Kevin Cash's decision to pull Blake Snell was an error. But the war on analytics? That's stupid.

Last night, the Dodgers won their first World Series since 1988. Clayton Kershaw finally secured his ring. Mookie Betts blasted a homer. The feel-good Kershaw columns predictably found editors late last night as he exercised his demons and shoved a postseason narrative into a casket. But it’ll be impossible to discuss the 2020 World Series without highlighting two managerial moves.


I don't have the brainpower or sanity to discuss Justin Turner being removed in the eighth inning of a World Series game because of a positive COVID-19 test. MLB implored Turner to isolate. He gave it the middle finger and celebrated on the field with his teammates, removed his mask for photos, kissed his wife, hugged and high-fived teammates and even sat beside his elder manager who beat Hodgkin’s lymphoma years ago. For lack of a better description, it's a very frustratingly 2020 way for baseball's season to end.


But we are going to dissect how Tampa Bay used its pitching staff. And by dissect I really mean I'm going to pour every thought from my brain onto the page. So, I recommend a coffee break. Actually, make it a beer because there are a lot of words.

Kevin Cash and Blake Snell will remember this mound meeting for a long time. AP Images


In the sixth inning with one out and his team leading 1-0, Rays manager Kevin Cash confidently and calmly exited his dugout and soon motioned to the bullpen. Blake Snell, who had given up just two hits, no walks and punched out nine of 18 batters he faced, left the mound visibly upset and frustrated with the decision. He had just given up a routine single to Austin Barnes, who was hitting ninth for the home team.


Snell entered the game with 21-straight starts of fewer than 6.0 innings. But this wasn’t the Snell of the previous 21 starts. This was the Cy Young winner who was supposed to be the exemption to Tampa Bay’s analytical-based method of pulling starting pitchers before the dreaded third time through the order. Because that's when the best numbers warn managers of big danger ahead for starters. If it were possible to illuminate text with LED neon lights inside a binder, the third time through the order numbers would beam brightly and annoyingly so they would be nearly impossible to avoid.


Snell wasn’t just performing at a high level. He was as dominant as he’s ever been with an absurdly good whiff percentage of 47% and full command of his four-pitch arsenal, an important distinction for pitchers if they're asked to attack the same hitters the third time. Through his first 73 pitches, Snell threw fastballs harder than 97.0 MPH seven times. He threw just one heater harder than 97.0 MPH during the regular season. According to Baseball Reference, Snell had the highest game score all-time of any World Series starting pitcher who didn’t complete six innings. All-time!


The three guys set to come up for the Dodgers were Betts, Corey Seager and Justin Turner. The trio was 0-for-6 with six strikeouts against Snell, who showed no signs of fatigue. Staring toward the mound in joyous disbelief, a weight was lifted off the Dodgers when they watched Snell leave the mound in favor of Nick Anderson. Some LA players even chuckled about it when asked after the win.


Arguably, Anderson has been the best reliever in baseball the last two seasons. But he wasn’t the same guy in the postseason. The hard-throwing righthander, who relies on his four-seam fastball up and breaking ball down, was called upon despite giving up at least one earned run in five-straight appearances. Los Angeles already saw him twice and had more hits (four) than Anderson had strikeouts (three in 2.2 innings), an alarming sign for a pitcher who gave up just five hits in 16.1 regular-season innings.


Betts smacked a fastball up in the zone for a RBI double, a wild pitch led to another run, Betts then scored on a fielder's choice and one-run lead felt insurmountable for an anemic Tampa Bay offense basically relegated to whatever Randy Arozarena, who treated us all to a prime Barry Bonds postseason performance, could do. The Rays probably wouldn’t have won game six 1-0, and the real reason they lost the World Series was because they couldn’t hit. Not because Snell was robbed of his masterpiece.


However, they let the Dodgers, MLB’s version of the Buffalo Bills until last night’s final out, off the hook by removing one of the game’s best pitchers on the verge of perhaps his best start of his career. Again, Dodgers players were as close to an odd blend of puzzled and giddy as professional players could be about a pitching change. They were still in gleeful disbelief during the postgame.


The Rays entered the season with the third-lowest payroll in Major League Baseball. Their bullpen was decimated by injuries to key pieces. Yet they had the best record in the American League and made it to game six of the World Series. Make no mistake, the analytical approach they implement is an enormous reason why they made it as far as they did and will contend again next season. There isn’t an organization in baseball better at dissecting data and using it to guide all types of baseball decisions. Tampa Bay is consistently ahead of the curve and is forced to mine the data for any edge while other organizations enjoy larger margin of error thanks to much more expensive payrolls. The Rays are like the richest baseball Bitcoin organization.


Ironically, the Dodgers are kind of the Rays but with a lot of money. The original architect of the Tampa Bay analytical and scouting machine runs Los Angeles’s operations now. So, the team across the field in the other dugout is into the numbers many people are blasting today just as much as Tampa Bay. Need proof? Look at what the camera caught Betts checking out during the game:

Yes, those darn analytics were in the Dodgers dugout too during the game's defining moment. And let’s not forget Dodgers manager Dave Roberts stuck with starting pitchers – hello, Clayton Kershaw narrative – too long in previous World Series appearances. Just days earlier, Roberts pulled Kershaw despite the pitcher disagreeing and his third baseball colorfully voicing his displeasure on the mound. It worked. The Dodgers won.


