Analytics for the win
How the losingest team in baseball found a competitive edge with analytics and rode it four years later all the way to the World Series
Pitcher Charlie Morton was on the mound the moment the Houston Astros clinched their first-ever World Series title with a 5-1 victory over the LA Dodgers in Game 7. It was something of a poetic ending to the season in which both team and player rose from the ashes of back-to-back losing streaks.
Following a 2016 season with only four starts, a mediocre ERA and a series of injuries that had him on the bench, the Philadelphia Phillies cut Morton loose. At 32, it looked like the death knell of his MLB career. But when the Astros’ front office noticed a velocity jump in Morton’s pitching, they signed him to a two-year, $14 million contract. One year later, he closed out Game 7 of the World Series.
The potential the Astros saw in Morton – and how the team climbed their way from the worst record in baseball to winning the championship – was based largely on insights derived from novel statistical analysis techniques.
Creating decision tools for decision makers
Though many in 2014 saw the Sports Illustrated projection that the last-place Houston Astros would win the 2017 World Series as further than a long shot, the Astros’ front office was confident in what was at the time an unconventional strategy. That strategy hinged on the work of former NASA engineer turned sabermetrician Sig Mejdal, who first walked into the offices at Minute Maid Park in 2012.
If you’re going to try to build a championship team from scratch, “you want to figure out how to take advantage of what’s discoverable and acquirable,” says Mejdal, then the Astros’ Director of Decision Sciences. Now Special Assistant to the GM, Mejdal describes his remit as improving the processes by which the Astros’ decision makers – scouts, coaches, even players – apply the findings of the team’s analytics group. Often in exploring the wealth of data the team has now amassed, analysts can see potential in a player that traditional scouting may have missed or share coachable insights with decision makers in the bullpen.
“Analytics permeates the entire [Astros] organization,” Mejdal says, “the decision to draft a particular player, the instructions he gets in player development, the decision to promote him, how he attacks a specific pitcher, where he stands on defense – analytics is a part of all of that. It's impossible to separate analytics from what we do.”
Analytics, no longer a novelty, is now ‘table stakes’ in baseball
Baseball has always been a game in which stats mattered. But Mejdal says that as the means of collecting information have gotten better and better, data has become truly ubiquitous. For example, new technologies like radar devices can now measure the release point, speed, movement, spin and location of every pitch – and record the same for batted balls. When combined with traditional scoring methods, these technologies provide teams with a vast number of data points. “We can get information on a center fielder's reaction time, his ability to accelerate, the efficiency of the path he takes to track down the ball, his top speed, and of course, whether he made the out or not,” Mejdal explains.
In fact, every major league team in America now has at least one quantitative analyst – someone like Mejdal working behind the scenes to interrogate the numbers. Whereas in the beginning of his career in sabermetrics, nuggets of statistical insight in any amount were enough to give an analytically savvy team the edge over a data-naïve competitor, quantitative analytics in baseball today is just “table stakes,” Mejdal explains. “When there were only four analysts in baseball, it was easy to stay ahead of the curve – the inefficiencies were giant. But now, like in any mature industry, the inefficiencies become smaller and harder to come by,” he says. Take all this readily available data and throw in capable sabermetrics programs at every front office around the country; what you get is as competitive a playing field as baseball has always been. But that doesn’t mean there’s not a competitive edge to be found.
The Astros gain a competitive edge through teamwork
When analytics is an arms race, as Mejdal describes it, the competitive advantage comes increasingly less from data access – or even the kinds of statistical insights you can glean from the data. Rather, teams that are most successful in their analytics programs are those that, like the Astros, have built a culture where individuals across all parts of the organization work together to take advantage of the data they’re given.
“I don’t think what we are discovering is too different from what others have already or soon will discover,” Mejdal says. “Where I think we (the Astros) do very well is in taking advantage of the findings. From the GM to the manager to the coaches to the scouts to the players – they all are embracing analytics and the efforts to do something a bit better. It’s incessant.”
