Changes to Starting Pitcher Ranking System

July 3, 2018
 

Modifications to the World’s #1 Starting Pitcher System

 

              For the first time since we introduced the system five or six years ago, we are making some small changes to the system for ranking starting pitchers.   There are four or five changes, depending on what you define as a "change".   The changes are:

              1)  We have changed the weight given to the last start from 3% to 3.5%,

              2)  We have changed the "Entry Point" for a new pitcher from 300.00 to 325.00,

              3)  We have increased the "decline rate" for inactive older pitchers (will explain shortly), and

              4)  We have amended the "start" concept so that it counts a relief appearance in which the reliever enters the game in the first four innings and pitches 4 innings or more as if it were a start.

              The first two points. . .the system as it has been was a little too slow to move a young pitcher up to the top ranks, a guy like Luis Severino or James Paxton or Jose Berrios.  

              I was sort of aware of this anyway, while I was developing and testing the system, but I just decided that I liked it that way.   I like seeing a young pitcher work his way through the pack, and I like to force him to sustain his performance for 40-50 starts. I still do, actually.

              But at the same time, it was sort of self-indulgent of me to design the system that way because I personally like to see it that way.  I recognize that it is more realistic the other way.   As of June 18, 2018, we still had Chase Anderson (5-6, 4.54 ERA) ahead of Blake Snell (8-4, 2.58).   We still had Masahiro Tanaka (7-2, but a 4.58 ERA and on the DL) ahead of Charlie Morton (8-1, 2.94, and a fine season last year as well.)   That’s not really realistic. 

              If you start pitchers out at a higher platform (325 rather than 300) and move them up a little bit faster, young pitchers on the way up pass older pitchers on the way out more quickly.   They pass pitchers who are just kind of on idle—a couple of good starts followed by a couple of not so good starts—more quickly.  Most pitchers most of the time are really just idling.  On the other hand, the big cluster of pitchers who start the season in a tie at the bottom of the system will be even larger, which I do not like, but. . .you gotta do what you gotta do.   In the system as it was, 60% of a pitcher’s score was based on his last 30 starts, and 40% was what he had done before the last 30 starts.  Now, 66% of a pitcher’s score is his last 30 starts, and 34% is based on what he had done before the last 30 starts.

              The fourth change is to count a 4-inning relief appearance starting in the first four innings of a game as a start, and score it by the Game Score System as if it was a start.  This is necessitated by the Rays’ "opener" experiment, which may spread to other teams.   If the "starter" pitches one inning and the first reliever pitches 7, we treat the first reliever as the starter. I’m guessing you see the point?   It’s screwy to credit a one-inning reliever as a "starter" because he pitches the FIRST inning.

              The final thing, which we could call the R. A. Dickey adjustment, is that we are going to move OLDER pitchers out of the system more rapidly when they retire.  R. A. Dickey, who is 43 years old and has not pitched all year, is still in the system; he still shows as the #132 pitcher in the system (as of June 18).  That’s not TOO bad, but early in the season he still ranked in the top 35 or something, which was kind of embarrassing.   The system was that when a pitcher was inactive for seven days or more, he would lose 0.25 points per day due to inactivity.   He lost 0.25 points for day until day 200 of his inactivity, at which point it increased to 1.00 points per day. 

              We have modified that in two ways. . .not sure whether that makes this four or five changes.   Anyway, we changed the 200 days inactivity to 190.   It’s dicey either way.  A lot of times a pitcher isn’t quite ready to start the season in April, but he’s still fine; he just need a little more time to get ready.   This adjustment will hurt us on THOSE guys, but it will help us on the pitchers who retire; it will help us get them out of the system faster.   Well. . .ten days faster.

