Blake Snell used a bad word Tuesday night, and no one could blame him. If Snell had chosen to follow up that bad word with one more, there was an easy choice: “Analytics.”
The Rays have turned into this era’s “Moneyball” team in part because of their full embrace of statistics to find any advantage they can. In Game 6 of the World Series, that meant Kevin Cash pulling Blake Snell at the height of his powers, a move that cost Tampa Bay its lead and gave the Dodgers a 3-1, series-clinching win. The long view of analytics points to reasons why pulling Snell was the right move, but it also ignores the individual context clues that Snell’s Game 6 domination provided.
There will always be two sides to the analytics coin in sports — kick the extra point or go for two, whether to encourage even bad shooters to take 3s, whether there’s ever a good time to bunt. So we’ve taken a look at both perspectives of Cash’s decision to pull Snell. Hindsight is 20-20, and Cash said after Game 6 that he regretted pulling Snell “because it didn’t work out.” The Rays might have also benefited, though, from better foresight.
What analytics told Kevin Cash to pull Blake Snell?
Before getting into the third-time-through-the-order penalty or anything else deeper analytically, we can start with this: Snell hadn’t pitched six or more innings in a game since July 2019. Part of that was because of injuries, and Snell wasn’t at full strength at the start of the pandemic-shortened 2020 season. Fox’s broadcast said Snell hadn’t felt fully right this season until late September. That fact alone surely had Cash ready to leap out of the dugout at the first sign of trouble.
Snell’s pitch count was only in the 70s, though, after Austin Barnes singled in the sixth. He’s thrown more pitches than that this season, and those who study arm injuries would note that Snell’s pitches Tuesday night were almost all of the low-stress variety, meaning that he shouldn’t yet have been physically tired.
Cash likely took most into account the success major league hitters have when facing pitchers a third time in the same game. According to Baseball Reference, Snell has allowed a .742 OPS the third time he faces hitters in the same game for his career. In a small-sample size (24 batters) against righties in 2020, Snell allowed a 1.072 OPS the third time, per ESPN (for context, Snell has allowed a .664 OPS to batters overall in his career). Coming to the plate was the right-handed hitting Mookie Betts, who has seven regular-season hits off Snell in his career from his time in Boston.
After Game 6, Cash said he didn’t want Snell to face Betts, Corey Seager and Justin Turner for a third time. All season, that’s the way the Rays have operated. ESPN wrote that Snell only faced 34 hitters a third time in a game all season, and because Cash has confidence in his “whole damn stable of guys who throw 98” out of the bullpen, it’s generally worked to pull Snell early.
Cash also spoke after the game about the pitcher he was bringing in, Nick Anderson. The recency bias toward Anderson is poor: He had allowed a run in six-straight postseason outings (a record) before coming in. But again, the Rays likely took the long view here that saw Anderson as one of baseball’s best relievers since Tampa Bay acquired him from Miami. In 19 regular-season games in 2020, Anderson had a miniscule 0.55 ERA and 0.49 WHIP. Analytics are meant to trust the large sample size, not the small one, and so the long-view numbers told Cash he was bringing in an utterly dominant reliever.
The case for leaving Blake Snell in the game
That’s the thing about analytics, though — they take the long view. For advanced statistics to be reliable, you need large sample sizes. The Rays wouldn’t put on an extreme shift just because the first at-bat of a player’s career was a ground ball to his pull side. They shift because hundreds or even thousands of data points say they should.
Working against Snell were his long-view sample sizes, the not pitching deep into games, the struggles the third time through the order. The case can be made that Cash managed this situation exactly like a computer would.
But analytics also depend on some sort of formula, and the variable that wasn’t accounted for here was Snell’s performance Tuesday night. He’d struck out nine batters en route to his first 16 outs, along with inducing 16 swing-and-misses on his 73 pitches. He had total command of all four of his pitches (four-seam fastball, curveball, slider, changeup). And the top three of the Dodgers’ order that Cash didn’t want Snell to face? They were 0-for-6 against Snell in Game 6 with six strikeouts.
Data sets don’t appreciate outliers, because they make the final analysis a little messier. Snell, the 2018 AL Cy Young winner, may have been pitching better Tuesday than he ever has in his life. That’s an outlier. And while a normal-performing Snell has plenty of data points to rely on in making decisions, the outlier Snell doesn’t. The Rays simply couldn’t know how this extremely dominant version of Snell would perform the third time through the order, because he’s rarely been this dominant.
That’s why a sentence that came across Twitter on Tuesday night stood out: “(Analytics are) facts meant to inform decisions, not make decisions.” The Rays should not only have had information about Snell’s performance the third time through the order, but also on specific pitch details about why he was doing so well Tuesday. The strikeout and swing-and-miss rates alone show that Snell was on a whole different level.
The only other pitcher in World Series history to strike out nine batters through 4.0 innings was Sandy Koufax in 1963, according to the Fox broadcast. Koufax’s opponents had a higher batting average against him the third and fourth time through the order in his career than the first two, according to Baseball Reference. But at his most dominant, Koufax would never have been pulled for that reason.
What Cash and the Rays didn’t recognize was that they had Koufax-level Snell on the mound Tuesday night. The analytics from mortal-level Snell didn’t apply. This was a different night, and the Rays didn’t realize they needed a different equation.
Sadly, there’s no alternate universe or time travel to find out what would’ve happened if Cash walked out to the mound, patted Snell on the butt and said, “Let’s get it.” All we know for sure is that Snell came out and the Rays blew a 1-0 lead and lost the World Series. A misapplied long-view of the numbers cost the Rays and Snell the most important single game in their histories.