Did the Rays misuse “Cybermetrics” when removing Blake Snell?

\"IMG_6701.jpeg\"

By Nic Osanic

By now you have probably heard that the Los Angeles Dodgers are the 2020 COVID Series Champions. In past World Series-clinching games, the defining moment is usually the result of a game-changing home run, a timely strikeout or clutch individual performance. Unfortunately, this year’s was all about the clutch performance that could have been. The controversy is so divisive that it gave rise to a new term, Cybermetrics.

Just in case you didn’t watch the game and haven’t been on social media since the start of the sixth inning of Game Six, I’ll fill you in. 

\"Photo

Photo credited to Kevin Jairaj — USA Today

Game summary

Thanks to Randy Arozarena’s tenth home run of the postseason, the Rays staked a 1-0 lead before Blake Snell had even taken the mound. Considering the Dodgers gave their pitchers an immediate lead in all three “road games”, this was a welcomed change and suggested that the Rays were not going away without a fight. In the bottom half of the first, Snell backed that up by striking out the side of 12 pitches. Through five dominant innings by Snell and the Dodgers bullpen, the score remained 1-0 Rays. After failing to get any additional run support in the top half, Snell entered the bottom of the sixth inning with nine strikeouts and a lowly single allowed under the hit column. 

Snell was in his 2018 CY Young form, and it seemed as though the monster which had plagued him all season – pitch count – was finally in his favour. He had needed just 69 pitches to get up to this point. The first batter of the inning, A.J. Pollock, popped out to Brandon Lowe on one pitch. Next up was Austin Barnes. On the third pitch of the at-bat, he lined a single into centre field. After a mere 73 pitches thrown, Kevin Cash brought in the Rays top reliever, Nick Anderson. 

\"Photo

Photo credited to Smiley N. Pool — Dallas Morning News

I can sort of see where the Rays were coming from with this move, in some sense. Snell was about to face the Dodgers lineup for the third time, and Nick Anderson was arguably an injury away from being the best reliever in baseball this year (he had a 0.55 ERA but limited to 16.1 innings pitched). However, Snell had struck out Mookie BettsCorey Seager, and Justin Turner six times in their combined six at-bats against him, and Anderson had struggled during the playoffs. 

Anderson immediately fell behind 2-0 in the count to Mookie, forcing him to leave a fastball down the middle of the plate. As you probably know already, Betts turned on it and ripped it down the left-field line for a double, putting runners on second and third with just one out. With Seager batting, Anderson spiked a curveball which got away from Mike Zunino, allowing Barnes to slide home, tying the game. Betts also advanced to third. 

On the very next pitch, Seager rolled a ground ball to the drawn-in infield, but Ji-Man Choi’s throw home was no match for the speedy Betts, giving the Dodgers a 2-1 edge. The Dodgers would ultimately hold on to the lead after gaining an insurance run via a Mookie Betts solo homer in the eighth and a solid shut-down relief performance by Julio Urias

Finally, we get to the main question: Should the Rays have removed Blake Snell?

As an analytics believer (with a soft spot for the Rays), I want to say they made the right move. After all, it is known that pitchers tend to struggle when facing a lineup for the third time much more than the first. Let’s also not forget that the Rays offence only scored one run. It would’ve been extremely difficult for Snell and the Rays bullpen to hold the Dodgers scoreless for the rest of that game. Regardless, I wanted to see just how much of an impact the third time through the lineup has on pitchers. For the sake of sample size, all our data is from 2018 to 2020. Here are the results:

MLB’s “Times Through the Order” splits 2018-2020 

\"IMG_6704.jpeg\"

From a purely ERA standpoint, the third time through the order makes a huge difference, as it consistently makes large increases the deeper into a game you pitch. However, ERAs can inflate during this time as many pitchers leave games with runners on base. It is then up to the reliever who inherits these runners to bail them out. Unless they are pitching past the seventh inning, the relievers who inherit their baserunners are likely not the top bullpen arms, and therefore often allow runs to score. 

