THE BOOK--Playing The Percentages In Baseball

Should we assume the same rate of decline for all starters regardless of quality?  I am wondering if we should use a less aggressive aging factor for better pitchers and perhaps a harsher one for less talented pitchers.  So someone like Johan Santana might decline at a slower rate than someone like say, Jason Marquis?

My blog entry was merely an illustration.  What we should use should be based on empirical evidence.

We know based on the Buehrle thread that the Rule of 10 applies well for those kinds of pitchers at that age.

I don’t have my numbers in front of me (I am still out of town), but it is tough to do the emprirical research to arrive at a “one-size fits all” formula for aging, both in terms of IP and performance rate.  There is a lot of selective sampling and it depends upon the how good a pitcher is to start with (if he is not very good, he won’t last long as he ages) and it depends on his injury history (a pitcher like Sabathia or Buehrle, who have a history of several injury-free years will far better in future IP than pitchers in general or pitchers with a history of injury).

In any case, I just looked at a report I did on Sabathia for the Indians.  He is an example of a starting pitcher with just about the most optimistic outlook you can come with, as he has a history of several years, injury-free, as well as a great rate of performance, at least over the last few years.

I have him (actually pitchers like him) losing 10-15% IP per year.  That includes the chance of not pitching at all, which is significant for all pitchers in any given future year.  In fact, it appears that for a god pitcher like this, with several years of history, still have around a 5% chance of not pitching in any subsequent year, so that, for example, in 5 years hence, a pitcher like Sabathia, has a 25% chance of not pitching at all (I did not look at how often this is temporary - i.e., he pitches the year after that).

Anyway, the rule of ten with respect to IP, is too liberal I think.  The average pitcher is going to be closer to 15%, and the best, most historically durable pitchers are still going to be between 10 and 15% in decreased IP per year.

As far as performance rate, it appears that for a young pitcher (mid 20’s lile Sabathia), the decrease in performance as around .1 runs per 9, maybe a little more.  I don’t know, without looking at the data for all pitchers, whether that is a constant rate or a percentage such that a good pitcher would lose less than a bad pitcher (IOW, it is like a 2.5% decline).

Anyway, if it is a .15 run decrease per year for all pitchers combined, that would be around 3 runs a year, or less than .5 wins.

I also think that that .15 runs per 9 or 3%, or whatever it is, increases at a greater rate as a pitcher approaches 30 and beyond.  It is probably closer to 5 or 6% in a pitcher’s mid 30’s, but that is mitigated by many fewer expected IP, in terms of wins.

And once again, one has to realize that the reason why you don’t see such a decrease in IP among pitchers is that if you are still pitching, the decrease is only around 5-6% per year.  The 10-15% comes from the 5% per year chance of not pitching at all (or significant blocks of time not pitching due to injury).

For offensive players, there is also a significant decrease in playing time per year, but it is a lot less than for pitchers (maybe half, IIRC), and it definitely is related to a player’s past injury history.  And of course, we are much more certain of an offensive player’s (at least all of them combined and averaged) aging curve than we are of a pitcher’s, although there is much controversy over peak age these days.  (In the past, it was pretty clear that it was 27 or so, but recent research has suggested that it may be closer to 29 or 30, although obviously there is a lag in which we can identify changing peak ages because we need lots of years to create a large sample size, AND it may be that widespread use of PED’s may have affected peak age in the modern era and that it might go back to what it was in prior eras, which was clearly around 27.)

Here’s an excerpt from an article I wrote for Heater Magazine looking at this subject, using all pitcher seasons from 1946-2006.

***

What if I were to raise the bar considerably by looking only at pitchers with three consecutive seasons of 200+ innings pitched? That way, we would only be selecting guys with true reputations for durability.

There are in our dataset 849 such pitchers (note that I’ve thrown out pitchers who had three consecutive seasons of 200+ innings pitched after 2001 because we’re going to be looking at their next five years of performance). Here are their averages in those three seasons:

Season IP ERA
Year 1 249 3.24
Year 2 252 3.28
Year 3 249 3.33

Now these are some damn great pitchers, huh? You would expect them to continue to dominate and endure. But do they? Here’s how these same pitchers fare in the next five years:

Season IP ERA
Year 4 211 3.48
Year 5 188 3.55
Year 6 166 3.58
Year 7 142 3.63
Year 8 119 3.64

Their ERAs slowly rise but the real story is how quickly and how dramatically the innings pitched fall. In fact, between year three and year four, these pitchers see an almost identical decline in innings pitched (19 percent) as did the pitchers with just one season of 200 innings pitched (18 percent).

The fall in subsequent years also fluctuates around that number. In other words, even great pitchers are expected to lose about 15 percent of their innings from one yet to the next. But perhaps you’re thinking that the average numbers aren’t really fair to these pitchers.

Some pitchers retire and don’t pitch at all, thus dragging the total averages way down. Maybe most pitchers actually continue to throw a good number of innings. Well, then, let’s look at the percentage of pitchers in each year who still throw at least 200 innings:

Season 200+ IP
Year 4 0.65
Year 5 0.51
Year 6 0.42
Year 7 0.35
Year 8 0.30

So even for the most durable pitcher, the odds that he won’t be totally healthy two years from now are as good as the odds that he will. Four, five years from now, there is just a 1-in-3 chance that will still be able to throw 200 innings a year.

It looks like the “rule of 10” may be closer to the “rule of 15 to 20” in terms of IP.

I have said for years that we generally underestimate the expected IP of a pitcher in years hence.  That is probably due to optimism bias or whatever social scientists call it.

My rule of 10 was documented in the Buehrle thread, and applies to the mid-to-late 20s pitchers who had a ton of IP:
http://www.insidethebook.com/ee/index.php/site/comments/do_pitchers_today_have_it_better_because_of_medicine_and_technology/

Whether it applies to other groups, I don’t know.