Houston's Grand Experiment
In the late 1980s, when people got too drunk and were kicked out of the other casinos in Lake Tahoe, they ended up at High Sierra, a place where there was no such thing as being too drunk. Sometimes they staggered over to a blackjack table manned by a young dealer named Sig Mejdal.
Mejdal was an undergraduate at UC Davis, studying mechanical engineering and aeronautical engineering. During the summers he'd head 120 miles west, clip an oversized bow tie to his collar—"Looked like a dead cat around your neck," he says—and sling cards at Tahoe's seediest betting house. He loved the job. It was fun, it was social, and he learned things that he could not back in the lab at Davis. He learned that human beings do not always make decisions that serve their own long-term self-interest, even when they are equipped with a wealth of experience and knowledge of the mathematical probabilities that ought to guide their choices.
Blackjack is a probabilistic game. For any combination of cards, the player's and the dealer's, there is an optimal action for the player to take to increase his chances of winning—or, as is generally the case, of losing less. Sometimes the course of action is obvious: You hit a 10 no matter what the dealer is holding. Often, though, players know what they ought to do—but they do something else because their intuition has told them to. "Hitting a 16 against a dealer's seven, it doesn't feel right," Mejdal says. "With a hundred-dollar bet, it feels even less right. But that doesn't mean it isn't right."
Sometimes players would ask other dealers what they ought to do with a difficult hand. The dealer would, without meaning to, offer the wrong advice. "This person sees a million hands a year, with immediate feedback," Mejdal says. "I thought that illustrated well the limitations of human capabilities"
Mejdal, who is now 48 and married with a stepson, would go on to earn two master's degrees from San Jose State, in operations research and cognitive psychology. He would perform research for NASA in which, essentially, he disproved the perceived utility of napping. All along, though, Mejdal's mathematically driven career had stemmed from his passion for the most mathematically driven of sports: baseball. In 2003, when he was 37, he read Michael Lewis's Moneyball, and he realized that there might be a place in the game for someone like him.
Soon Mejdal was sending out résumés and proposals in an attempt to land his dream job. He traveled to the 2003 winter meetings, in New Orleans, hoping to get a general manager's attention. Finally, in '04, one of his pitches caught the eye of a baseball executive whose CV was almost as unusual as his: Jeff Luhnow, who had joined the front office of the Cardinals the year before.
Like Mejdal, Luhnow had two undergraduate degrees (in chemical engineering and economics, from Penn), as well as a master's (an M.B.A. from Northwestern) and a varied professional career. He had designed suits intended to protect troops from nuclear, biological and chemical warfare. He had helped start an Internet business, PetStore.com, and another that produced customized apparel on a large scale. He had also spent five years as a management consultant at McKinsey & Company, and he believed that one engagement (in the company's parlance) he'd worked on there had prepared him for his job in baseball more than any other. The project involved advising one of the world's largest casino operators.
"I learned a lot about how the gaming industry works, and about probabilities," says Luhnow, a trim 48-year-old with neat gray hair. "How if you have a large number of occurrences, even though luck is involved, you can still make things pretty predictable. For the player, when you do start to follow your gut, or you've had a couple drinks and think you've seen a lot of 10s, you're just basically giving the house back some money. The odds are the odds."
Luhnow hired Mejdal to run the franchise's new analytics department in 2005, around the same time that Luhnow was elevated to director of amateur scouting. Over the next seven seasons the Cardinals would draft more players who became big leaguers than any other organization. Of the 25 players on the team's World Series roster last October, 16 were drafted under Luhnow's watch. But he was not in St. Louis to see the Series, because in December '11 the new owner of the Astros, Jim Crane, hired him to be Houston's general manager.
Luhnow brought Mejdal aboard to be his director of decision sciences. The new director of amateur scouting was Mike Elias, a 31-year-old Yale graduate who had worked in the Cardinals' scouting department. The new assistant GM would be David Stearns, a 29-year-old Harvard graduate who had most recently worked for the Indians. The new director of pro scouting would be Kevin Goldstein, who had been a respected writer for Baseball Prospectus but had never worked in pro baseball.
The job facing Luhnow was different from the one he'd faced in St. Louis. There he had to keep a healthy organization healthy. In Houston he was asked to figure out how to defibrillate a club that was dying.
