January 25, 2007
A Triple Redux
"I don't know why people like the home run so much. A home run is over as soon as it starts....The triple is the most exciting play of the game. A triple is like meeting a woman who excites you, spending the evening talking and getting more excited, then taking her home. It drags on and on. You're never sure how it's going to turn out."
Last week in this space we laid out the case for just why it is (and how we can tell) that the quality of play in the major leagues has steadily improved since the game's early days right up through the dawn of the 21st century.
The arguments and evidence examined included the following:
In discussing the increased athletic prowess of today's athlete, I mentioned that the average player of today is over two inches taller and 20 pounds heavier than the average player in the dead-ball era. To be more precise, the following table lists the average height and weight by decade for players that debuted in the major leagues in that decade (debut information is available for 99% of the over 16,800 players that have played the game).
Average Height and Weight By Decade of Debut Decade Height (in) Weight (lb) 1870s 68.9 162.9 1880s 69.8 170.0 1890s 70.2 171.7 1900s 70.7 174.4 1910s 71.0 172.5 1920s 71.1 173.6 1930s 71.8 179.6 1940s 72.2 183.0 1950s 72.6 186.1 1960s 72.9 188.3 1970s 73.1 188.5 1980s 73.3 191.2 1990s 73.5 193.7 2000s 73.7 196.8
Comparing the 1910s with the 2000s, we see a difference of 2.7 inches and 24.3 pounds, thereby confirming the generalization. While that difference can be primarily explained by better nutrition and medical care in the general population, the increase in height may also be attributed to more stringent selection within a larger population pool as well as the attendant advantages of economic success that professional athletics brings. This is suggested by data that shows that height among major leaguers has risen faster than that of the general population (of Americans anyway) over the past 30 years.
When you look more closely at the trend--as illustrated in the following graph--it is clear that while height increase has been almost perfectly linear since the 1870s, weight increase has outstripped height, as the slope of the pink line begins to turn upward in the 1980s.
This divergence can likely be attributed primarily to the widespread adoption of weight training (with the shadow of steroids always looming) in the late 1980s, a practice long eschewed by the industry.
Peter Morris, in his excellent book A Game of Inches: The Stories Behind the Innovations That Shaped Baseball (Volume I), notes that the Big Red Machine of the mid 1970s is usually credited with being the first club to make extensive use of weight-training machines, although it should be added that a smattering of players including
One would also expect that weight gain might outstrip gains in height for the simple reason that humans as a species possess a genetic height potential that is not unlimited, while changes in weight can be more variable.
As an aside, I'd be interested to hear reader theories as to why average height and especially weight dipped in the period from 1910 through the 1920s and then picked up where it left off in the 1930s. No systemic cause or data problem jumps out at me and it doesn't seem likely that the slashing style of play popularized by the Baltimore Orioles of the 1890s would have such a latent effect.
But be that as it may, I'm guessing you're wondering why I'm reiterating some fairly well-established trends. Read on.
As you might have guessed from the title of this column, the reason for these musings has to do with triples. Put more directly, back in September I wrote a column chronicling the sad decline of the triple, and although the changing demographics in baseball were brought up at the time by several discerning readers, I hadn't gotten around to considering how they might have had an impact on that decline. The remainder of this column is an examination of that interaction.
As you'll perhaps recall, the previous column documented the decline in triples using the following graph.
To recap, a player who accumulated 500 at bats-plus-walks during the dead-ball era through the mid-1930s would have averaged from 6 to 7 triples. Since then the average has fallen steadily, with a slight upturn in the late 1970s, and has now settled in around a paltry 2.5.
While we explored several of the reasons for this decline, including better fielders, differing park configurations, risk aversion, and the general aging of the player population, none seemed to offer a complete and satisfying explanation.
What if, as several readers questioned, the decline of triples can be in large part explained by the increasing bulk of the players themselves? As players get larger and larger do they also therefore get slower or lack the quickness to stretch doubles into triples?
To investigate this theory I calculated the Body Mass Index (BMI) for all major leaguers who debuted from the 1880s onward. As a measure of obesity BMI doesn't really apply to athletes: because muscle density is greater than fat density, it overestimates their body fat percentage. Since here we're comparing only athletes, it should be a good indicator of the coupling of increased height and weight due to better nutrition and increased bulk due to weight training. The graph below shows the trend of BMI for players that debuted per decade.
