September 28, 2006
Schrodinger's Bat: Baseball's Trifecta
by Dan Fox
"I think if you come to the ballpark and you see Carl hit a triple, you've had a pretty good day. It's pretty much a signature play for him, because when he hits the ball down the line, or in the gap, he's thinking three. He never thinks two. He breaks triple. He wants triple, he takes triple."
--Devil Rays manager Joe Maddon after Carl Crawford's triple on September 24.
"Hey, big mouth, how do you spell triple?"
--Shoeless Joe Jackson, to a heckling Cleveland fan who taunted him by asking if he could spell "illiterate." This was his response after hitting a triple.
In the bottom of the sixth inning of last Sunday's Yankees/Devil Rays game, Carl Crawford pulled Mike Myers' 1-0 slider into the gap in right-center. The ball skidded past Bobby Abreu, and by the time he retrieved it and hit the cutoff man, both runners had scored and Crawford had coasted into third. It was his 15th triple of the season.
As much as I disdain more or less arbitrary statistical milestones, the hit did draw some attention, since it made Crawford the first player in 76 years to hit at least 15 triples in three straight seasons. In that year, 1930, no fewer than three players were finishing a run of three or more years with 15 triples or more:
1930 1929 1928 1927 1926
Earle Combs 22 15 21 23
Paul Waner 18 15 19 18 22
Charlie Gehringer 15 19 16
By comparison, Crawford hit 19 triples in 2004, 15 last year, and now 15 this season. When asked after the game why he thought it had been so long since a player accomplished the feat, Crawford replied, "There are fast guys in the game who can hit, so I have no clue why guys haven't done it. That's not a stat that you go out and try to do every year. That's a stat that just happens."
Crawford's achievement and his comment provide a springboard for this week's column, where we'll discuss triples and their accompanying historical trends.
The simple and somewhat tautological answer to Crawford's consternation regarding the lack of triples is that the triple has become increasingly rare over time. And just as a rising tide lifts all boats, a low tide grounds them. The following graph shows the number of triples per 500 at-bats plus walks for each year from 1901 through 2005:
Notice that, as it did for offense in general, the robust environment of 1930 marked the high point for triples, with 6.8 triples hit per 500 AB+BB. The rate dropped immiediatly thereafter, to 5.7 in 1931 and 1932, and it never again reached as high as 5.3. It now seems to have stabilized at around 2.5.
If there are fewer triples being hit, then it becomes less likely that an individual player will be able to hit 15 in three consecutive seasons. For example, a player who hits 15 triples would have a rate of 12.5 per 500 AB+BB. In 1930, a player who hit triples at 1.8 times the rate of the average player would end up with 15 triples, and eleven of the 73 players with 500 or more AB+BB hit 15 or more triples in 1930. In 2005, however, a 15-triple player would have to hit triples at a rate more than five times that of the average player, and just two players (Crawford and Jose Reyes, who hit 17) out of 140 with 500 or more AB+BB could do that. It should be noted that 2005 was a comparatively good year for three-baggers: since 1992, there have been ten seasons in which no player has hit 15 triples. In contrast, when 10-15% of the players hit 15 or more triples every year, there is a very good chance that one or more of those players will repeat for three consecutive years.
As an aside, the general reduction in triples makes the performance of Cory Sullivan on April 9 even more of a fluke. In the top of the fifth inning, Sullivan hit two triples in a seven-run outburst that helped the Rockies beat Jake Peavy and the Padres 10-4. Those two triples tied a record held by ten others, although most recently accomplished by the Senators' Gil Coan in 1951.
The graph also shows the spike in triples that occurred between the years 1974 and 1980. As you can see, triples had been declining steadily since 1930, reaching a low of 2.72 in 1973. From there they began to climb again, reaching a high point of 3.71 in 1977 before gradually declining to settle back down at the 1973 level by 1986. Note that 1977, like 1930, was a relatively big offensive year, with teams scoring 4.47 runs per game. Offensive levels continued to rise throughout the period, and so it can't simply be chalked up to more hits resulting in more triples. But since there weren't new parks being introduced, and expansion occurred in the middle of the spike and not at its start, it's not obvious what might have caused it the outburst.
At first blush, one might posit that there was a general trend towards valuing speed that began in the early to mid-1970s, as young players like Ron LeFlore, Tim Raines, Willie Wilson, and Omar Moreno began to establish themselves. The increase in stolen bases, however, is more gradual than that for triples, and actually began around 1959 (with the "Go-Go Sox" and Maury Wills playing a large part) with a steeper increase in the mid-1970s that peaked in another year that was good for offense, 1987, as shown in the following graph:
While all of this is interesting, it kind of tiptoes around the answer that Crawford is looking for (he probably doesn't really care, but play along). Although there doesn't appear to be consensus among the analytical community, the following are the theories most often discussed as to why the triple has become relatively rare:
- Better Fielders. One of the more interesting questions is to consider how the game has changed as the players have become more athletic. Clearly the speed, strength, size and athletic ability of the average professional baseball player in 2005 exceed that of one in 1920. The question is, how does this affect the game and the evaluation of performance? This was recently touched on by Phil Birnbaum on his Sabermetric Research blog, and was the subject of a thought-provoking chapter by Nate Silver in Baseball Between the Numbers.
