RESEARCH: Pitchers' secret weapon for delaying the adverse effects of aging

Recently, I predicted that the leaguewide increase in strikeouts would finally reverse trend in 2015, after discovering that the rise could be largely attributed to unsustainable churn in the player pool, citing that the fountain of higher-strikeout minor league players (both batters and pitchers) that have been gradually replacing outgoing lower-strikeout major league players may have finally run dry.

One of the key assumptions made in that analysis is that the perceived increase in strikeout rates among established players (those who have played in the year analyzed as well as the years immediately before and after) is a statistical mirage caused by recent higher-strikeout incoming players cumulatively accounting for a higher and higher portion of the established group over time, and not that individual established players have somehow been generating more strikeouts with each passing year due to some other factor(s).

Today, we will more closely examine the aging patterns of major league players (specifically pitchers) with respect to strikeouts, both to further validate the assumption above and to see what new insights we can unearth in the process.

How Aging Impacts Strikeouts and Strikeout-Related Skills

The following charts depict the change in skill of major league pitchers as they pass from one age to the next. The highest point on each chart represents a typical peak age for that skill (though some pitchers peak earlier or later than others). The points surrounding that peak on either side illustrate the typical path toward and away from that peak. Each specific point along the path is calculated as the aggregate rate change in skill for all pitchers who played at that age compared to the skill level that those same pitchers exhibited the previous season. (Each pitcher-age pair is weighted by the harmonic mean of batters faced in each of the two seasons to wash out small sample size outliers.) This calculation is repeated for all ages from 22 to 38, and the rate changes are then chained together on the chart to visualize the peak.

Let's start by staring directly into the sun:

Strikeouts (K/PA)

Despite speculation to the contrary, strikeout rates remain relatively flat through a pitchers mid-20's and then start to fade with age. Even in an environment where the leaguewide strikeout rate is increasing by roughly 0.3% (K/PA) per year, when you look at the year-over-year rate change of individual pitchers, their individual strikeout rates are on average declining (by about 0.4% per year). This supports the earlier hypothesis that the leaguewide rate increase is due entirely to churn and playing time shifts (and thus susceptible to reversal if/when even higher-strikeout replacements from the minor leagues fail to arrive).

One subtle difference since we last ran aging curves at BaseballHQ.com (which was 2011; it is good to revisit them every few years) is the steadiness or even slight increase in K/PA during pitchers' early 20s, as opposed to a slow-but-accelerating decline from debut through retirement. Let's look at a few other indicators of strikeout ability -- fastball velocity, swinging strike rate, and first pitch strike rate—to see if any of them offer clues to how pitchers may be able to hold onto their strikeout ability longer than previously believed.

Fastball Velocity

Four-seam fastball velocity declines mostly linearly with age. Even as leaguewide fastball velocity has increased approximately 0.2 mph per season during the period analyzed (again due to churn and playing time allocation in favor of harder-throwing pitchers), individual pitchers tend to lose about 0.2 mph per season off of their respective fastballs.

Swinging Strike Rate

Swinging strike rate follows almost the exact same pattern as strikeout rate itself: more or less steady through a pitcher's mid-20s before turning toward decline after age 30. Again, compare this to the leaguewide swinging strike rate, which on average has increased by 0.1% per year, whereas individual pitchers' swinging strike rates tend to drop about 0.2% per year, and even more so when past their prime.

First Pitch Strike Rate

Now here is something that we've never seen before (anywhere). We know that the leaguewide first pitch strike rate has been rising, particularly over the last five years. But unlike overall strikeout rate, fastball velocity, and swinging strike rate, which all saw individual pitchers exhibit a pattern of deterioration with age, first pitch strike rates for individual pitchers do seem to increase with age until about age 29, and then level out (the extra bump at age 36 is likely a fluke, though interestingly that same bump is present for both LHP and RHP; not shown).

What this implies is that younger pitchers may in fact be steadily "learning how to pitch" (or simply harnessing their control) early in their careers, which enables them to hold or even slightly increase their strikeout rates until their mid-to-late 20s before giving way to the physical effects of aging (namely velocity loss) that result in their eventual strikeout rate decline. It is possible that some of this perceived age-related increase in FpK% is due to a year-over-year churn of incoming batters with higher first pitch strike rates replacing outgoing batters with lower first pitch strike rates. I wasn't able to readily isolate this effect using the data available from Baseball Info Solutions, but even if the aging curve shown above did become more flat, it would very likely remain directionally true.

Conclusion

We now have even more evidence to support our prediction that the leaguewide strikeout rate will finally reverse trend in 2015. Perhaps more importantly, we discovered something new about pitcher aging with respect to strikeouts and associated strikeout indicators—specifically that individual pitchers do seem to throw more first pitch strikes as they gain more experience.

We'll finish by mentioning a few players who, even if the leaguewide strikeout rate does plateau or decline in 2015 as we've predicted, may yet be able to drive their own individual strikeout rates above last year's levels based on supporting factors. The following players were <26 years old in 2014, faced at least 100 batters, and had above average swinging strikeout rates but below average first pitch strike rates. They may be our best bets for eking out a few extra points in the strikeouts category in 2015 if they can bump up their first pitch strike rates as our new finding suggests that they could:

Technical notes: The aging curves in this article were constructed using what is called "the delta method," where each pitcher-year is evaluated individually vs. prior year, and then these individual pitcher "deltas" are aggregated for all pitchers of that age to determine the overall aging pattern. One common flaw with this method, at least when using a composite measure of player value such as WAR, is that there is a survivor bias inherent in the methodology that makes pitcher peaks seem earlier than they otherwise might occur. The idea is that pitchers who "get lucky" in Year 1 are more likely given an opportunity to play again in Year 2, when they inevitably regress and thus contribute a "false decline" to the aging curve. Conversely, pitchers who are unlucky in Year 1 (notably those already nearing the border of replacement level) are never given the chance to redeem themselves, where simply playing up to their true talent level would have contributed positively to the aging curve, thus possibly extending the peak age to a later year or, if nothing else, decelerating the post-peak decline. There are ways to adjust for this bias when constructing aging curves, but these methods were not employed in this study for two reasons: 1) it would have minimal impact on the interpretation of the results (even if the final analysis would technically be more accurate), and 2) I am not convinced that at the specific skill level, the same theory of survivor bias always applies as it does at the result (e.g., runs allowed) level.

All data for this analysis comes from Baseball Info Solutions for seasons 2002-2014.


Click here to subscribe

  For more information about the terms used in this article, see our Glossary Primer.