THE BOOK--Playing The Percentages In Baseball

I often see the difference between starters and relievers presented in the form of ERA+.  This statistic is, obviously, flawed for a number of reasons and is particularly misleading in this context due to the dependence of “earned runs” on inherited runners and the large portion of incomplete innings pitched by relievers.  What is the best “true” measure of the performance difference between the two groups?

First of all, the idea that ERA for relievers is not a good stat because of “inherited runners” is WAY overstated.

ERA is not a great stat for any pitcher, but it is pretty much just as good for relievers as it is for starters.

Inherited runners scored and those kind of stats for relievers are worthless for evaluation and projections.  In fact, I can’t stand them.

However, if you want to use a stat that has nothing to do with inherited runners or actual runs scored, then use component ERA.  It is not that easy to figure out and most sources do not list it.

DIPS ERA or FIP is not that bad either.  Again, it is essentially a regressed component ERA, so it has nothing to do with actual runs allowed, inherited runners, etc.

Anyway, when we talk about the difference between the two groups, we are talking about if the same pitcher pitches in relief and if he pitches as a starter.

Because a reliever can throw much harder for 1 (or 2) innings(s) than a starter can, and because batters generally only see a reliever one time, a pitcher is MUCH (over a run per 9, on the average) better when he relieves than when he starts.

But, relievers as a group are worse pitchers.  That is why they are relievers, generally.

In fact, if we compare exclusive relievers to exclusive starters, they both have around the same ERA (relievers a little lower, I think).  That means that relievers are a lot worse, pound for pound.

But, since when a pitcher relieves, he can shave over a run off his ERA, we can take just about any crappy starter and turn him into a decent reliever.  That is why it is so easy to replace a starter in the 5th, 6th, 7th, etc. innings with a reliever who is a least as good, unless the starter is an absolute ace.  And even if he is, like Santana, if you have a good reliever in the pen, he is going to be as good as or better than a Santana after 100 pitches and 3 times through the batting order.

MGL, I have to give a lot of credit to you and Tango for improving my understanding of relief pitchers through what you wrote in the Book and what you’ve said on the site here.

However, I do have a question about this assertion:

Because a reliever can throw much harder for 1 (or 2) innings(s) than a starter can, and because batters generally only see a reliever one time, a pitcher is MUCH (over a run per 9, on the average) better when he relieves than when he starts.

Do you have any proof (other than anecdotal) that relievers as a group can throw much harder than they could as starters?  The second part makes sense to me since we can measure how much worse a starter is the second and third times through the lineup, but I have never seen a study that has shown that pitchers can throw harder in relief than as starters.  I’m sure some individual pitchers can, but I’m not sure I believe that to be true about relief pitchers as a group.  The one study that I am aware of, by Dan Turkenkopf, found only a small effect (less than one mph).

Mike, depends on what I/you mean by “much”.  1 mph is a lot.  I have never done any studies using pitch TLV (type, location, velocity) data.  I assumed that my assertion is true, and I guess Dan’s study backs it up.

You guys that work with the pitch f/x data simply have to take all pitchers who have thrown in short relief and also as a starter and look at the difference in their average fastball speed as a starter and as a short reliever, and weight that difference by the the minimum of the two pitch numbers (e.g., if the number of fastballs thrown as a starter is less than the number of fastballs thrown as a reliever, than use that number to do the weighting). Then get a weighted average of everything.  That will tell us exactly what the average difference is when a pitcher starts versus when he is a short reliever.  You should be able to do this in about 5 minutes, no?

Example:

Pitcher A
starter 100 fastballs 91.3 mph
reliever 30 fastballs 92.7

Pitcher B
starter 27 fastballs 88.6
reliever 56 fastballs 89.1

((92.7-91.3) * 30 + (89.1-88.6) * 27) / 57

The third factor that I did not mention was fatigue, independent of pitch speed.  Since the research on how starting pitchers’ fastball speed changes as the game goes on suggests that it doesn’t (change much), we have to assume that there are other manifestations of fatigue that plague a starting pitcher, which would not plague a short reliever.

Further support to our expectations is that basically the only component stat that changes between starters and relievers is K/PA (with the expectation that K implies speed, for the most part).  I think it’s easy to see and say that pitchers throw harder as short relievers.  The only question is by how much.  I would think the range is 0 to 4 mph, with a mean around 2.  Just a guess.

