Monday, June 29, 2015

ON THE NOTION THAT TOP HITTER YEARS HAVE SHIFTED TO "PRE-PEAK"...

We read an essay recently with some semi-elaborate "controls" imposed to "eliminate survivor bias" which came to the conclusion that hitters in the present age come up and have their peak years in the first few seasons of their careers.

It's an astonishing idea...if something like that were true, it would mean that scouting is literally everything (imagine a so-called sabermetric study that proved it all came down to scouting!!) in terms of developing hitters.

So is it true? It could be partially on the money for a reason that bypasses the classical "age-27 peak" formulation--that players don't earn significant playing time until they are already pushing up against what most analysts have called the "peak years" (once upon a time thought to be from 27-31, but more recently seen as 26-29).

Of course, if players don't get significant playing time in MLB until they are 25, such a notion that their "first years" are their best is in "alignment" with what studies have identified as the "peak years."

As always, things can get muddled fast. Such alignment doesn't really tell us much. Its predictive value is tenuous at best. What we need to see is not a comparison for individual players from year to year: such a study actually suffers from "anti-survivor bias."

What we need to see is a chart that shows the degree of elite hitting that occurs at each age, what its distribution over time looks like, and what patterns emerge from that distribution.



And so we give you the great OPS+ distribution chart for ages 25 thru 28, collated for the years 1990-2015. You can see how many hitters had OPS+ figures equal to of higher than 150 at each age for those years, followed by the number of hitters ≥ 140, ≥ 130, and ≥ 120. A weighted numerical formula is applied to create the values in the "Tot" columns at right.

Finally, these are color-coded to show the values for the age-25 hitters in any given year as they move through the next three years, up to age-28. You can see them cascade upward and to the right in the columns at the right.

What you see there is a great deal of individual year fluctuation in the initial age-25 values. But what you also see is that there is often an ongoing correlation with the original established value in those age-25 years as they move across time (to age-26, age-27, age-28). There are "fat years" and "lean years" as represented in the fluctuating totals that you see in the far right column.

And, yes, as Brock Hanke surmised some years back from a different examination based on individual hitter career patterns, there is a good bit of "iambic" progression here--where the numbers go either up, down, up, down or down, up, down, up.

Mostly, though, we see that a really good showing of age-25 hitters in any given year will almost always lead to a high scores across the sequence of following years. Such a pattern is there in the age-25 hitters from 1993: they not only hold their value as they move into "peak seasons," they increase it over those years. There are similar patterns in 1994-97, 1999-2002, 2001-2004, 2005-2008, and 2009-2012.

There is no indication that hitters are peaking at age 25--either here or in the five-year average chart that is the analogue to the above table. Looking at the chart (at right), you can see that the age-25 average trails across most of the time covered by the chart--though it does have a brief flurry in the last 4-5 years. (However, that flurry does not suggest that hitters are having their peak seasons earlier in their careers: age-28 was the leader in the late 90s, only to be replaced by age-27 since 2006 or so.

The chart shows that, on average, the age-27 year five-year averages are rebounding strongly as we get closer to today. While age-28 is showing some slippage from its heyday (1995-2001: probably an artifact of the offensive explosion itself...you can say "steroids" if you must), it's not enough to convince us that any serious shift to younger peaks is underway.

We will endeavor to expand this data into the past, in order to compare it with what we're seeing in recent years, and we'll also try to extend it on either edge of the ages represented here, to see just what kind of curve manifests itself from the total set of ages (and find out if hitters over age 33 are really just "washed up" or not. For right now, though, retain your faith in age-27 as the likeliest year for players to reach elite hitting status. Reports of its demised have been highly exaggerated...