Scherzer’s Time
Max Scherzer moved ahead of Clayton Kershaw in the Starting Pitcher Rankings last week, and I tweeted this out. . ..OK, in unnecessarily dramatic language.
IT'S HAPPENED! Max Scherzer has passed Kershaw to become the #1 starting pitcher in baseball.
This proved a rather controversial comment, and a number of people asked me to explain or defend my analysis. What, in Twitter?
A surprising number of people responded that in order to say who was the best starting pitcher in baseball, we needed to define what we meant by "best" starting pitcher—surprising, since I had not used the word "best". I was being asked to define a word that I had not used.
Let’s assume that I had used the word "best". . ..is it then true that I need to define the word?
I wouldn’t say so. If I said that one of the trees in my back yard is now the tallest tree in the yard, would I need to define "tall"? If I said that Justin Turner had the reddest hair on the Dodgers, would I need to define "red" or "red hair"?
One does not need to define terms which have a common understanding. I would be inclined to believe that the terms "best starting pitcher" or "good starting pitcher" have a common understanding. I would argue that neither you nor I could write a definition of what we mean by the best starting pitcher in baseball which is preferable to the common understanding of the term. Therefore, what one needs in order to say who is the best starting pitcher in baseball is not to write a definition of "best" starting pitcher, but rather, to write a set of formulas which matches the common understanding of the term.
That is what I tried to do, in the Starting Pitcher rankings: I tried to write a set of formulas which matched the common understanding of the term. To the best of my knowledge, I did pretty well.
  One gentleman who is well known and well respected in our field has insisted that Scherzer (a) is a half a run a game behind Kershaw, and (b) is not one of the ten top pitchers in baseball, or is barely one of the top ten pitchers in baseball, or something along that line, and offers a list of the top ten pitchers in baseball, at least four of whom are currently on the Disabled List. Well, of course he is entitled to his opinion, but it would not seem to me that that statement has enough credibility to warrant a response.
In the common understanding of "best starting pitcher", what is meant is the pitcher who has the most value to his team. A pitcher who makes 35 starts in a season has 40% more value than a pitcher who makes 25 starts in a season. A pitcher who pitches 7 innings a start, or 6.2 innings a start, has more value than a pitcher with the same level of effectiveness who pitches 6.1 innings a start.
Part of the problem is that our friend, I believe, is looking at the problem through the lens of a gambler. The gambler’s interest is that given that two pitchers are pitching, which one is more likely to win? To the gambler, the distinction between 35 starts in a season and 25 starts in a season doesn’t mean anything, because the gambler doesn’t bet on the pitcher when he doesn’t start. The only thing that matters is how he pitches when he pitches.
In this way, the gambler’s definition of "best starting pitcher" is not the common definition of the term. But it is letting our friend off the hook too easily to suggest that this issue is the ONLY problem with a ranking of starting pitchers that has difficulty identifying Scherzer as one of the top ten.
One of the key questions is what weight one places on recent performance, as opposed to what weight one places on performance that happened years ago. I have heard from people who think that Chris Sale should be #1 based on how he has pitched this year—and a year ago, I heard from a lot of people who thought that Noah Syndergaard should be ranked higher based on how he pitched in 2016. Two years ago, I heard from people who thought that Arrieta should be number one because of what had done in 2015. But in my view, you have to prove what you can do over a period of time longer than one season to be considered the #1 starting pitcher in baseball.
On the other hand, I have heard from people who site career ERAs and career won-lost records and career WAR, as if what Scherzer or Kershaw did in 2009 was as relevant to where they are now as their last start. That makes NO sense.
My system rates the pitcher’s last start more heavily than his next-to-last start, and his next-to-last start more heavily than the one before that. It does that, because that’s the right way to do it. The last start IS more relevant than the previous start.
I weight each start at 3%, so that the last 23 starts are 50% of where the pitcher ranks, and what he did more than 23 starts ago is 50% of where he ranks. I could easily change that, right? I could easily make it 4% for the last start or 2% or 2 ½%. I chose 3% because it seems to me that that makes more sense. 3% for each start, then after a full season (33 starts) 63% of the ranking is what you have done this season, 37% is what you have done in previous seasons. That seems more right to me than saying either:
  a) What happened in 2009 counts the same as what happened this year, or
b) What happened before this year is totally irrelevant.
If you go to 2% for the last start, then even after a full season, where you were a year ago still accounts for more than half of your rating. That doesn’t seem to make sense, to me. If you go to 4% for the last start, then it seems to me that the past is wiped out too rapidly, and you wind up with guys at the top of the list who have not proven that they can handle a 33-start, 200-inning workload year after year. But more research on this issue might be in order.
  We make park adjustments for every start, and also, we adjust for the quality of the opposition offense on every start. We do it the way we do it because we think it’s the right way to do it. If you can persuade me there is a better way to do it, I’ll do it the other way.