key point: the anomalies associated with World War II were tremendously overstated by the media in that era, and have been exaggerated by historians ever since. The perfect age score did go down in World War II—barely. The figure for 1945 was the lowest since 1913, but the general impression that baseball in the war years was played by teenagers and old men is just wrong. The perfect age score was higher in 1943 than it was in 1928, 1929, 1930, 1931, 1932, 1933, 1940 or 1941. It only really declined in 1945, the final year of the war, and even then the change is not all that notable.
7) After shooting upward after World War II, the number declined again due largely to the Whiz Kids, the Philadelphia National League champions of 1950, and the Bonus Baby Wars of the post-war era, which brought into the majors a substantial number of very young players.
8) The Perfect Age Score dropped sharply in 1960-1961 due to the influence of the "Kiddie Corps" in Baltimore. In 1960 the Baltimore Orioles, managed by Paul Richard, used a starting rotation of Milt Pappas (aged 21), Steve Barber (21), Chuck Estrada (22), Jack Fisher (21), Jerry Walker (21) and one other, older pitcher. They led the American League in ERA. This made other managers, for some years, dramatically more willing to put 21-year-old pitchers on the mound.
9) The Perfect Age score dropped further after expansion, reaching its lowest point since World War II, and then recovered in the late 1960s.
10) The numbers went steadily but slowly upward from the mid-1960s until the mid-1990s, reaching the highest peak in the mid-1990s since the early Babe Ruth era.
11) The steroid era changed aging patterns, kept older players in the game, and thus drove the numbers sharply lower, down to their lowest point since the 1960s.
12) Since the banning of steroids the Perfect Age Score has moved up. The 2011 figure was the highest since the mid-1990s.