Jeff put together a sweet graph showing every leverage index where a Royals reliever has entered...

Jeff put together a sweet graph showing every leverage index where a Royals reliever has entered the game this year, with Joakim Soria's situations highlighted. While Soria has the highest total leverage index this season, he hasn't been put into the most important situation yet this year.

(A take-off on Steve's sweet idea.)

about 2 years ago Sky Kalkman 8 comments 0 recs |

Story-email Email Printer Print

Comments

Also telling

are the 2+ LI appearances by others on days that Soria looks available and ended up sitting around twiddling his thumbs.

I wonder if there is a way to correct for sequencing… i.e. the best reliever was user in a high lev situation, but then later in the game a higher lev situation came up.

Big props to Jeff and Tango for taking by crude graph and making it MUCH more useful.

Beyond the Box Score | Play a Hard Nine | Twitter

by stevesommer05 on Apr 28, 2010 10:48 AM EDT reply actions

Trey Hillman

He essentially lost a series to the Rays last year July 17-19 where he did not use Soria in 8th inning situations that were high leverage. All 3 games were lost in 8th with Soria unused in BP.

If I am commenting you need to hear it.

by DeadeyeRR on Apr 28, 2010 11:32 AM EDT reply actions

Not justifying or criticizing any decisions, but...

We can’t look at this and say whether or not the right move was made for two reasons:

1. It could easily be that Soria was unavailable on nights of very high LI, for some reason.

2. When a manager’s managing a game, they have to decide if they think the LI will go any higher at some point in the game. It’s a best guess sort of deal. So a bullpen ace might not enter the game because the manager thinks the LI might get higher later, or conversely, he might enter the game before the highest LI because a manager thinks the most critical situation is occurring/has already occurred.

If that makes sense.

by themastah on Apr 28, 2010 11:37 AM EDT reply actions

I definitely agree. To look at historical LIs and compare decisions to the perfect usage pattern is really unfair.

But there’s a lot of useful stuff here, and you could probably come up with an algorithmic way to compare bullpens/managers.

Beyond the Boxscore Not a member? Sign up.

by Sky Kalkman on Apr 28, 2010 12:36 PM EDT up reply actions

I agree, and...

I think the best way to do it is with some sort of generative modeling, but that might be extremely nontrivial. It’s also going to be highly dependent on the metric used for evaluating the relievers’ skills. Still, I’d love to see work in this…I always thought managers should just use some sort of computer program for making their player swap decisions…maybe I’ll make a fanpost on it in the future.

by themastah on Apr 28, 2010 6:25 PM EDT up reply actions