As an example of such an effect, the late Stephen Jay Gould argued that a rising level of play inching closer to the "right-wall" of human ability coupled with stabilization of the game itself have conspired to decrease the variability in seasonal batting averages, making it far more difficult to hit .400 now than in years past. The epitome of Gould's argument is that Tony Gwynn had less opportunity than Ty Cobb to exploit the inferiority of others.
Something like this may be happening with triples as well. The theory is that as fielders have become bigger, faster, and boast better throwing arms, would-be triple hitters have had a more difficult time exploiting their opponents, and thus rack up fewer three-baggers. In addition, the standardization of positioning (including the idea that outfielders played shallower in the past) and cutoffs have added to the difficulty. Although baserunners have also become faster, this theory would argue that the improvement in fielding ability and techniques has outstripped the increase in baserunner speed.
- Park Configuration. This is a corollary to the first theory. Early in the century, ballpark dimensions were far less standardized than today. For example, the Huntington Avenue Grounds where the Red Sox played from 1901-1911 featured a left-center field fence 440 feet away, and a centerfield wall 530 feet from home plate from 1901-1907, and then at 635 feet starting in 1908. Similarly, the center field fence at Forbes Field was 462 feet away in 1909, and at the Polo Grounds, center field ranged from 430 feet in 1931 to 505 feet in 1949. Don't forget that while these and other ballparks in the two eight-team leagues had one or more long distances, they also had lots of corners and edges that made for unpredictable caroms. All of this adds up to situations which surely allowed hitters more opportunity to leg out triples.
Over time, standard dimensions (335/375/400/375/335) made their way into the game, diminishing the opportunity for strange bounces and balls rolling towards distant fences with outfielders in hot pursuit. For my money, the combination of this and the first cause probably explain the lion's share of the overall historical trend.
- Risk Aversion. As mentioned previously, as offensive levels rise, the relative importance of stolen bases decrease. The same reason causes triples to decline in value; the marginal benefit of stretching a double into a triple is lessened as the probability of scoring from second base increases. A quick look at Run Expectancy Matrices from various years would bear this out as well as the graph presented in my column on Win Expectancy. The argument, then, is that as offensive levels have risen over time, triples have decreased as a result of their lessening strategic importance.
While the premise of this theory is certainly true, one doubts whether calculations like this are taken into account either consciously or subconsciously. More problematic, however, is the fact that runs per game have not increased over time, therefore cutting the legs out from under this theory. Contrary to the steadily downward-sloping line in the first graph, run scoring was actually higher throughout the 1920s and into the early 1930s than at any other time, and after diminishing to reach its low point in 1968 (3.42 runs per game per team), it has steadily increased since then:
- Player Aging. No discussion of triples would be complete without at least a brief look at the effect of age. As Clay Davenport noted in his essay "Graying the Game" in Baseball Prospectus 2002, and subsequently reinforced by Nate Silver last season, the player population is aging, and has been for quite some time. This has an impact on triples, since older players lose foot speed and don't hit as many as younger ones do. The graph below shows triples per 500 AB+BB for all players since 1901, player-seasons from 1901-1935 and seasons from 1936-2005:
Even at a time when triples were much more common (the orange line), triples peaked at age 22 and steadily declined through age 40. You'll also notice that the slope of the line for players in the first part of the 20th century is not quite so steep as it is for those since. I also find it very interesting that the slope of the line from ages 22 through 35 for all players is very nearly straight, indicating an extremely uniform decrease with age.
Clearly, players don't hit as many triples as they get older, but the general aging of the player population cannot account for the overall decrease in triples. Just considering players 25 years old or younger, those who played since 1936 hit triples at a rate of 3.72 per 500 AB+BB, while those who played before 1936 hit them at a rate of 6.73. Keep in mind that it's also very likely that the average speed of players 25 years old and younger in the major leagues today is greater than that in the Deadball Era, meaning that other factors such as fielding prowess and changing park configurations are much more important to the overall trend.
"How do you spell triple?"
The triple is often called the most exciting play in baseball, and for good reason. There is no play that involves as many players, lasts as long, and concludes so often with a bang-bang crescendo. As we've seen, there are a variety of reasons that have conspired to make it a much rarer event today than it was in days past. These include increased standardization and ability on defense, less variability in park dimensions, risk aversion, and perhaps an aging player population. Whatever the combination and relative importance of these different causes, rather than wring our hands at its disappearance, let's instead appreciate the feat for its increased difficulty and marvel at those, like Carl Crawford, who can do it with regularity.