I am going to second that “guess” of around 2 mph.  Wasn’t Joba supposedly throwing 98 as a reliever?  The other night, when he pitched real well, he was at 93, I think.

It just surprises me to see outright guesses, a few anecdotes, and “it’s easy to see” substituted and accepted in the place of actual evidence or research.  I don’t expect that here.  Didn’t Bill James teach us that if we could measure something, we should measure it and draw conclusions based on a well-designed study, rather than just assuming that the few instances our eyes have seen and the biased filter we’ve applied to our memories is seeing things correctly?

It’s easy to see that Derek Jeter is a great hitter or great in the clutch, and in fact, he made a great play the other night when I was watching.  Right?

I’d be fine with saying that pitchers perform better as relievers than starters, and we don’t why.  That’s better than going around saying that we know it’s because they throw harder.  We don’t know that.  Dan’s study found an effect of less than 1 mph.  That’s in the noise.  We need to study it more.  We need a bigger sample.  We need to understand if certain types of pitchers can gain several mph on their fastball by going to relief and others won’t gain anything.  Does it depend on body type, does it depend on repertoire, does it depend on the particular pitcher’s fatigue patterns, etc.?

In Dan’s study, some pitchers LOST speed on their fastball when they moved from starting to relief.  That’s not consistent with a universal “every pitcher can throw harder in short relief” mantra.

The reason I’m harping on this point, is that Rany Jazayerli quoted you (Tango) on his blog, saying that you had proved that pitchers pitch better in relief because they can throw harder and, ergo, a pitcher (Joakim Soria) moving from relief to starting would lose 3-4 mph on his fastball.  We don’t know that.

In The Book, the research was very clear that the effect was fairly drastic.  It was also clear that I didn’t prevent any evidence to support the reason as to why.  If Rany or anyone is saying I did so, I’d like to see the quote.

With such a drastic difference (1 run in ERA) clearly something happened.  It’s huge.  It’s not something that will be buried in the noise.  Whether that something is that you throw harder or have more movement on the ball or a combination of both would be the basic reason.  The only other thing that a pitcher does (location, mixing up pitches) wouldn’t seem to apply here.

That said, I haven’t gone any deeper because I’ve yet to go to any more granular data.

I always have a callout for researchers on ideas, and this is definitely one.  We also discussed this a few months ago, where I gave out the list of relievers that could be studied.  As Bill James once said “I can’t do this by myself”.  The sabermetrics world will be in good hands once I hang up my keyboard from what I’ve seen here and at THT and elsewhere.  At this point, I’d prefer seeing more fishermen than me catching more fish.

Mike, I really don’t understand your point.  I mean you are basically correct, but I don’t see anything that anyone has said that is deserving of criticism.

I will bet all the money in the world that on the average a pitcher throws harder in relief than when starting.  Without seeing any data.  No, I’ll bet more than that!

If you are going to question something, question something that you can win.  You can’t win this one.

I realize that you are NOT suggesting that pitchers do NOT throw harder in relief, only that it hasn’t been proven.

Yes, it is nice when we see the numbers, but it is OK to speculate on things that are 99.9% obvious, as long as we at least occasionally say that we are speculating, and we don’t make things up, like, “We have studied the issue and we have shown...”

We have not in this case, and no one that I know of has said so.  If Rany said that, it is his mistake and not Tango’s.  Tango never said that we have “shown” (with the data) or have proven it. 

If I wanted to take the time, I could probably do a Google search and come up with a dozen snippets of me saying something like, “We assume it is because relievers can throw harder because they don’t have to pace themselves,” etc.  Not always, because it is not always necessary.  But you will find exactly ZERO snippets of me saying, “It has been proven,” or, “The data show that...”

And it goes without saying, although I do occasionally say it just to remind people and to explain to the newbies, that when we say something like, “Pitchers pitch over a run better in relief than as a starter,” that obviously that does not apply to all pitchers, but to pitchers collectively.  Does that really need to be said all the time or more than occasionally? If it does, I might as well hang up my keyboard because my fingers wouldn’t take it.

BTW, I admire and respect Bill James, and I am indebted to him (and others like Pete Palmer, my mother, father, etc.) but as far as I know, he hasn’t “taught” me any particular principles. I take full credit for my accomplishments as well as my mistakes.