March 3, 2004

Baseball Prospectus Basics

How We Measure Pitcher Usage

by Rany Jazayerli

So to understand the methods we use to analyze pitcher usage, it's important to appreciate that while every team in baseball today employs essentially the same usage pattern--starting pitchers work in a five-man rotation, with four or five days of rest between starts, and never relieving in between--that usage pattern is far from the norm historically.

• As recently as 30 years ago, starters were expected to start every fourth day, with only three days of rest between starts. This does not appear to have had a detrimental effect on the pitchers of that era; in fact, over half of the 300-game winners of the live-ball era were in the prime of their careers in the early 1970s.

• There is no definitive proof that pitching in any kind of rotation is a necessary ingredient for successful pitching staffs. Through the 1950s, starting pitchers would routinely get six or seven days off to pitch against a team they matched up favorably against, then return to the mound on just two days' rest for their next start.

• There is no evidence that starting pitchers who relieve on their days off between starts suffer adversely for doing so. Starting pitchers routinely made 10 or 15 relief appearances a season for the better part of half a century.

So if starting pitchers have been used in many different ways over the years, and there's no hard evidence that any one usage pattern was more likely to keep pitchers healthy, how do we determine whether a pitcher is being used in a manner that's likely to get him hurt?

One thing we have learned is that for starting pitchers, how many days off they get between starts does not seem to correlate with injury risk. This series of articles carefully looks at the track record of pitchers working in a four-man rotation vs. pitchers in a five-man rotation, and finds that pitchers who worked in a four-man rotation stayed just as healthy as pitchers working every fifth day. It also showed that a pitcher working on three days of rest is no less effective than when he works on four days of rest, and in fact that he might have better command on less rest.

What seems to matter isn't how often a starter pitches, but how much he pitches when he does take the mound. About five years ago, we unveiled a system known as Pitcher Abuse Points (PAP for short) that attempted to measure just how much is too much. The system is based on the following principles:

1. While pitching is an inherently unnatural motion, throwing a pitch does not necessarily do permanent damage to a pitcher's arm. It's only when fatigue sets in (and a pitcher's mechanics start to waver) that continued pitching can result in irreversible injury.

2. There is a certain number of pitches that a pitcher can throw before that fatigue sets in.

3. Once a pitcher is fatigued, each additional pitch causes more damage, and results in more additional fatigue, than the pitch before.

The original version of PAP operated under the assumption that fatigue set in at 100 pitches, and after 100 pitches a starter was awarded Abuse Points for each additional pitch. The number of points he received per pitch slowly increased as he threw more pitches.

Two years later, Keith Woolner performed the definitive study that examined the relationship between high pitch counts and injury risk. First, Woolner looked at whether there was a relationship between high pitch counts and decreased effectiveness over the pitchers next few starts. What he found was that, while the relationship was there, the formula for PAP needed to be changed--that until that point, the system did not penalize pitchers enough for really high pitch counts (120 and up) compared to a 105 or 110-pitch outing.

Then using the new, refined formula for PAP, Woolner showed that there was, indeed, a link between high PAP scores and future injury risk.

The way PAP scores are calculated is quite simple. Simply take the number of pitches thrown in any given start, and subtract by 100. (If the pitcher threw fewer than 100 pitches, he automatically receives zero PAP for that outing.) Then the resultant number is cubed to arrive at the PAP score for that start:

100 pitches - 100 =  0^3 = 0 PAP
105 pitches - 100 =  5^3 = 125 PAP
115 pitches - 100 = 15^3 = 3375 PAP
130 pitches - 100 = 30^3 = 27000 PAP

As you can see, by this method a 130-pitch outing is eight times more damaging than a 115-pitch start, and 216 times worse than throwing 105 pitches.

There's one other factor that needs to be considered when evaluating whether a starting pitcher is throwing too many pitches. As first explored by Craig Wright in his landmark book, The Diamond Appraised, starting pitchers under the age of 25 appear to be particularly sensitive to how many innings they are allowed to throw. Some of the most talented young pitchers of the last forty years--from Gary Nolan and Don Gullett in the 1970s, to Dwight Gooden in the 1980s, to the Mets' "young-guns" trio of Bill Pulsipher, Paul Wilson, and Jason Isringhausen in the 1990s--went on to suffer career-threatening injuries that prevented them from ever reaching their full potential. Conversely, two of the most durable starters of our generation, Nolan Ryan and Randy Johnson, weren't even full-time starters in the majors until they turned 25.

Indeed, when 21-year-old Kerry Wood blew out his elbow, the spring after one of the most exhilarating seasons ever by a rookie pitcher, it proved to be the spark needed to convince major-league organizations that lowering the pitch counts of their starting pitchers might prevent a significant number of injuries--and save them millions of dollars in the process.

So to recap, here's everything we know about the usage of starting pitchers:

• There is no evidence that the current system of employing a five-man rotation is any better at accomplishing what it was created for--keeping pitchers healthy--than the four-man rotation. It appears that most pitchers simply don't need more than three days of rest between starts.

• In the era of the four-man rotation, teams were able to get six or seven more starts, and 50-75 more innings, out of their best starters than teams do today.

• Starting pitchers have, historically speaking, thrived without use of a fixed rotation at all.

• Starting pitchers have, historically speaking, been used as relievers between starts without adverse consequences.

• What seems to put starters at risk of injury is throwing too many pitches per start.

• Roughly speaking, "too many pitches" seems to translate to "over 100".

• Once a pitcher hits his fatigue point, his risk of injury goes up very quickly with each additional pitch.

• Pitchers under the age of 25 are exquisitely sensitive to overuse.

The ideal usage pattern--something I think we'll see some teams try to emulate over the next decade--would probably look something like this:

• A reversion back to a four-man rotation, giving a regular starter 40 to 41 starts over the course of a season.

• More careful observation of pitch counts, with most pitchers probably averaging about 90-95 pitches a start, and rarely going over 110 in any given outing. Older, more established pitchers might average closer to 100 pitches a start, with a soft limit of 120 pitches in an outing.

• Judicious use of a starting pitcher on his standard throw day between starts could net another seven or eight appearances and 10-15 innings over the course of a season.

Individual starting pitchers throw fewer innings, and therefore have less overall value, today than at any point in baseball history. But it doesn't have to be that way. By following these guidelines, teams should be able to safely get 280-290 innings out of their #1 starter, instead of the 220-230 innings we see today.

Rany Jazayerli is an author of Baseball Prospectus. 
Click here to see Rany's other articles. You can contact Rany by clicking here