The Astros reached the World Series in 2005, but in '10 they finished at least 10 games below .500 for the third time in four years. In '11 they went 56--106, their worst record to that point. Their club's longtime core was gone—Lance Berkman and Roy Oswalt had been traded; Jeff Bagwell and Craig Biggio had retired. Worse, the farm promised no quick replenishments. Before the 2010 season, Baseball America ranked Houston's minor league system as the game's worst.
Luhnow and his men envisioned a decision tree, with that 56-win team at its roots and a sustainable championship club at its tip. Their only goal, with Crane's blessing, was to reach the top as quickly as they could. That meant every decision they made, no matter how painful, would be based upon the probability that it would be helpful in the long term. They would, in other words, hit on 16 against a seven every time. "We didn't want to be mediocre for a decade," says Elias. "We wanted to be really good as soon as possible."
They would not make cosmetic decisions, such as wasting money on a free agent or hanging on to a veteran who might instead be converted into future assets, in an effort to keep up appearances. This was partly a financial decision: When Crane bought the team from Drayton McLane, it was running, sources say, an annual deficit in the tens of millions. Crane was not driven to spend more than necessary while the team was losing—a period he planned would last no longer than a few years.
"You look at how other organizations have done it, they've tried to maintain a .500 level as they prepare to be good in the future," says Luhnow. "That path is probably necessary in some markets. But it takes 10 years. Our fans have already been on this decline, from 2006 to 2011. It's not like we're starting fresh.
"Would it be the right strategy for somebody else who had a great farm system and up-and-coming players already at the big league level? No. But for us, it was. When you're in 2017, you don't really care that much about whether you lost 98 or 107 in 2012. You care about how close we are to winning a championship in 2017."
It is one thing to commit to only making decisions that will lead to a long-term goal, and another to figure out how to make those decisions. Blackjack is an exercise in hard probabilities. Evaluating baseball players is something else. Some information you can gather about a baseball player is hard: how fast he can throw a fastball, how quickly he can reach first base. But much of it is soft: how diligently he will work, how his power stroke might develop, how likely he is to become injured. "How do you combine the soft information with the hard information in a way that allows you to make the best decisions?" asks Luhnow. "That is the crux of what we're trying to do here."
They are trying to do it in a way that synthesizes quantitative and qualitative information about players. This represents an evolution from the processes that Billy Beane's A's used a decade ago—at least as they were described in Moneyball. "For all the wonders that the book did, the portrayal was a dichotomous one," says Mejdal. "It's either the scouts or the nerd in the corner of the room. But from the very beginning in St. Louis, Jeff framed it as an and question. The question was not which one to use, but how to combine them." The goal is to use all that information to produce a metric that will render a decision on a player as simple as the one in blackjack: hit or stay.
To that end Mejdal and his analytics team—which has grown to four and occupies a room in the Astros' offices that they have named the Nerd Cave and decorated with a Photoshopped image of scientists examining Vladimir Guerrero in mid-swing—created an evaluation system that boils down every piece of information the Astros have about prospects, and about every player for that matter, into a single language. The inputs include not only statistics but also information—much of it collected and evaluated by scouts—about a player's health and family history, his pitching mechanics or the shape of his swing, his personality. The system then runs regressions against a database that stretches back to at least 1997, when statistics for college players had just begun to be digitized. If scouts perceived past players to possess attributes similar to a current prospect, how did that prospect turn out? If a young pitcher's trunk rotates a bit earlier than is ideal, how likely were past pitchers with similar motions to get hurt?
The end result is expressed as a numerical projection which roughly translates into how many runs a player can be expected to produce compared with what the team is likely to have to pay him—a single value partly derived from a player's stats but mostly from scouting reports. "They're not asking us to be sabermetricians," says Ralph Bratton, a Texan with a thick white mustache who has spent a quarter century as an Astros scout. "They're asking us to do what we've always done." The twist is that Luhnow's front office processes that information differently and makes decisions largely based on the result—even when that result, like a directive to hit a 16, feels wrong.