              The other thing is, for an OLDER pitcher, a pitcher who is 35 years old or more, we increased the decline rate after the pitcher has crossed the 200-day threshold (now a 190-day threshold).   For a pitcher aged 35 or older, that 1 point per day decline in his score is now 1.50 points.   If a pitcher is good enough to hold a job and isn’t pitching at age 34, we’ll assume that he may be coming back; if the same thing happens at age 35, we’ll assume he may be retired.   And either way, we may be wrong, but it’s a better percentage bet. 

              Thank you. 

 
 

COMMENTS (18 Comments, most recent shown first)

MarisFan61
(how which manager got Severino to try what?)
1:29 PM Jul 17th
 
Brock Hanke
Bill - I don't know how the manager got Severino to try this. Back in the 1990s, when I actually had a press pass, I tried to interview Tony La Russa. Tony is as intimidating as his rep, but I realized, once I'd gotten out of there and calmed down, that he'd answered the question I asked. I asked, "Why did you give up on the "three pitchers, three innings each, three-man rotation concept?" He basically replied that he could not get anyone to start. No one wanted to go into contract negotiations with an ERA of 2.75, but a Won/Lost record of 0-12.
8:58 AM Jul 15th
 
Steven Goldleaf
An excellent illustration of why I stopped listening to Francesa.It no longer became fun to get irritated by his bloviation.
8:33 AM Jul 8th
 
MarisFan61
Not that Mike Francesa or his callers are necessarily any store of sabermetrism, but early last season there was a call that encapsulated the change being made here. A caller asked if Francesa thought Severino was now the Yanks' "ace." Francesa ridiculed it (in the extreme way that he does), saying that Severino still had a lot to prove and that Sabathia and Tanaka, although not going as well as him and in fact being pretty poor, were established. Leaving personality aside (as well as what anybody means by "ace"), it was basically about how much we should weigh "recent."
11:26 AM Jul 7th
 
chuck
How about the Cart for the relief-starter, and the Horse for the long man?
12:20 AM Jul 7th
 
klamb819
It's probably too late to call them Appetizer and Entree. ;-)
11:51 PM Jul 6th
 
JohnPontoon
Good job on the changes. Great job, as well, of choosing examples of guys who are obviously better right now than other guys you named, hence your reasoning. Mr. James, you've still got the stuff, sir!

For the long-innings pitcher/headliner/opener/etc, I suggest "Rotation Man" or "Rotationer." As always, I accept that humanity will take or leave my suggestion as it pleases.​
6:41 PM Jul 6th
 
jemanji
Ya. In chess ratings, we have juniors - age 14, 15 - who are gaining strength VERY rapidly, and they "victimize" opponents in this way.

If you have a junior whose actual RATING is 1600, but whose actual PLAYING STRENGTH is 1900, then the person who loses to him gets a huge rating loss for losing to a "1600" player.

So in chess there's real incentive to "get it right" in terms of where a quickly-rising player actually stands. Juniors get big bonuses for "out of alignment" performances.

Even though the BJOL Starting Pitcher Rankings don't have this incentive to be accurate, it's still nice to see this tweak for players like Severino and Paxton. Agree very strongly with the changes to the system.

- Jeff


5:15 PM Jul 6th
 
tangotiger
Headliner is the best name so far.
4:14 PM Jul 6th
 
MWeddell
Well, I think the name "opener" is already in widespread use. Did Brian Kenny coin it first? Anyway, that ship has already sailed.

For the guy who the manager intends to pitch the most innings but didn't start the game, yes, that guy, the name is still in flux. Tangotiger likes the "headliner" too and that might catch on given that he directly influences what Statcast calls it.
12:18 PM Jul 6th
 
cderosa
Love the starting pitcher rankings.

If the Rays' first-inning pitcher strategy does catch on, I don't think "opener" is the optimal name, though it is better than any variation on LOOGY, such as first-inning guy (FIGGY). After giving due consideration to the opener, the fore-closer, the teaser, the gambit pitcher, and the preemptor, what I think we should call this pitcher is: the spoiler.

