Blog
June 4, 2012 posted by Patrick DiCaprio

Closer Identifier Algorithm Results as of June 4-An Additional Criteria

Print Friendly

Joel Hanrahan, PIT

Here are the latest Closer Identifier Algorithm results as of June 4. There are still a lot of players with “lose” status. Sadly, CIA chalks up another “miss” as Steve Cishek has ceded the closer role to Heath Bell.

So far, the two misses are very enlightening. The first was Sean Marshall losing the job to perhaps the most talented pitcher that ever lived, and Cishek losing his job solely because of the salary of his competition. Both cases might result in additions to the algorithm at season’s end.

Other changes:

-Frank Francisco rides regression to the mean in his skill set back to “hold” status. The same will happen to Joel Hanrahan shortly. Both are as good bets as anyone in the game.

-Seattle has no pitchers evaluated. It makes no sense at this point; none of the contenders got a save last week and now we have Stephen Pryor in the mix.

-Alfredo Aceves also rides an improved skill set to “hold” status. His case is a good one for the use of an algorithm versus traditional analysis; there are so many conflicts and ambiguities in his skill set that it is impossible to properly weight them all.

-Jim Johnson continues to confound CIA, which still predicts “lose” status. The obvious reason is the fact that his BABIP is below .190 and he has a 100% LOB%, in large part due to his success. This suggests an obvious tweak, which is to incorporate extraordinary BABIP and LOB% rates into the algorithm.

I am hesitant to add any BABIP or LOB% criteria since we expect these to regress as the season goes on. But extraordinary luck will manifest itself in one easy to examine stat: Save%.

-Jose Valverde is another that suggests this tweak, at least last year. There is no explanation right now for a continued hold on the job.

Jim Johnson is 17-17 in converting saves. Valverde was perfect last year. So, we will add the following addition:

Does the player have a SV% greater than 95% with more than 10 chances? If so, CIA predicts “hold” automatically.

For now, based on this addition, we change Jim Johnson to “hold.” Why 10? If we are going to give a player automatic “hold” status we want to make sure it is not a fluke insofar as it is possible. 10 is the minimum I would use. It doesn’t have to be the “optimal” inflection point; we are trying to predict managerial choice to a large extent. And we are trying to set up minimum conditions.

Pitcher CIA result Reason for Change, if any
Jim Johnson Hold New criteria added
Alfredo Aceves Hold K/BB over 2.5
Addison Reed Hold
Chris Perez Hold
Jose Valverde Lose
Jon Broxton Lose
Ernesto Frieri Hold
Matt Capps Lose
Rafael Soriano Lose
Brian Fuentes Lose
Generic Seattle Reliever N/A no saves this week and in flux
Fernando Rodney Hold
Joe Nathan Hold
Casey Janssen Hold
JJ Putz Hold
Craig Kimbrel Hold
James Russell Lose
Aroldis Chapman Hold
Rafael Betancourt Hold
Brett Myers Lose
Kenley Jansen Hold
Heath Bell Lose
Steve Cishek Hold removed from role–MISS
John Axford Hold
Frank Francisco Hold K/BB and BPV gains
Jon Papelbon Hold
Joel Hanrahan Hold New criteria added
Dale Thayer Hold likely losing job on Street return
Santiago Casilla Hold
Jason Motte Hold
Tyler Clippard Hold
Hits
Brandon League Misses
Hector Santiago Sean Marshall
Scott Downs Steve Cishek
Jordan Walden
Brian Fuentes
Rafael Dolis
Heath Bell
Scott Downs
Henry Rodriguez
Carlos Marmol
Javy Guerra
Grant Balfour
Share

You must be logged in to post a comment.