The Astros' decisions since the end of 2011 seem to have genuine promise. The farm system is now ranked among the game's best. The major league team, buoyed by recent promotions of top prospects like outfielder George Springer (who energized the club with both his constant dance moves and his 13 home runs in his first 58 games) and first baseman Jon Singleton, went 15--14 in May, its first winning month since September 2010. Springer and Singleton have complemented holdovers like diminutive second baseman Jose Altuve (who is batting .336 with an AL-leading 26 steals), but more impressive has been the improvement of the young staff. Since May 1 the Astros have an ERA of 3.75, the league's sixth lowest, behind suddenly maturing starters like 26-year-old Dallas Keuchel (2.45), 24-year-old Jarred Cosart (2.84) and 24-year-old Brett Oberholtzer (3.32).
The progress made in the last few years, however, has come at a cost. The Astros are not a restaurant that, when faced with dwindling returns, can shut down, renovate, hire a new chef, reimagine the menu and relaunch. They had to stay open for business. Business has been bad.
In Luhnow's first two seasons in charge, the Astros were 106--218. They drew a combined 3.3 million fans to Minute Maid Park—an attendance figure they had nearly reached in 2007 alone, when they topped three million. Several games have gotten local TV ratings of 0.0. The franchise has been accused of violating the most basic element of a baseball team's social compact—that it tries its best to win every game—and has angered the players' union with its low payrolls ($22 million as of Opening Day 2013, the lowest in the majors; $44 million this season, the second lowest). The team has even been made fun of by Alex Trebek, on Jeopardy! The answer, last November: "The large valve used to control wellbore fluids on oil rigs is this 'preventer'; the Astros could have used one." The question: "What is a blowout preventer?"
One result of their poor performance was that the Astros this year became the first team to have the first pick in three consecutive amateur drafts. This was never a goal, they insist, but a by-product of their long-term plan. Even so, it represented an opportunity. The right player might be the finishing piece on the championship teams they envision. They dreaded making the wrong decision.
TO CLUBS picking first overall—one-one, in baseball shorthand—high school pitchers are terrifying. They have displayed a greater chance of flaming out, due to injury or a failure to develop, than any other category of player. "There have been some wild successes," says Elias, "but the list of those picked high is littered with injuries and disappointments."
Between 1965, the first year of the draft, and 2013, clubs picked a high school hurler one-one just twice. In 1973 the Rangers chose a lefthander from Houston named David Clyde. Arm injuries ended Clyde's career when he was 26; he had a record of 18--33 and an ERA of 4.63. The Yankees tried again in 1991, when they selected a southpaw from North Carolina named Brien Taylor. Taylor tore up his shoulder in a fight in 1993. He would become one of three one-ones to never play in the majors at all.
The Astros had decided on less volatile categories of players with their two previous one-one picks. In 2012 they selected a 6'4" high school shortstop from Puerto Rico named Carlos Correa. The pick surprised the industry, but Elias had deep convictions about Correa from scouting him extensively when he was with the Cardinals, Mejdal's system liked him, and Correa had indicated that he would sign a contract that would be relatively cheap for a one-one. This year Correa was rated by Baseball America as the sport's seventh-best prospect, although he is now on the DL with a leg injury. Last year the Astros went with Stanford righthander Mark Appel, considered as risk-free a pitcher pick as has ever been made. This spring Appel was BA's 39th-rated prospect, though he has an 10.48 ERA through 221/3 innings in Class A this year. He has had tendinitis in his right thumb and an appendectomy, underscoring that the ride isn't always smooth even for the safest of prospects.
As 40 members of the Astros' front office staff—including Luhnow, Elias, Stearns, Goldstein, Mejdal and the other inhabitants of the Nerd Cave, all of their scouts and certain special assistants like Biggio—assembled in a conference room on the second floor of Houston's old Union Station, which abuts Minute Maid Park and contains the club's offices, they knew this year could be different. It was 10 a.m. on June 4, the day before the draft, and the men were there to provide their expert opinions on the six players who were still in the running for one-one. Two of the prospects were high school pitchers. "All right," said Elias, who was running the meeting, "this is your opportunity to air it out."
For the next 100 minutes the room discussed the prospects one by one. As each player's name was announced, his video clips were projected on a screen. First, the area scout who was responsible for the player would introduce him. Then anyone else who had seen him—Elias, national cross-checker David Post, special assistants—would chime in. Luhnow, who had also seen each of the six in person, would ask questions. Analyses of the player's swing or pitching mechanics, to which coaches within the organization contributed, would be read aloud. Finally, Mejdal's team would weigh in with its statistical projections.
