Monday, April 04, 2011
There’s nothing magic about age 27
Posted by Derek Carty at 5:11amThere are some fantasy baseball analysts who consider age 27 to be a magic number for players. Eric Mack at Sports Illustrated gives his reasoning for the theory:
History shows 27 is the age many players outperform their draft position because a man's physical peak comes around then and years of preparation allow them to blossom statistically.
It may sound good in theory, but I'm not convinced that history actually shows this to be true. For me, it seems a little simplistic, so I decided to run some numbers and see whether or not it's true, and if so, to what extent.
The study
My study looks at all players since 1953 (the beginning of the modern baseball era) with at least 400 plate appearances or 130 innings pitched in one season. I then compare each player's numbers in that season to the year after, grouping by age.
I place no restrictions on the number of plate appearances in the following season because this would introduce a lot of survivor bias. That is, if a player fails to break out and posts poor numbers, it's likely that his playing time will be reduced.
If we set a plate appearance cutoff, these players would be ignored since they wouldn't reach the cutoff, which we don't want. Sure, we'll get some injuries and some part-time player weirdness mixed in, but I don't think that's as important as capturing these legitimate non-breakouts.
After I had my player pool, I decided to define a breakout as a player outperforming his previous year's production by at least 20 percent. I decided to use four categories: wOBA to measure raw offensive production plus the three main fantasy skill categories for hitters (home runs, batting average and steals). All stats are rate stats, with the rates being HR/(AB-K), H/AB, and SB/AB. For pitchers, I used ERA, WHIP, and strikeouts per nine innings (K/9).
The results
Here is what we get for all ages that had at least 100 player seasons. The numbers list tell us the percentage of players that exceeded their previous year's production in that category by at least 20%.
+-----+--------+------+-----+-----+-----+ | Age | Sample | wOBA | HR | BA | SB | +-----+--------+------+-----+-----+-----+ | 22 | 114 | 9% | 39% | 3% | 27% | | 23 | 256 | 7% | 36% | 7% | 36% | | 24 | 461 | 5% | 31% | 6% | 35% | | 25 | 653 | 6% | 36% | 6% | 31% | | 26 | 854 | 4% | 33% | 5% | 32% | | 27 | 932 | 2% | 32% | 3% | 28% | | 28 | 949 | 3% | 29% | 4% | 30% | | 29 | 902 | 4% | 31% | 5% | 27% | | 30 | 845 | 3% | 29% | 4% | 25% | | 31 | 761 | 3% | 29% | 3% | 26% | | 32 | 660 | 3% | 28% | 4% | 26% | | 33 | 534 | 2% | 27% | 2% | 23% | | 34 | 429 | 2% | 29% | 2% | 29% | | 35 | 331 | 2% | 23% | 3% | 27% | | 36 | 249 | 2% | 24% | 2% | 24% | | 37 | 173 | 1% | 21% | 1% | 17% | | 38 | 110 | 2% | 22% | 1% | 28% | +-----+--------+------+-----+-----+-----+
Essentially, what we see is that the "age 27" theory holds little water. Most of the breakouts happen at younger ages (though part of that may be smaller sample sizes), and 27 is essentially like any other mid-20s age in terms of breakout potential. Once you get into the 30s, the likelihood of a breakout starts going down.
If I change the criteria to either 10% above the previous year or 30% above the previous year, the results are very similar, with age 27 not meaning very much.
But what about pitchers?
+-----+--------+-----+------+-----+ | Age | Sample | ERA | WHIP | K/9 | +-----+--------+-----+------+-----+ | 22 | 102 | 37% | 18% | 14% | | 23 | 208 | 28% | 14% | 21% | | 24 | 352 | 32% | 15% | 17% | | 25 | 457 | 34% | 12% | 17% | | 26 | 552 | 31% | 14% | 13% | | 27 | 578 | 32% | 13% | 15% | | 28 | 550 | 31% | 14% | 15% | | 29 | 495 | 36% | 15% | 12% | | 30 | 458 | 34% | 13% | 15% | | 31 | 369 | 36% | 14% | 10% | | 32 | 313 | 33% | 17% | 13% | | 33 | 263 | 36% | 15% | 14% | | 34 | 209 | 36% | 15% | 19% | | 35 | 163 | 36% | 15% | 12% | | 36 | 127 | 32% | 16% | 13% | | 37 | 100 | 39% | 19% | 12% | +-----+--------+-----+------+-----+
Nope, same story. Pitchers don't tend to break out more often at age 27, either.
Initial, ill-fated idea for the study
Originally, I intended to run this study using projections (namely Jeff Sackman's infinitely useful historical Marcels), comparing a player's actual performance to his preseason projection. When I ran the tests, however, I got some strange results.
I found that hitters in their 30s were the "most likely to break out," which seems incredibly counter-intuitive. My best guess as to why is that there's likely some survivor bias going on that Marcels —being a bare-bones approach to forecasting —isn't accounting for.
You see, if a hitter lasts into his mid-30s as a major league player, he's likely of a different breed than a hitter who fizzles out before then. If a player in his 30s plays poorly, he's less likely to receive another chance than a player who performs poorly in his 20s and still has "potential."
As an example of this, since 2000, hitters over 30 have posted a 3.9% HR/(AB-K), while hitters under 30 have posted a 2.8% rate (that's a difference of about five homers over a full season). Marcels doesn't recognize this, though, and regresses all player stats to the major league average.
But these over-30 hitters aren't major league average; they're better than that. Because they're actually better than what their projections say, they outperform the projections, giving the illusion of a "breakout."
So, while using projections for this kind of test would have been ideal, Marcels doesn't seem suited to the task, and no other projection system goes back far enough to give us a large enough sample. If you were curious, though, when compared to other "20s" ages, the age 27 didn't turn out to be anything special using this Marcels approach.
Concluding thoughts
This is far from a definitive study, but it is one piece of data to consider and certainly doesn't help the "27 is a magic number" theory. If you have any questions or suggestions, as always, feel free to let me know.
Derek Carty, 23, has also been published by NBC's Rotoworld, Sports Illustrated, FOX Sports, and USA Today. This season, he'll be contributing to FanDuel and will be linking to all of his work at DerekCarty.com. In his three years competing in expert leagues, he has won 2 titles with 4 top three finishes, including a LABR NL title in 2009, making him the youngest person to ever win a major expert league title. Derek is a proud graduate of the MLB Scouting Bureau's Scout Development Program and is a firm believer in the importance of combining stats and scouting. He welcomes questions via e-mail, Facebook, or Twitter.
Your sample peeked at 27-28, which seems to support the theory, I still believe you have better stats, less injuries around the 27-28 time frame, and in fantasy, injuries can kill you.