How well did Diamondback starters pitch when the game was close?

At the end of each game, saves measure whether bullpen pitchers held a lead when the game was close. What about the beginning of a game? Is there a similar measure?

Finding an answer required several steps.

Let’s look at the last two seasons (2019 and 2020) for each pitcher in the 2021 rotation. One caveat is that this season’s starters will include additional pitchers.

Let’s limit some sources of variability that could add noise to the answer.

• To reduce the impact of facing batters the third time through the lineup, let’s only look at innings 1 through 5. That can easily be done by a data filter at Baseball Savant.
• To reduce the impact of luck when balls are in play, let’s measure pitching with an emphasis on quality of contact (as determined by exit velocity, launch angle, and for topped/weakly hit balls sprint speed of batter). That emphasis on quality of contact can be found by looking at Statcasts’ expected statistics (xBA, xOBP, xSLG, and xwOBA.) These are available at Baseball Savant. Three of the x-statistics have a link to their definition – just click.
• To reduce the impact of small sample size, let’s exclude pitchers with less than 10 relevant PAs. This had impact. Taylor Widener had only 2 relevant PAs. He was excluded.

What filter is available at Baseball Savant to separate close games? A situation filter allows selection of just PAs with the following: go ahead run on base, go ahead run at the plate, tying run on base, and tying run at the plate. That will work!

The answer was interesting.

The filtered data had a surprise. Three of the 2021 bullpen pitchers had significant PAs in the first five innings, so I added them to the chart. First, let’s look at the chart and then let’s discuss what it shows. The dark green cells are at least 10% above average, while the light green are above average.

Innings 1-5, close games. Data from Baseball Savant.

Two observations:

• The last column in the chart (SDiff) is the sum of differences between the four x-stats for close games (first 5 innings) and the x-stats for all games (first five innings). The purple shaded cells indicate that Zac Gallen and Caleb Smith were the most consistent with their average pitching.
• Please notice the row with averages (weighted by PAs). Zac Gallen and Caleb Smith were above average for all four x-stats in close games. That’s not too surprising because of the Diamondback pitchers who started games in 2020, only Zac Gallen, Caleb Smith, and Merrill Kelly had an ERA below 3. Perhaps the absence of Merrill Kelly at the top was a surprise.

Two observations about players excluded from the chart because they are no longer Diamondbacks:

• If Zack Greinke was a Diamondback, he would be above average for all four x-stats in close games. Zack Greinke was traded away for Seth Beer, J.B. Bukauskas, Corbin Martin, and Josh Rojas.
• On the other end of the spectrum, former Diamondbacks Mike Leake and Robbie Ray were below average for all four x-stats.

Three relief pitchers had more than 10 PAs in the close games (first five innings):

• In 2019 & 2020, Alex Young’s x-stats were an indication that he could possibly be a starter. This indication was strengthened by September of 2020, when he started 4 games. In the best 3 of those games his ERA was 4.70 with 5 innings pitched per game.
• In 2019 & 2020, Taylor Clarke’s ERA when he was a starter was 5.88 and 5.09, while his ERA as a reliever was 3.54. Clearly he pitches better in relief.
• I have no conclusions about Duplantier – his 21 relevant PAs in 2019 was a small sample size which is now 2 years old.

How did the Diamondbacks compare to MLB average for 2019 and 2020?

The MLB averages (close games, first 5 innings) from Baseball Savant follow:

• .247 xBA
• .314 xOBP
• .424 xSLG
• .319 xwOBA

It’s most remarkable that when it matters (close games) every Diamondback in the starting rotation had an xOBP better than the .314 MLB average.

When it matters (close games), the Diamondback were better than the league average in preventing hits and preventing batters from getting on base.

• The Diamondbacks average .234 xBA was better than the .247 MLB average.
• The Diamondbacks .300 xOBP was better than the .314 MLB average.

When it matters (close games), the Diamondbacks allowed slightly higher slugging while keeping the expected runs below average.

• The Diamondbacks .432 xSLG was slightly worse than the .424 MLB average.
• The Diamondbacks .314 xwOBA was better than the .319 MLB league average.

Despite the disappointing 2020 season, in close games the overall Diamondbacks pitching was better than league average.

Summary.

Baseball Savant provided a way to address the question, “How well did Diamondback starters pitch when the game was close?” They performed better than league average, especially for expected OBP! The best two Diamondback starters were Zac Gallen and Caleb Smith.