It became clear that while the room liked each of the players very much, they were narrowing their focus to four: Carlos Rodon, a lefthander from N.C. State; Alex Jackson, a slugger from Rancho Bernardo High in Southern California; and Brady Aiken and Tyler Kolek, the high school pitchers.
A year ago Rodon had been considered almost a sure thing to go one-one, but a slightly down junior season had engendered some doubts in the industry about his command and efficiency. You wouldn't have known it based on the report that Tim Bittner, the area scout who had covered him, delivered. "The big thing for this guy is he has a pitch you don't see normally: it's a 70-grade slider"—out of 80—"at 88 to 91 miles an hour," Bittner said. "It's a weapon. It's a weapon now, it's a weapon on all levels." Mejdal's team revealed that one of the players to whom their metrics suggested Rodon was comparable was Chris Sale, the White Sox' ace and an annual Cy Young candidate.
You could sense the scouts' views of Jackson before the discussion of him had even begun. He had hit 47 home runs in high school. "Mmmm," they grunted, each time he unleashed his violently powerful swing on the video screen. "Mmmm." "Physically, he looks like Magglio Ordoñez," the area scout said. "A three or four hitter. Potentially hits 30 homers, with a .300 average."
"What about his swing?" Luhnow asked.
"Graded 80 out of 80," came the reply.
Area scout Brad Budzinski was similarly unequivocal about Aiken, a 6'4" lefty from San Diego's Cathedral Catholic High who had committed to play at UCLA and who threw a mid-90s fastball to go with a plus curveball and changeup. "I love everything about this kid," Budzinski said. "To me, we're getting possibly the next Andy Pettitte. Makeup-wise, I feel like it's Peyton Manning on a surfboard. A lot of people say they want to be a Hall of Famer, but I believe for this kid it's a realistic goal."
"If the stuff stayed the same as it is right now," said Post, "it's more than enough to pitch and have success in the big leagues."
Though Mejdal's department does not incorporate high school statistics into its formula—they are too misleading—it recited Aiken's stats anyway: "K's per nine of almost 17."
"Did he say 17?" one of the scouts in the back of the room whispered.
Kolek was also an attractive option. He stands 6'5" and weighs 260 pounds, and his fastball touches 102 miles an hour. "The stuff is as good as we've ever seen from a high school kid," Elias said. "I think we can all agree this is as seriously as we've considered taking a high school righty, and with good reason."
Kolek had attended Shepherd High, less than an hour northeast of Minute Maid Park, and the allure of drafting a local boy was considerable. "They've got a cool setup out there on their ranch," Elias told the room. "They've got a pond to fish in. They've got tractors that they drive around, chasing animals."
Elias grew up in the Washington, D.C., suburbs, the son of a Secret Service agent, and his novice description of hunting drew laughs from the outdoorsmen in the room. "Now, where are you from, Mike?" boomed a deep Texan voice. It belonged to Nolan Ryan, who is the father of Astros president Reid Ryan and serves as an executive adviser to the club.
"Nolan, how hard did you throw at his age?" a scout asked the alltime strikeout king, who had watched Kolek pitch in person.
"There weren't radar guns in those days," Ryan said. "But I can tell you, Nolan Ryan wasn't even close to what this kid is as a senior in high school. "
The meeting drew to a close at 11:40 a.m. "All right, it's a good group," Luhnow said to his 39-man brain trust. "Flip a coin now, or later?"
"If we take one of the high school pitchers, we have to be really [convinced] that this guy is the guy, and that's not real easy to settle on," Elias said later. "Especially when you've got other good options." The Astros' decision engine had one more day to make its choice.
*****
The Astros anticipated backlash against the rebuilding effort they planned to conduct with a purity that to their knowledge had never before been attempted. They have received it. It came most fiercely at the end of last season, after they had traded away the last of their mature assets in closer Jose Veras, outfielder Justin Maxwell and starter Bud Norris. They finished out the season with a 15-game losing streak. Their record, 51-111, tied for the majors' worst in a decade.