The bigger question will be what to call the what we now term starting pitchers when they no longer start the game. They could be long men, assignment pitchers, rotators, horses, anchors, stackers, and/or headliners.
7:40 AM Jul 6th
 
tangotiger
Bill, does this mean that you can have two "starts" for one team potentially? If so, ok.



7:28 PM Jul 4th
 
Arrojo
Love Baseball Musings. Speaking of Dave pinto, he was a member of our Strat-O-Matic league, The SOMBILLA, for a number of years. Although he never lived down drafting Steve Decker in the first round one year.
6:48 PM Jul 4th
 
StatsGuru
This is David Pinto. I just saw the article and will implement the changes as soon as I have the opportunity.​
5:45 AM Jul 4th
 
bjames
Responding to Chuck. . ..I think Baseball Musings is run by Dave Pinto, who is an old friend. I’m sure he’ll see this on his own, but if he doesn’t it’s fine for you to tell him. Thanks for plugging his site; I should do that myself more often.

Responding to Maris. . . I frequently try to measure things which
(a) Are subjective, or
(b) Have unclear definitions.

What is a dynasty? What is a Superstar? What player today is most similar to Willie Mays? What does “most similar” mean? When I do that, I have to make choices, and I am, in essence, choosing for other people; I am trying to choose answers in such a way that:

1) I believe they are correct, but
2) Other people will accept them.

It doesn’t seem to me unusual to run into this problem of choosing between what I think is best, and what other people think is best. In this case I chose wrongly, but again, I don’t think that’s unusual; sometimes I get it wrong on the first effort, and the second, and the third, and the fourth.

12:19 PM Jul 3rd
 
MarisFan61
Like it.

BTW, to me, almost the most interesting thing is the fact of a distinction between what you "like" and what you saw to be best!

Bill, I'd love to know, are you aware of any other such examples where you did a sabermetric thing a particular way because of how you preferred a phenomenon to be, and then saw that the thing would be better with less of that preference in it? I know that you've refined just about everything, but the things you revised never seemed to have involved some 'world view' or 'ethics' factor, as it seems this did (pardon if I'm characterizing it wrong). I wouldn't be surprised if this example is unique in that regard.
11:40 AM Jul 3rd
 
chuck
Bill, thanks for this update and for the fun concept and system to begin with.
I was working with game scores and with starting pitcher rankings recently, and the Rays thing was kind of a chore to work around.

On that issue, for those occasions when their True (2nd inning) starter gets lit up and doesn't complete the 4 innings, his start wouldn't be counted, right? May not happen that often, but if this thing catches on with other teams, it would mean a bunch of poor starts get left out. This thing is probably automated, but perhaps there's a way to identify these True starters from their innings in previous outings, and count their game scores if they're coming in in the 2nd, regardless of outcome?

Also, there's a site- Baseball Musings- that has a cool feature which allows one to see the starting pitcher rankings and scores on any date, going back to 1957, I think:
www.baseballmusings.com/cgi-bin/PitcherRatings.py
It gets updated to 2 years prior to the present date. Are you ok if I let them know about these updates to the system? I'm guessing they may want to implement them.

The studies I've done looking at Hall of Famers' median career scores and the average of their best six scores (by "scores" I mean their score the morning after the last day of the season)... the median career scores typically start around 515, and the average of their six best start around 550.
There are only a couple of exceptions of pitchers not meeting those thresholds (Glavine is the main exception), while there are several pitchers who are comfortably above both levels but not yet inducted (Schilling, Mussina, Halladay).
10:07 AM Jul 3rd
 
Steven Goldleaf
This is a great and useful concept--is there any way you could develop a similar system for batters? Thanks for all the work you've put into developing this one. It's one of my favorite metrics.
6:02 AM Jul 3rd
 
 
©2018 Be Jolly, Inc. All Rights Reserved.|Web site design and development by Americaneagle.com|Terms & Conditions|Privacy Policy