I don’t remember the site, or the author, or I would give the link here. This crossed my screen just before I set out on The Grand Tour, and I was preparing for that trip rather than consciously banking ideas to investigate once I came back.
I do remember the assertion made, though: that young baseball players, by nature, are more prone to streaks and slumps than those who have been in the majors for several years. The key ingredient is posited as experience, the accumulation of general knowledge and specific meetings against pitchers that makes each at-bat more like every other, less likely to surprise, pleasantly or otherwise.
Nice theory you’ve got there. It’d be a shame if something happened to it. Like contrary data.
Taking age 28 as the average peak of a baseball player’s career, I chose cohorts equal distances from this point, age 23 and age 33, to represent young players and vets, as well as the sweet spot of 28 itself. I looked for seasons between 2003 and 2012 when players of these ages qualified for the batting title (502 plate appearances). I also required that they have a minimum of 60 plate appearances in each of the six months of the season, to avoid small samples creating artificially large swings.
(For purposes of this survey, late March games count together with April, and early October games count along with September. Postseason play is excluded.)
I found the first potential hiccup in my data in how many players qualified fully for the survey. Among young players, 49 qualified for the batting title, and six fell short on monthly totals. For the prime players, it was 147 and 40; for the veterans, 78 players batted enough for the year, but 20 couldn’t keep up 60 PA per month.
The youngsters drop out on monthly totals less than half as often as the older players. One can theorize that older bodies are accumulating more minor injuries that cost them half a month here and there (though it happens somewhat less often for the age-33 cohort than at age 28). Given that nagging injuries could produce more slumps, both before and after a DL stint, this could flatten out the vets’ bumps.
(But is this necessarily bad? The hypothesis is that younger players are more streaky because of inexperience. Physical durability isn’t part of that equation. Trimming out seasons due to injuries, presumably covering more age-28 and 33 players than age 23, may get us closer to answering the specific question, if further from the general one.)
Another hiccup is that this method probably has a bias toward youngsters getting off to hot starts. A 23-year-old who runs cool in April has a much bigger chance to find himself demoted or benched than the 33-year-old does. Granted, you can bench the veteran, which cuts his chances of making the cutoff lines plenty too. Teams, though, are likelier to play someone with a contract the size of the average age-33 player’s longer into a hitting drought, waiting for the rebound. Similar rationales exist for sticking with the age-28 player.
There’s also something of a problem with varying sample sizes, but there is little I can do about that. Youngsters are just less likely to get regular playing time. Going back more years for everyone won’t really alter the ratios, and doing it just for the kids may confound the numbers. I go with what I’ve got.
As for determining streaks and slumps, I used monthly figures for each player. Ups and downs surely come in smaller sizes, as well as larger, and they don’t necessarily conform to break points on the calendar. Again, I go with the data I have available—which in this case provides something admirably suited to the work.
There is this wonderful abstruse statistic in the Baseball-Reference records called tOPS+. The OPS+ part you probably know: on-base plus slugging, adjusted for park effects and normed to the league average at 100. The ‘t’ part here means the norming is done instead to the player’s own total performance. You can thereby measure a batter’s splits against what the batter does overall. It works for lefty-righty, home-away, and in this case, month by month.
I take the variation from 100 as the magnitude of streaking or slumping for each month. The direction of the variation by month does not count for my purposes, only the magnitude. A tOPS+ of 120 or 80 will produce the same variation, 20.
For every player season in the survey (some players got in twice at different ages), I took their monthly tOPS+ splits. I then adjusted them further, against the league-wide tOPS+ splits for the months in question. If July of Year X had a tOPS+ of 106 compared to the overall year, a batter’s monthly split of 120 would be less of a variation from the norm than it appears. I’d revise the variation from 20 points down to 14.
For an example of how this works, I’ll give you Joe Mauer‘s age-23 season in 2006. Note that Mauer’s numbers are relative to himself, not the league, which is why all his monthly numbers aren’t well above 100. Also, due to different PA totals in each month, the numbers won’t necessarily average out to an even 100.
Mar/Apr May June July August Sep/Oct Mauer's tOPS+ 74 110 146 79 76 104 League tOPS+ 100 98 99 106 100 97 Mauer's Variation 26 12 47 27 24 7
That’s a total of 143 variation points over six months for Mauer, which is dead average for his age cohort, as you will soon see.
Before diving into the streak numbers, I’ll take a quick side-trip to overall monthly performance for the age groups. Different ages could plausibly have differing ebbs and flows in how well they bat. While there are some suggestions of that in the numbers, actual steady patterns are more elusive.
tOPS+ by month M/A May Jun Jul Aug S/O Age-23 cohort 104.0 100.8 101.0 93.2 100.8 99.1 Age-28 cohort 97.1 99.8 95.5 105.0 100.8 96.6 Age-33 cohort 93.0 97.0 105.7 98.1 101.9 101.8
For each group, the tOPS+ numbers average out to less than 100, which, the way I’ve set this up, actually makes sense. Players are naturally likelier to receive more plate appearances when they’re running hot, but those bigger PA clumps count as just one month, same as the cold ones. This pulls the average down.
