Do Analytics Add Up To Wins?
The Efficacy of Playing by the Numbers
It was a cold chilly October night at Chavez Ravine, the crowd buzzing with excitement. The Dodgers were hosting the Red Sox in a pivotal game four of the World Series. If the Dodgers could win this game then they would tie up the series and have all the momentum going for them for the next three games.
The starting pitchers that game were Rich Hill and Eduardo Rodriguez and for the first five innings they were both lights out, not giving up a run for either side. Then in the bottom of the sixth the Dodgers started to rally and Yasiel Puig cracked a mammoth three-run home run to left field to give the Dodgers a 4–0 lead. Fans waved their blue towels in unison to create a beautiful sea of blue around the stadium. It seemed like a sure thing that the Dodgers would record these final nine outs and even up the series.
Then, came the top of the seventh. Rich Hill was dominating the game, showing no signs of faultering and yet every Dodger fan was looking down in the bullpen — afraid of the inevitable, that the hook was coming for Hill.
A few years ago, no one would be warming up down with a starter pitching so well. But Dodgers’ manager Dave Roberts usually doesn’t let Rich Hill go more than 100 pitches or face the opposing lineup a third time, whichever comes first. When the seventh inning started, Rich Hill had only thrown 82 pitches. He walked the lead-off man on six pitches and followed that with a strikeout. With the win seeming to be in no danger, Dave Roberts still came out of the dugout to pull Hill. Though the move was consistent with the Dodgers’ data-driven managing style, most fans were surprised because Hill still seemed unhittable. The Dodgers’ bullpen blew the lead and the game and few doubt the loss sealed the Dodgers’ fate as they went on to lose the World Series for the second year in a row.
The million dollar question remains: did it have to be this way? Why take out a dominating pitcher, Rich Hill in this case, when he’s showing no signs of losing his mojo. The answer is analytics.
“Roberts is a [bleeping] dumbass,” says my neighbor, a diehard Dodgers fan. “It was a very bad decision because instead of rolling with the hot hand, Roberts felt analytics and numbers were the right way to go in that situation and clearly it was not.”
Analytics is a growing and controversial trend in baseball. A few years ago when analytics and data were introduced to baseball, it changed the game radically. Managers stopped relying on their gut and their eyes and started relying on numbers. That means limiting pitchers to predetermined pitch counts regardless of how they seem to be flowing and pulling pitchers for favorable lefty-versus-righty matches even if the starting pitcher is dominating. Even the best pitchers in the game rarely pitch a complete-game victory, once considered the ultimate achievement for a pitcher outside of a no-hitter. It all adds up to increased use of the bullpen.
For hundreds of years pitch count was secondary to how a pitcher was actually performing. It was about results. If you were rolling, you stayed in. If not, you were given the hook. This was how the game was played until analytics took over.
What has changed from then to now? Pitchers arms are relatively the same if not better than they were 50 years ago, so why can’t pitchers in today’s game throw as many pitches as the older guys?
Are analytics the new bible for baseball? Perhaps it depends on how you use them.
They probably can, but the data game management now relies on shows that it is better for the team if a pitcher gets pulled sooner rather than later. By crunching data, teams know what the numbers say, what the percentages are in any given matchup at any point in a game. The numbers tell them what pitch count to go to, when to move to a lefty-versus-righty matchup, how to shift the infield against certain batters. It’s all about what the numbers tell managers is the best opportunity, percentage-wise, for success in a game situation.
Analytics also tell teams that it is more efficient to score runs with one swing of the bat rather than two or three. It suggests that runs can be scored more often if hitters swing for the fences by using “launch angle” for their swings. The idea is that one hitter can score a run by himself as opposed to three hitters getting a single and bunting them over to score a run. That’s why the Dodgers’s batters seem to strike out so much. They have been managed to swing for the fences. They are not the only ones: the amount of home runs being hit in Major League Baseball is at an all time high. So is the amount of strikeouts.
All 30 MLB teams have adopted this “money-ball mindset” to an extent. Some rely on it more than others but all the teams believe that playing by the numbers helps them be more successful. Are analytics the new bible for baseball? Perhaps it depends on how you use them.
Numbers can be used in two different ways. Some teams allow their managers to use analytics as a guide and not a rule. The manager will have a little card in the dugout with him and look to it when he feels he needs to do so, sort of like an advisor with data to back up his advice. But the ultimate decision still comes down to the manager
The past two World Series champs have used analytics in this way. They crunch the numbers, but give ultimate decision-making to the manager, allowing room for what his gut and his eyes tell him. They call this the human factor. For example, analytics would have told the Astros manager, A.J. Hinch that it was best to take out Charlie Morton after his fist turn through the order in Game 7 of the 2017 World Series. But Hinch saw that Charlie was dominating and he felt that he was their best guy to finish the game for them. His decision turned out to be the right one. But other teams use analytics and rely on it to be there decision maker.
Other teams don’t give their managers that much leeway. For example, the Dodgers and Brewers stick almost religiously to analytics. For the Dodgers, analytics tell Dave Roberts when to pull a pitcher, when to shift his infield, and what hitter should be hitting that day. These analytics tell Roberts the numbers and what is likely to happen during that situation, and he plays the percentages almost 100 percent of the time.
The fact that the religiously analytical Dodgers have lost the past two World Series have led to questions about whether the numbers game, even if it’s likely to work over a large sample size, is the best one to play in short playoff series. We have also this with the Oakland A’s who perform well over the course of the season but are not successful in five-to-seven-game long series. Across the bridge from the As, the San Francisco Giants used very little analytics during their recent run of winning three World Series in six years. More recently, the Astros and Red Sox did not rely exclusively on analytics, but allowed for a combination of playing the percentages and letting the manger use his eyes and experience. The Brewers, another religiously analytic team, lost to the Dodger’s in the playoffs last year.
While the jury is still out, recent history indicates that teams that rely too heavily on analytics in October end up losing and teams who use it as more of a guideline and reference come out on top.
For me, its all about how a pitcher looks out there.
One interesting point is that analytics and pitch count haven’t yet taken over high school or college baseball. My old high school pitching coach said, “For me, its all about how a pitcher looks out there. If he is still going strong and doing well then we’ll keep him out there, but if he’s starting to tire then we’ll get a fresh arm in there. We tell our pitchers to go as hard as they can for as long as they can to try and get the most out of them.”
This is the way baseball has been played for years and, so far, is still being played at the high school and college level.
