Analytics can’t predict unpredictable playoffs

Despite more data, even the best baseball managers’ decisions tend to be hit or miss

Los Angeles pitcher Yu Darvish had an outing he would like to forget Friday, giving up six hits and four runs and recording just four outs before getting the yank in the second inning. If you were to survey a group of baseball people and come up with a list of the best active managers, names such as Bruce Bochy, Terry Francona and Joe Maddon would be at or around the top of it. They’ve each had a lot of regular-season success, they have proven they can adapt to the modern game, and they have World Series titles on their resumes.

And yet, Francona has now lost six straight post-season games in which his Cleveland team had a chance to close out a series.

Maddon took his World Series champion Chicago Cubs and piloted them straight over an NLCS cliff, and then either fired his coaching staff or watched as they were fired, depending on whom you believe.

And Bochy, as bulletproof as they come with three championships this decade, won 64 games this season with the Giants. He has now been a major-league manager for 23 seasons and has won 1,853 of his 3,708 games. That is a winning percentage of .4997, which rounds to an even .500.

That last statistic seems particularly appropriate. Just like flipping a coin, manage enough baseball games and the outcome will eventually become 50-50.

Some names no longer on the list of best active managers include John Farrell, Dusty Baker and Joe Girardi, because all in recent days have been fired. Farrell won a World Series in Boston in 2013 but went 1-6 in his last two playoffs, and that was enough. Baker, despite getting to Game 5 in the NLDS twice with the Washington Nationals, and winning 95 and 97 games on the way to it, was also dismissed.

Even more puzzling was the case of Girardi, who won a World Series in 2009 with the Yankees, kept the team from cratering as it rebuilt, then guided a young team to the precipice of the World Series far sooner than anyone expected. But they couldn’t hold a 3-2 ALCS lead against Houston, so, see ya.

If these are cases of front offices replacing their managers because of post-season failures, that is nuts. Was it Girardi’s fault the Astros traded for Justin Verlander in late August, and he somehow went from fading glory to throwing thunderbolts and lightning the inexperienced Yankees couldn’t hit? Was Baker really to blame for asking Max Scherzer, the likely Cy Young winner, to nurse home a Game 5 lead he couldn’t hold? Good decisions don’t always work out, whether in April or October.

There is also a theory the latest spate of firings is just the latest iteration of Moneyball. First the data revolution altered front offices, then scouting staffs, and now it’s finding its way into the dugout, with managers who speak the language of an analytically inclined baseball operations staff. Sports Illustrated reported on Thursday this was the thinking behind Brian Cashman’s cashiering of Girardi. He didn’t just want someone who was merely accepting of data, but apparently someone who believed it in his soul.

But while it only makes sense to have a manager who is pulling in the same direction as a front office — there is no point in using objective data to fill out a roster, then turning it over to an old coot who is all about the eye test, and heart, and gut instinct — those in baseball ops should also remember, even the most analytically inclined manager is only fiddling about in the margins of wins and losses.

Modern baseball, from roster construction to in-game management, is about increasing the probabilities you will score runs while at the plate and prevent them from being scored the rest of the time.

The most cleverly deployed managerial strategies might bring a few extra wins over the course of a 162-game season, but there’s so much randomness in baseball — hall of famers get hits three of 10 times and all that — that good and bad decisions end up as just statistical noise.

And in the playoffs, outcomes are even less predictable. Maddon has long been considered the gold standard of managerial nerds, but in the NLCS, he brought John Lackey in to pitch in the ninth inning of a tie game. Lackey had given up more home runs than anyone in the National League. He gave up a walk-off home run.

Francona has been deservedly lauded for his new-school approach to using Andrew Miller whenever big outs are needed, but the ALDS turned when Miller gave up a home run to New York’s Greg Bird — who was injured most of the season and hit .190 when he returned. It was the only run in Game 3.

In the World Series, Houston’s A.J. Hinch and the Dodgers’ Dave Roberts are both relatively young managers who have fully bought into the big-data era. All they did in Game 2 on Wednesday night was make pitching change after pitching change that blew up in their faces.

For all of the ways in which analytics have changed baseball, even the manager who has best arranged the probabilities can’t escape one fact: In the playoffs, the unlikely often happens.