Blog

When right is wrong, or why the nerds yelled too

On Tuesday night, Major League Baseball’s 2020 season came to a conclusion – but not without controversy. Actually, a couple of controversies – but I’m not going to devote much time explaining to you that letting players play while getting an inconclusive COVID-19 test reviewed is bad, and letting the player celebrate with his team after he’s tested positive and you’ve put him in isolation is worse. No, we’ll stick to to the strategical, on-field decision that many felt cost the Tampa Bay Rays a shot at coming back in the World Series, a decision that proved once and for all that..

ANALYTICS HAS GONE TOO FAR!

Sorry, primal instinct there. Anyway, Rays manager Kevin Cash chose to pull his starting pitcher Blake Snell from the game after giving up a single in the sixth inning. Not a run, just a single, and just his second hit conceded of the game. He wasn’t just playing well, but in the midst of one of the all-time great World Series Elimination Game performances.

Tampa Bay are a very process-oriented team, though, and bet on the known fact that pitchers tend to decline as they get deep into the pitch count, and that their odds of fooling their opponents on their third lap at the plate are a fair bit lower than their first two. Following their process to the letter got the team this far, so who would he be to question it now?

“The only motive was that the lineup the Dodgers feature is as potent as any in the league,” Cash said to reporters after the game. “Personally, I felt Blake had done his job and then some. Mookie (Betts) coming around the third time. I totally value and respect the questions that come with it.”

“I guess I regret it because it didn’t work out. The thought process was right. If we had to do it over again, I would have the utmost confidence in Nick Anderson to get through that inning.”

In fact, one could probably argue that the decision didn’t even really backfire as much as it felt like it did. At the end of the day, the Rays defence gave up five hits and three runs to the best team in baseball – they exceeded the overall expectation and did their job, and it only feels otherwise due to the fact that two of the conceded runs came right off the hop, and their offence only delivered a single run and five hits of their own. Had the Rays got their typical 4-5 runs, we are very likely watching a Game 7 some time in December or January (because of, you know, the super spreader event on the field).

So, if the decision was sound in its nature, and the results weren’t as disastrous on paper as they felt in our hearts – why has the backlash gone beyond the typical “analytics bad” casual and old school crowds? Why were even your nerdiest friends in shock, dismissing the move as absurd?

When situations like this happen, where a team makes a controversial bet based off of their data process and it doesn’t lead to a winning result, we tend to hear a lot about how analytics are ruining the sport. This is true no matter what sport we’re talking about – hence why I’m writing about something that happened in the World Series on my personal hockey blog right now. If you make a bet based on data, and the overall result doesn’t work out, the data will always get blamed, sometimes regardless of whether or not the actual bet over or under performed.

Much of this is rooted in the emotion of it all. Take this situation with Snell – millions of people tuned in to see a former Cy Young (Pitcher of the Year, for non-baseball fans) winner dominate in the most important game of his life, only to get pulled out midway through – not because he fell apart, but because his team didn’t want to take the risk of him falling apart. In previous generations of the game, the bet would have instead been placed on “riding your horses”, or in other words, letting your best players play until they truly give you an excuse to pull them out. Is that the safest bet? Not necessarily, which is what we’ve learned over the years thanks to the data. But it’s the one that feels right.

That “feel” is the real crossroads, that doesn’t get spoken of correctly when it comes to the analytics debate. Too often, people in denial of the value of probability and projection will justify their reliance on gut feelings and intangibles by saying that the data misses them, while their brains truly know how to win. These people are very good at pitching themselves as a benevolent product, despite the fact that most of them come out of most of their seasons empty-handed – these days, even more so than those who trust their processes. Say what you will about the Rays blowing it at the end of this game, but this is the same process that got them to Game 6 of the World Series with the sixth-lowest payroll in the majors.

Instead of focusing this debate on whether or not the data works – we know that the answer is “generally yes and more so than the hunch” – we should be focusing on whether the data is contributing to the product, and whether or not the data is causing the problem, or is merely a tool of the problem. The problem in the case of the Snell pull wasn’t that available evidence disagreed with it – it’s that no one, except perhaps his opponents, wanted it. Snell didn’t want it. The fans didn’t want it. His teammates didn’t want it. Cash, in his heart, probably didn’t want it.