This harmony within the organization is what makes the Astros great. For one, the growth of decision sciences in the Astros’ front office hasn’t detracted from the team’s reliance on expert valuations and the opinions of scouts and coaches; in other words, the art of baseball has not been eclipsed by the science. And coaches and players have been eager to experiment with the insights they receive from analysts.
When the scientific method is the strategy, JMP® takes you from start to finish
More astute analytics programs like the Astros’ front office understand that data insights go both ways; this is a key part of the culture General Manager Jeff Luhnow has fostered. Analysts listen to what the experts intuit and then help test the value of certain attributes. “Analysts have come into baseball, and what they’re bringing in is a process to test hypotheses in a structured way,” Mejdal explains; the scientific method is the strategy. And sabermetricians are positioned to apply that method – with the help of new technology and statistical tools – to investigate the merits of conventional baseball wisdom.
Having come into sports analytics in the field’s early days, Mejdal had to build that toolkit for himself. “At the time, we had R and XL Stat,” Mejdal recalls. So it came as a surprise when Mejdal got a call one day inviting him to speak at a conference hosted by statistical software company JMP. “I had not used JMP, I did not own JMP, and then when I got to the conference I could see the passion these users had with this software program and I couldn't believe it. That got me curious to try JMP, because frankly I didn't care for another off-the-shelf software program that I needed to learn. But then when I took a look I was just wowed with the product…. That combination of intuitiveness, capability and speed is hard to find.”
Interactive data visualization in JMP® facilitates important communications
Data visualization helps the end user too; in baseball, the front office’s data insights are only as useful as the players and coaches make them. “We need to connect to the [on-field end users]. We’re asking them to some degree to change their behavior. In my experience, that doesn’t just come with saying ‘Hey, I did some analysis. Here’s a number. Use it.’ Instead, the more you illustrate what’s behind [the stats], connect an anecdote to it, the better off you’ll be.”
As they say, a picture is worth a thousand words; similarly, graphics help on-field experts understand and operationalize data insights. “The more you can remove [coaches and players] from the black box of analysis and toward an understanding – a visualization of what’s going on – the better. And JMP is great at creating intuitive illustrations of your data quickly,” Mejdal explains.
That the Astros would use a tool whose key value lies in its exploratory potential is no coincidence. Mejdal says that same curiosity is at the heart of Luhnow’s vision for the team: to drive, question, improve and then take advantage of any insights they can glean. “JMP is an excellent tool with which to determine these relationships and predictive ability,” Mejdal says. “Rarely does a day go by in which I don’t use JMP.”
A prophetic ending
Fast forward to Game 5 of the 2017 World Series. “It was the loudest crowd I have ever heard in Minute Maid park,” Mejdal recalls. The pandemonium was set off by Astros third baseman Alex Bregman who, late into the game’s fifth hour, put the torturous 13-12 game to an end with a walk-off single. Bregman was seen as a highly unconventional pick when the Astros drafted him in 2015. Mejdal says that at the time, the team’s analysts noticed a number of attributes pointing to his potential success. That insight combined with scouts’ significant interest in the player was enough to draft Bregman out of LSU. So, when two years later he became only the second player to drive in a run in each of his first five World Series games, the whole affair had something of a prescient shine to it. “It's wonderfully satisfying to have played a role in acquiring a player and then seeing him years later succeed,” Mejdal says.
Even more satisfying, the Astros finished off the World Series with an infield shift – a move designed to defend against batters with a penchant for pulling hard right hits through the gaps between players. Seeing the team’s infield move into this now ubiquitous defensive position, Mejdal took a moment to reflect on just how far they’d come. In earlier years, the Astros had been the first to take the shift to a level that had seemed almost laughable at the time. “I thought it was appropriate that the last out in Game 7 of the World Series was into the shift,” says Mejdal. Yet further proof that analytics programs – at least in baseball – are here to stay.