Late Inning Shutdowns and Meltdowns by Starting Pitchers

In a follow up from my article on Monday, I decided to specifically look at the performance of starting pitchers in late game situations to quantify late leads or close games blown, for Yu Darvish specifically and in the context of the league.

I borrowed a concept used for late inning relievers called Shutdowns & Meltdowns, which was initiated from a discussion on Beyond the Box Score then hashed out The Book blog, and is available on FanGraphs. In order to gain a Shutdown or a Meltdown, a relief pitcher's total cumulative WPA must either improve the team's odds of winning by 6% or more (shutdown) or hurt the team's odds of winning by 6% or more (meltdown).

In order to take a similar view for starting pitchers, I used the Event Finder tool from Baseball-Reference.com to pull every plate appearance from the beginning of the 2013 season through Monday, August 26th. I aggregated the events in the inning by pitcher for an Inning Summary that had a total WPA accumulated and rated each inning as either a Shutdown, Meltdown or Low Impact using the same 6% standard for the Shutdowns and Meltdowns and everything in between for Low Impact.

A pitcher can receive a meltdown even if no runs score by leaving an inning early after allowing runners on base for a relief pitcher to deal with, or a low impact after allowing lots of runs if the game is a blowout and the lead goes from 10 to 7. Shutdowns are almost exclusively scoreless innings in close games with the lead or the score is tied.

I limited the pitchers to evaluate only those who have appeared in at least 100 innings this season (though not necessarily 100 innings pitched, because innings with zero or partial outs are considered as an inning appearance). There were 130 starting pitchers in the result set.

After tallying the results, I determined a meltdown percentage (MD%) and shutdown percentage (SD%) as the percentage of meaningful appearances (MD + SD), disregarding low impact innings in the ratio. Starters are ranked by the lowest MD%, with Yu Darvish coming in 45th out of 130, a little better than the average of all 130 starters. Last year's numbers for Yu Darvish were also included in the list for comparison. The table below lists the top ten lowest MD% starters in the 6th inning or later, as well as several selected pitchers with either a top of the rotation reputation or of interest to Rangers fans. Additionally, the last place pitcher is listed for context (and hilarity).

The 8 late inning meltdowns for Yu Darvish show that Eric Nadel's eyes did not deceive him, meltdowns that hurt the team's chances of winning have occurred at a middle of the pack rate.

An important note on the RA/9 metric listed in the table is that it does not consider inherited runners, just runs scored while the pitcher was on the mound. Also, for those unfamiliar with RA/9 versus ERA, RA/9 doest not differentiate whether a run was earned or not. All runs, regardless of scorer judgement are in RA/9.

Here is a Google spreadsheet with tables for the data on all 130 pitchers, including a breakout of the first five innings and overall game. Each inning-by-inning summary is included as well its associated date.

A final note of interest, in the linked spreadsheet that shows similar data for the first five innings, Yu Darvish has the 5th lowest MD% and Cliff Lee the 7th lowest, compared to their 45th and 55th respective late inning rankings. Both are doing a great job of infrequently putting their team in a difficult position to win early, but not excelling at shutting down the late innings. But hey, at least neither are having the year of CC Sabathia!

2013 Starting Pitcher MD/SD, 6th Inning or Later - Top 10 and Selected Starters (min 100 overall innings appeared in)