People wanted to see the elite pitcher pitch his heart out. People don’t tune into professional sports to watch a model play out, they tune in for the stories, for the big plays, for the big moments. Players don’t become professional athletes to be cogs in a machine – becoming a professional is a reward for years of amateur training and competition, and while the money and fame is nice, the main prize for these players is the chance to get to the top of the mountain they’ve dreamed of since they were children. When we talk about analytics taking the “emotion” out of the game, it’s too often brought up as a competitive error, and not often enough brought up as stifling the fun and passion that the game brings to its players, its staff, and it’s fans.

That’s not something that the analytics revolution brought to any of these sports, either. Data doesn’t inherently demand that you make the most risk-adverse decision every time. It doesn’t demand that you can’t ever trust your intuition. It doesn’t suggest that it will work every time, or even significantly more often than the alternative. Data simply gives you more information to take into account when you make your own decisions, and it’s up to you where you go with it.

What the current depths of information do, however, is enable micro-managers, and that’s where the real problem is. It enables those who will look for every single edge to win to sacrifice so much of what we enjoy about the game in order to get there. They have more situations to put their managerial fingerprints on, and gives them something to deflect blame towards if it doesn’t work. It allows for a world where you can make educated guesses on questions no one really cared to ask, and that’s the problem at hand.

It’s also not uniquely an analytics problem, though. We’ve seen similar micro-managing when it comes to sport science, we’ve seen it when it comes to contract talks, and for decades, we’ve even seen it in traditional systems. Take the “dead puck” era in hockey, for example – we didn’t need Corsi, Expected Goals, and RAPM models to end up at the neutral zone trap, or other strategies that exist to slow down the flow of play and protect leads. We’ve talked about “overcoaching” for decades as something that detracts from the entertainment factor of games.

Analytics didn’t create this problem. In many cases, listening more to analytics could even reverse some of it. Most of hockey’s general analytical concepts encourage a higher paced game with more scoring opportunities, and requires rosters to have more skill and talent than ever before to execute plays that are often less complicated, but also separate the elite from the mediocre in a more obvious way. Football’s biggest analytical trope is “go for it on fourth down instead of punting”, which is a thousand times more exciting than the old method. Baseball encourages home runs more than ever. Soccer asks for its players to move faster, Basketball encourages more technically impressive shots than it used to.

A significant chunk of the information we’ve picked up over the years, across all team sports, comes back to a general concept of “have your best players try to score, always”. It’s confirmed that tropes like “the best defence is a good offence” and “put pucks on net and good things happen” are the ones worth keeping around. It’s confirmed that game breaking talent matters a lot. It’s encouraged teams to replace players along for the ride with players who can add more impactful performance.

Data-driven sports management doesn’t have to be boring, and it doesn’t have to be timid – it’s those things when the people who use them are boring and timid. The growth of professional sports leagues into multi-billion dollar monstrosities, who operate as corporations even more so than they operate as teams, is much more to blame for analytics “ruining your sport” than the analytics themselves. When a mistake in a game isn’t just about a win or loss, but about shareholder value, public relations, and job security in a volatile, short-term career, you end up with more and more decision makers trying to go the safe route at just about every turn.

That is how you end up with a situation like the one that Blake Snell, Kevin Cash, and the Tampa Bay Rays faced last night. What we saw was a fear of betting on the less probable and losing out leading to a decision that no one wanted. It wasn’t a play to win, but a play to not lose. It was a glance at a story that could’ve gone right in a way we’d be talking about for decades, or wrong in a way that would be mocked by the probabilities, but respected by the masses. Because the latter seemed more likely, they decided to not find out at all. That may very well have been the best decision for their chances of winning the baseball game, even though they ended up losing. But it wasn’t the right one for the sake of the competition, or the entertainment product they’re selling.

Analytics, strategy, intuition, and blind luck can all co-exist in sport. In fact, one could easily argue that a combination of the four is going to lead to the best performances, most intense and rewarding competition, and most entertainment value. But so long as the stakes are so high, micro-management and risk aversion even in the slightest margins will continue to exist, and it is that which spoiled the moment last night. That is why the nerds yelled just as loud as the jocks – as much as we trust our processes, the process is worth nothing if it means the games lose their fun factor. Last night, the fun factor was squashed on a very, very big stage, and because of that, the blow-back will end up even greater than had we all been given what we wanted to see.

 

You might be interested in …

Leave a Reply

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