Strikeout percentage (K%) also backs up this notion. The more times you face a lineup, the fewer times they strikeout, albeit there is not much difference after you face the lineup once. Walk percentage (BB%) is the opposite. It seems as though most pitchers are wild early into their appearances and settle down the more they pitch. As for quality of contact (Hard%), the third time through the order is when pitchers get hit the hardest. But once again, it is not by much. 

Now for the most surprising metric, xFIP, which is a statistic that tries to isolate defense (which is considered to be out of a pitcher’s control) and focuses on strikeouts, walks, and home runs allowed. By removing defense from the equation, this metric is believed to measure a pitcher’s true skill level and proves to be better at predicting a pitcher’s future ERA than by using their current ERA. 

When looking at xFIP through the order, we can see that these are much smaller intervals than with ERA, suggesting that although pitchers do perform slightly worse the more they face a lineup, the difference is much smaller than ERA indicates. In 2020, the difference between the first and third times through the lineup is about the same, as Milwaukee’s Josh Lindblom (5.16 ERA, 4.24 xFIP) and the Yankees’ Deivi Garcia (4.98 ERA, 4.63 xFIP) (who some have called “The Next Pedro Martinez”) illustrate. Such results tell us that the third time through the order does somewhat affect pitchers, but not to the extent in which you would remove a pitcher who had been cruising. So how does the third time through the order affect Blake Snell?

Blake Snell’s “Times Through the Order” splits 2018-2020

\"IMG_6705.jpeg\"

Snell clearly has some problems with the third time through the order. His ERA inflates significantly, his K% drops by nearly 6% per additional lineup rotation, and his xFIP indicates that his ERA inflation is legit. However, it isn’t that simple. Here is a look at some of his Statcast data:

\"IMG_6706.jpeg\"

Although he loses some of his strikeout ability, Snell continues to allow soft contact and still generates swinging strikes at an above-average rate. Additionally, his fastball velocity stays relatively consistent by only dropping half an MPH. If Snell still has his swing-and-miss ability and generates soft contact, then why does his ERA inflate? The answer lies in his pitch mix. 

\"IMG_6707.jpeg\"

Despite getting hit hard the second time through the order, overall Snell’s high fastball generates good results when hit into play. It also does a good job at generating swinging strikes, as it’s consistent considering his numbers don’t deviate a whole lot from his 15.9% average. However, Snell decreases the usage of this pitch by 8.7% from the first time through the order to the third. If Snell’s usage of his main pitch decreases the further he goes into games, then the usage of a secondary pitch must be increasing.

\"IMG_6708.jpeg\"

Snell’s curveball and slider both generate tons of swinging strikes, and although his slider’s whiff ability drops as he goes deeper into games, 20% is still a commanding rate. Neither pitch is great when contacted, but both produce the greatest results during the third time through the lineup. Since his usage rates essentially stay the same, neither of these pitches are inflating his ERA. Since it’s not one of Snell’s breaking balls, let’s look at his offspeed pitch:

\"IMG_6709.jpeg\"

This appears to be the pitch that hurts Snell the most. During his initial trip through the lineup, Snell’s changeup is great. It generates a high swinging strike rate and weak contact. His second time through the lineup with the pitch is also good but does show signs of decline. During Snell’s third time through the lineup, the pitch’s effectiveness plummets and is worse than his high fastball, which is significant because his changeup was the pitch that Snell had replaced his high fastball with.

\"Photo

Photo credited to Getty Images

Was removing Snell a mistake?

In the end, it\’s hard to tell whether the Rays should\’ve yanked Snell, and I don’t feel like it was a mistake either way. The data leans towards yes, but there are several ways to interpret it. 

His ERA and xFIP are both inflated during his third time through the lineup, and his strikeout rate drops significantly, meaning that batters are putting the ball into play more often. On the other hand, his walk rate also improves slightly, and his quality of contact allowed stays the same despite the number of balls in play increasing. 

Interestingly, it appears as though Snell’s problems with the third time through the order stem from just one pitch; his changeup. If I were Kevin Cash or Kyle Snyder, I think I would have made a mound visit and let Snell stay in the ballgame under the condition that he doesn’t throw any more changeups. Snell’s high fastball, curveball, and slider had been working all night, baffling the Dodgers’ lineup. Additionally, Betts had hit eight of Snell’s changeups into play during his career and produced a .533 xwOBA on them. On all of Snell’s other pitches combined, Betts had a .433 xwOBA, which is still high but much better.  