That anticipated selection bias toward fast starts by the youngsters does appear to exist. Less easily explained is their collective slump in July. It could be this is the time when pitchers start getting their second looks at young hitters, start figuring out the holes in their swings, start benefiting from their adjustments before the batters can adjust back in following months. Or maybe it’s luck.
The veterans get off to a decidedly slow start, which is just the stereotype one might invent about older players struggling to get back into playing shape and rhythm. Of course, you’d also stereotype them as getting more worn out by season’s end, and in reality they hang in pretty well down the stretch. You may insert your own explanations here. You can also try to explain the age-28 spike in July: unless it’s pure variance, I’m stumped.
The numbers for streaks and slumps likewise avoid being clearly decisive, but they do give some food for thought.
Variation M/A May Jun Jul Aug S/O Year Per Mo. Age-23 26.9 19.4 24.0 27.3 21.9 23.4 142.9 23.8 Age-28 23.1 25.6 21.7 23.2 20.7 22.8 137.1 22.9 Age-33 27.9 23.9 23.8 23.9 21.2 24.9 145.6 24.3
The veterans show a bit greater variance over the season than the youngsters. As I allowed earlier, the selection bias for the age-23 group may exclude a few early slumpers, suppressing the numbers a little. Even with that taken into account, there’s no evidence to say that young players are more inconsistent than the long-timers. The answer to the question that originated this article is “No.”
The prime players, though, may have a case for themselves. They show less variation than the other cohorts, and consistently too: they have the lowest monthly splits five out of six times. The biggest gap is in March/April, 3.8 points below their nearer competitor. Getting off to a relatively steady start explains much of their margin, but not all.
The consistency shows through in individuals’ numbers as well. The age-23 cohort had six seasons with a variance of 200 or higher, and only two at 75 or lower. For age-33, it was eight at 200+ and three at 75-, a similar ratio. At age-28, however, there were 11 apiece going very high or very low. Great streakiness was about a quarter less common for the prime players, and great steadiness was more than twice as common.
The distribution suggests that prime-age players get a part of their prime performance through consistency. They may avoid the worst dips that knock down the averages of both younger and older players. If experience has its effect on steady performance, maybe the slow decline of the human body does as well, and the age-28 peak represents a sweet spot for steadiness as well as overall ability.
More study may be indicated. (That’s intellectual-speak for “Let someone else handle this.” I like how that sounds.)
There are some individual performances that, while not doing much to illuminate the overall question, still have interest. Someone in this study had to have the most streaky performance, and the winner is Jason Kendall, age 33 in 2007. This is what a real up-and-down year looks like.
Mar/Apr May June July August Sep/Oct Kendall's tOPS+ 32 55 148 78 202 68 Adjusted Variation 61 42 48 22 99 37
Kendall racked up 309 variation points, the highest total of the 208 players in the survey. The trick seems to be getting two big spikes in the same direction, which is admittedly tautological: the trick to being streaky is to have streaks.
The competition for the steadiest performance was tighter, for a while. Big names cropped up here and there: Albert Pujols, age 23 in 2003; Derek Jeter, age 33 in 2007. But one guy beat them all, and it wasn’t close.
It was Ryan Braun, just last year. The 28-year-old put together a monthly tOPS+ line of: 98, 108, 103, 93, 102, 97. After monthly adjustments, his season variation point total came to a microscopic 21, less than the mean variation for a single month. His closest challenger was 2009 Paul Konerko, with 51, a veritable seismograph compared to Braun.
Two questions pop up from the Ryan Braun outlier. The first, regrettably, is what often pops into mind regarding Braun: PED suspicion. I suppose it’s possible this is somehow a result of performance enhancers. A brief check of players I surveyed who have been linked to PEDs in various ways came up inconclusive: three above average, three below. Braun’s previous four years give two results below average and two above. Absent a lot more evidence, I’ll call it luck.
The second question is how much Braun’s rock-steady 2012 bends the numbers. Luckily, not much. Without him, the age-28 cohort’s monthly variance would be just a tenth of a point higher. Their steadier performance is not due to him alone. Besides which, age-28 players had 107 of the 208 seasons I surveyed. By pure chance, odds were that the lowest variation would belong to one of them. Then again, that’s also true for the highest variation, and that ended up in the age-33 bucket instead.
Younger players do not appear to be more prone to streaks and slumps than older ones, but players at the peak age of 28 do look somewhat steadier month to month than either surrounding group. Also, older players are prone to slower starts, at least within the boundaries of this survey.
So the next time a long-time player talks about some fresh kid’s bat being on a roller-coaster, look instead at the fellow speaking. It’s just as likely he’s having his own highs and lows as well.