Sandy Alcantara and The Perfect (Game) Score

It happens every year when a (non-unanimous) Cy Young winner is announced, inevitably pitcher A will have the advantage in volume stats, and pitcher B will have the edge in rate stats, and the online warriors knock each other over the head with:

While it’s impossible to know what would have happened if Pitcher B threw 22.2 more innings, we can know what Pitcher A could have done throwing less innings.

The perfect game score isn’t the highest game score possible, it’s the idea of what a stater’s Game Score could’ve been, if the manager knew exactly when a start was about to go bad. Picture if you will, pulling a starter being like a quiz for the team’s manager, except the manager already knows what the perfect answer to the quiz is going to be before the game starts (much like the acclaimed 2004 feature film “The Perfect Score”).

To simplify this exercise I’m using Game Score Version 2.0 created by Tom Tango. The major difference between Bill James’ original Game Score and GSv2 is that there is no bonus for innings pitched after the 4th, and you are additionally penalized for giving up HRs. Also GSv2 doesn’t distinguish earned and unearned runs.

In every event of a game (batter faced, runner picked off, runner caught stealing, etc.), the Live Game Score for a starter changes… lets take Alcantara’s August 10th from last year as an example. This was a tough loss for Sandy, who pitched into the 8th with a 2-1 lead after giving up only 2 hits in the first 7 innings, only to surrender 6 hits and 3 runs in the 8th inning.

Had Sandy been pulled for a reliever after the 7th inning, his GSv2 would have been 77, by the time Sandy left the game his final GSv2 was 60, let’s illustrate the play-by-play below:

Sandy’s pitching line that night was:

Sandy’s Perfect Score pitching line would have been:

For illustrative purposes we’re taking a very intuitive example where it’s plain to see, “yes pulling Sandy after the 7th would have been clearly the right move”. However, there are more outlandish examples of this that involve Alcantara being pulled in the 4th, 3rd and even 2nd innings. Is that realistic? Absolutely not, but that’s why this is the Perfect Score, because it assumes Don Mattingly already knows what is the right move to make based on what was going to happen. Here’s Sandy’s 2022 game log with all the “Perfect Score” corrections highlighted in blue:

First the overview; Sandy Alcantara in 2022 made 32 starts. Of those 32 starts, 17 starts ended with the Perfect Score, meaning Sandy was lifted at the perfect time (or finished the game himself). This leaves 15 starts where Sandy should’ve been lifted earlier in the game to achieve an optimal GSv2 result.

Given the methodology above, hearing examples of Sandy being pulled in the 2nd and knowing that Sandy was left in too long in almost half of his starts, how many less innings do you think Alcantara would have pitched if he achieved “The Perfect Score” in all 32 starts? 40 less innings? 60 less innings? 100 less innings?

Sandy pitched 228.2 innings in 2022, most in baseball. Sandy being pulled at the “perfect” time would have only lead him to throw… 208.2 innings, still the most in baseball.

Meanwhile how many less runs would Sandy have given up? Sandy gave up 58 earned runs in 2022. With those 20 fewer innings Sandy would have given up… 25 earned runs.

The summary of season totals:

The obvious drawback is “you’re pulling Sandy in the 2nd inning and that’s crazy”. The 20 extra innings doesn’t sound like much, but remember that the 20 extra innings usually involved Sandy getting batted around, so in getting those extra 60 outs, he faced an additional 123 batters, which the bullpen would have had to absorb.

In the greater context of the 2022 Marlins season, starters faced 3630 batters and relievers face 2426 batters. So 123 additional BF for the bullpen would’ve been an increase of 5.1%, and this is for just 1 guy who only started a fifth of their games. If Mattingly used his future-vision superpowers to pull every starter early, that would be adding about 623 batters faced on the bullpen bringing the total above 3000 batters faced by the bullpen. There’s only been 4 teams in history that had their bullpen face +3000 hitters; 2019 Blue Jays, 2019 Rays, 2019 Angels, and 2018 Rays (basically teams that heavily relied on the opener all year).

Now that the methodology is established we can compare how close a starters season was to their “perfect score” season totals. The biggest question is going to be how close to “perfect” was Corbin Burnes pulled in every start in 2021, and how does it compare to Zack Wheeler?