Overall, Betts had previously been 7/25 with one home run, six RBIs, and more walks than strikeouts against Snell entering Game Six. Knowing this creates a legitimate agreement for removing Snell. However, the pitcher they brought in, Nick Anderson, has his own pitching problems. 

\"Photo

Photo credited to Getty Images

Was Nick Anderson the right guy to bring in?

Although I don’t mind removing Snell in that situation, Kevin Cash needs to have a better alternative than Snell to make this decision. With the outcome of this move aside, I don’t think Nick Anderson was a better alternative than Blake Snell in this postseason. If this were the regular season, I would 100% agree with Cash’s decision to bring in Anderson. As I mentioned earlier, he was one of baseball’s best pitchers this year and likely would’ve been in the running for American League Reliever of the Year had he not gone on the IL with right forearm inflammation mid-season. Surprisingly though, Anderson really struggled after the Wild Card round. 

\"IMG_6712.jpeg\"

Despite striking out six batters in 4.2 innings during the ALDS, Anderson only got one swinging strike on the 13 high fastballs he threw. Since this pitch was his bread and butter during the regular season (46% usage rate, 19.3% swinging strike rate), success without his fastball being effective would seem to be unsustainable. Anderson followed up this performance by struggling mightily against Houston and Los Angeles. In these two series, he surrendered seven runs in 7.1 innings and walked more batters than he struck out (4 BB – 3 K). 

Once again, his high fastball was bad. Anderson generated just one swinging strike on 39 high fastballs and seemed to be missing his location often with the pitch. His velocity was slightly higher during this stretch of the playoffs compared to the regular season, and Anderson’s high fastball spin rate was also down 50 RPM. Since velocity gains normally coincide with increases in spin rate, this suggests that Anderson might not have been 100% healthy. 

When battling arm problems, pitchers sometimes see their velocity and spin rate drop at their normal effort level. This causes them to overcompensate their velocity loss by trying to throw even harder, leading to a loss of pitch control, while their spin rate doesn’t return to the same level. Additionally, Anderson’s fastball thrives due to its ability to spin in a way that produces high amounts of vertical movement. Without as much spin, his fastball will sink and be more prone to contact. 

With so many red flags and a sample size of 12 playoff innings of high fastball ineffectiveness (relevant since his regular-season dominance only spread over 16.1 innings), I don’t agree with substituting Blake Snell for Nick Anderson.

\"Photo

Photo credited to Wally Skalij — Los Angeles Times

In the end, I feel like you have to leave Blake Snell in the game. 
 
The Rays are not wrong about how the third time through the lineup impacts Snell, but I believe they are wrong about how much it impacts him. Snell likely would not have struck out Betts for a third time (Betts is hard to strike out once, let alone in three straight at-bats, while Snell’s strikeouts drop as he pitches deeper into games), but there was a strong chance that he could’ve survived as long as he didn’t throw a changeup. 
 
The quality of contact allowed by Snell during his third time through the rotation is the same as his first run-through. All it would’ve taken to get out of that inning was a ground ball for a double play. Nick Anderson was also not a better alternative than Snell during the postseason. Likely battling some sort of fatigue or further shoulder inflammation, this version of Anderson is not the same pitcher from the regular season. 
 
There is also the hard-to-quantify mental aspect of the game. Snell’s brilliance surely frustrated the Dodgers, and his removal likely gave them hope given their previous success against the Rays’ bullpen. Maybe a different high-leverage reliever such as Pete Fairbanks or Diego Castillo would’ve fared better, but with a lefty in Corey Seager on-deck and Betts’ struggles to hit left-handers this year, I feel like the Rays should’ve trusted their former CY Young winner.

Statistics retrieved from FangraphsBaseball Savant (Statcast), and MLB.com

Title Graphic retrieved from LA Dodgers Twitter, Thumbnail retrieved from Getty Images

Leave a Comment

Your email address will not be published. Required fields are marked *