As we saw in the graph of height and weight, there is a clear dip that occurs in the 1910s and 1920s, and after seeming to reach a peak in the 1970s, BMI takes off again in the 1980s through the 2000s. Once again this documents the increasing bulk of players through time.
To get a feel for how the distribution of BMI has changed over time, I then divided the players by a measure called simply BMI Index. That index divides BMI into the following categories:
BMI Index Name Range Low <= 22 Med 23-25 Med High 26-28 High >=29
Grouping players by BMI Index then allows us to graph the percentage of players who debuted by BMI Index and decade, as shown below.
As might be expected, the distribution changes over time occur in the middle categories of Medium (23-25) and Medium High (26-28). Fully 70% of the players who debuted in the 1910s and 1920s fall into the Medium category, and only 10-15% in the Medium High category. That all changed rapidly to the point where there are an equal number of Medium and Medium high players by the 2000s, and a substantial increase in players with BMIs of 29 or greater, approaching 10%. Meanwhile the percentage of players in the Low category has changed the least amount, starting at 14.3% in the 1880s, stabilizing by the 1930s, and most recently sitting at 8.6%.
As an aside, the player with the highest BMI, coming in at 38.2, was and is 6'5" 322 pound
Finally, we can recalculate triples per 500 at bats-plus-walks for BMI Index by decade of player debut, as shown in the table and graph below.
BMI Index for Players Who Debuted in the Decade Decade Low Med MedHigh High 1880s 8.94 4.47 6.54 8.45 1890s 5.00 6.35 7.83 7.48 1900s 5.39 6.65 6.55 4.36 1910s 5.70 6.35 6.35 7.55 1920s 5.87 5.68 6.11 4.15 1930s 3.61 4.79 4.84 3.13 1940s 4.22 3.99 3.90 2.41 1950s 4.00 3.80 3.12 1.78 1960s 3.71 3.08 2.78 2.27 1970s 3.92 3.54 2.75 1.83 1980s 4.29 3.27 2.24 2.23 1990s 4.00 2.69 2.31 1.49 2000s 3.80 3.91 2.57 1.78
The general declining trend is immediately visible and indicates that players in the Medium High, High, and--to a lesser extent--the Medium group have seen their triple rate decrease fairly steadily over time. The Low group, on the other hand, has been fairly constant since the 1920s.
So while the downward trend is still evident, the changing distribution of the player population with regards to BMI does mitigate the decline somewhat. In correcting for that changing distribution, the following table shows the triple rate for each decade followed by the rate adjusted for the percentage of players that fell into each BMI Index category during the 1920s. In other words, the second column of the table below shows what the triple rate would have been during each decade had the distribution of BMI Index during that decade been equivalent to the 1920s.
Triple Rate per 500 AB+BB and Rate Adjusted for 1920s Distribution of the BMI Index Decade Rate 1920s Adj 1880s 5.78 5.40 1890s 6.54 6.40 1900s 6.44 6.45 1910s 6.26 6.27 1920s 5.76 5.76 1930s 4.67 4.63 1940s 3.95 3.99 1950s 3.56 3.71 1960s 3.00 3.11 1970s 3.29 3.46 1980s 2.95 3.24 1990s 2.59 2.80 2000s 3.18 3.68
By normalizing the data in this way, we can still see in the graph below that the triple rate fell precipitously starting in the 1910s through the 1960s before increasing again in the 1970s and declining once again in the 1980s and 1990s. So while the changing distribution of BMI in the player population has an effect, it accounts for only around 20% of the difference between the triple rates of the 1920s and those of today, and rates much lower than that for previous decades.
In the end, this leads us to the conclusion that ever-bulkier players are not the primary reason for the decrease of triples since the 1920s and 30s. In reviewing the other theories, we are then left with better outfield positioning, better throwing arms, better cutoffs – in short, increasing excellence on the defensive side of the game – along with the standardization of ballparks as the likely chief reasons for the waning of the three-bagger. Looked at this way, the decrease in triples provides additional support for the idea that there has been an ever-increasing level of play over time.