So, this isn’t and shouldn’t be about analytics. Turning last night into such an argument is disingenuous, lazy and ignores how the Dodgers operate. It’s about feel and trusting what you’re seeing, and combining the two to guide a decision-making process. Heck, maybe it’s about Anderson entering and not another reliever. Yeah, we'll revisit that in a moment. The best organizations, and the Rays are certainly included in this, blend the numbers and the scouting eyeballs. Cash erred last night because he ignored his eyeballs and let the computer totally dictate his decision-making; he made a 162-game regular season process decision and not a win-or-go-home World Series decision. How good and valuable is a Cy Young winner if he doesn’t get a longer leash? If Snell isn't going to get any benefit of the doubt when he's dominating the league's best lineup like it's the Pirates, when will he.


Keep in mind Tyler Glasnow threw 102 pitches (!!!!) during Tampa Bay's game five loss, and he dreamed of being as effective as Snell was Tuesday night. Also, it's only fair to keep in mind this is all easy to say and write about in hindsight.


Sure, the numbers say Snell’s performance, like almost every starting pitcher, gets noticeably worse when he faces a lineup the third time; and Anderson is one of the best, most dominant relievers in the game. You’ll find the most diehard #teammath baseball analysts today pushing back on the idea Cash made the incorrect decision and will shout process > outcome. Hey, I’m on that team too because over the long haul process wins.


But to rely only on the numbers implies a standard projection for Anderson, the hitters, and ignores what the numbers were also saying about Snell’s complete dominance. They’ll also say Snell probably would have soon faltered and that a starting pitcher is dealing and dominant until he’s not. And they’re right. But they can’t tell you for sure when or if that would happen. That’s the in-game risk assessment and balancing act for a manager. And Cash didn't just choose to remove Snell. He chose to remove Snell and bring in Anderson.


The numbers don’t account for Anderson lacking confidence, feel, his usual elite stuff and failing to meet those projections the previous five times he was on the mound. Well, actually, maybe some do, but you get the point. Speaking of, Eno Sarris wrote a great, thought-provoking piece at The Athletic about the real mistake the Rays made - choosing Anderson. While I don't agree with all the points, like Snell dealing not mattering, I do agree Anderson wasn't the same guy Cash wanted him to be.


Nick Anderson's fastball velocity and vertical movement, via Brooks Baseball and The Athletic


The Excel sheets and algorithms don’t account for what is being seen in games. They don’t account for Snell, a rare type of top-of-the-rotation arm teams cherish because it can provide all those dominant innings, having his best combination of stuff and command in a very long time. You can find the splits and tweet the third time thru stats, but not every start is created equally; it's rare when Snell goes to the mound and every single little thing is working. Those numbers tell us what an expected Snell outcome would look like based on his past. Last night, the lefty was in the very top percentile of his personal dominance; he wasn't sitting in the middle with the mean projection, and the numbers don't tell us what is projected to happen when he's in that tier. Eyeballs on the bench are fixated on his every move and every pitch to notice the decline. There was no decline. Not yet, at least.


This was a fascinating thread on Twitter today from Six Man Rotation founder Connor Kurcon. He explored similar starts to what Snell did last night.

Here's the argument: it's impossible for us to look at box scores and similar statistical starts and put them in a grouping with Snell last night.


  • What were those pitchers' pitch counts?

  • Did they have their best stuff, above average stuff, or decent stuff?

  • Were they facing a bad lineup, good lineup, or great lineup?

  • How often did they have the platoon advantage?

  • How many pitches did they throw last start?

  • What were the weather conditions?

  • What time of year was it?

  • How many pitchers didn't give up runs or only gave up a run?

  • What were the numbers for pitchers who gave up two or less hits and zero runs because that's what Snell left with?

  • Did they have great command, good command, average or worse command?

  • Were they carrying themselves with confidence or was the opposing team starting to see them better? Did all their strikeouts come really early?

  • How many hard-hit balls did they give up?

  • Were the predominantly two-pitch, three-pitch pitchers or could they attack with a full arsenal?


The list of questions, many of which can only be answered by the scouting/eyes, goes on. I'm not pushing back on the risk associated with pushing a pitcher through the lineup a third time because that's absolutely real. But we simply can't assume Snell's start can be lumped into a long list of other good pitchers and those results will predict his because we don't truly know how similar those starts are. Each start is its own baseball snowflake. "All of these pitchers were just as good (or better) as Blake Snell and pitching just as well (or better)," is an unfair statement based totally on box score results.


The numbers don’t account for the Dodgers looking totally overmatched. They don’t account for things like pitchers tipping pitches and hitters with slight injuries, new swing and timing issues, fatigue or poor mental approach. If things like confidence and mental approach didn’t matter, teams wouldn’t obsess over makeup reports for prospects. Trust me. I’ve filled out some MLB scouting reports. There is an enormous section for makeup.


And while some of the game’s most popular analysts imply the mental competitive side is always linear, the guys playing beg to differ. They flat out told us last night they didn't think they could hit Snell and were relieved when he was removed. And if teams didn't care about scouting and the stats played out on the field, pro scouts wouldn't spend countless hours hunting any little piece of human error or tendency that would gain an edge. Sure, the momentum element of game analysis is overblown. Unless you ask an actual player.


The analytics simply provide the best information about what’s projected to happen based on past results and predictive information. In some ways, it’s baseball’s version of playing Blackjack by the book. Tampa Bay has the book memorized and would almost never deviate. Many teams read the exact same book and could have interpreted the data differently or simply chose to go against it. That's why this isn't and shouldn't be a war against analytics. Tampa Bay should have been drunk on Snell’s dominance and thrown the book to the side at least until an actual sign of danger on the field and not in the predictive data. The Rays had the type of hot streak everyone dreams of, and they got up and left for another table before they knew it ended.

These photos are not mine. Thanks to MLB.com, Boston Herald, AP, and Angels.

©2019 by Mike Trout's WAR. Proudly created with Wix.com