Before the season they'd hired a manager they felt was the right man to guide their players through such a stretch. Bo Porter understands the necessity of losing, or at least he professes to. "We had to go through that," says Porter, who is 41. "The biggest mistake organizations can make is the misevaluation of their own players. Had we not gone through what we went through last year, we wouldn't be where we're at today, because we'd still be trying to figure out who can we move forward with, who do we need to cut ties with. "
Even though Luhnow intellectually understands why his Astros must lose, he maintains that doesn't make it any easier. "The hardest part for me is when people think we don't care," he says. "We desperately care. Would I prefer to be able to do this with losing 70 games a year instead of 100? No question about it. Do I think it's possible? I really don't."
As for concerns about the payroll, "we feel we're going to have the resources we need to add the appropriate players to complement what we have to win when we need to win," Luhnow says.
Other criticisms have surfaced more recently. In an article published in the Houston Chronicle on May 25—the day, as it turned out, the Astros began a seven-game winning streak—beat writer Evan Drellich detailed the ways in which, as the headline read, RADICAL WAYS PAINT ASTROS AS 'OUTCAST.' "They are definitely the outcast of major league baseball right now, and it's kind of frustrating for everyone else to have to watch it," Norris, who was traded to the Orioles last July 31, told Drellich. "When you talk to agents, when you talk to other players and you talk amongst the league, yeah, there's going to be some opinions about it, and they're not always pretty."
The criticisms fell into two categories. The first was that the Astros' analytics-based approach dehumanizes players. "It was a difficult thing for me to read, because I spend so much time personally getting to know our players, and so does our staff," says Luhnow. "There is a perception that anybody who is doing analytics in a serious way is doing that at the expense of the human element. It's just not true, in our case."
Adds Mejdal, "We realize these are human beings, not widgets. As far as assigning a number to a person—well, I assume you get a salary? Do you feel dehumanized because your boss has put a number on you?"
The other criticism stemmed from the Astros' use of new competitive tactics, such as a heavy reliance on extreme defensive shifts. The club's proprietary database—christened Ground Control by Elias's wife, Alexandra—contains not just projections of the future value of every player but also spray charts for every hitter on every count against every type of pitch thrown by every type of pitcher, as well as the probabilistically optimal way to position defenders in each scenario. This sometimes leads to shifts in which, say, the Astros' second baseman plays well to the left of second base against a pull-happy righthanded hitter—a violation of traditional baseball norms, though one that's becoming more common across the game.
Mejdal puts the Astros' tactics into perspective. "A year ago, with the defensive positioning that was going on, we were in the top half dozen, and there was tremendous pushback," he says. "Well, the rate at which we shifted last year, that would be below average in the major leagues now. Innovation, by definition, suggests change will be taking place. If there's change taking place, it's not likely going to feel right at first. If it felt right, it would have been done a long time ago."
The Astros' leadership bristles at the notion that it thinks it knows how to operate better than anyone else. All it knows is what it believes to represent best long-term practices, based on the information it has acquired and processed. "We're far from perfect," Mejdal says. Even what they believe to be optimal decisions often don't work out. Sometimes a righthanded pull hitter goes the other way. Sometimes players they discard, or decline to draft, turn into stars. "Sometimes you hit on a 16," Mejdal says, "and if you stayed, you would have won."
AS 6 P.M. Central approached on the evening of Thursday, June 5, the majority of the Astros' scouting and analytics staff milled around the club's draft room. The metal walls were covered with magnets, each bearing the name of an amateur player. The staffers were waiting, like the rest of the baseball world, to see who the team's leadership would pick one-one. The day before they had dressed in khakis and oxford shirts, but now they wore suits and ties. If there was any need to remind them of the caliber of player they hoped to draft, there was the dinner they had just been served: Nolan Ryan Beef Brisket and Nolan Ryan Jalapeño Sausages.
Finally, at 6:05, Elias emerged from Luhnow's office, where he had been huddling with the GM, Stearns and Mejdal. He nonchalantly slapped the magnet bearing their pick's name at the top of the draft board. Minutes later commissioner Bud Selig announced the pick from the MLB Network studios in Secaucus, N.J. On the fuzzy big-screen TV mounted at the front of the room the Astros' scouts watched as the player, whose reaction the network's cameras were covering live from his home, buried his face in his hands.
"Oh, no!" a scout called out. "I don't want him to cry!"
There would be no tears from Brady Aiken, whose name was printed on Elias's magnet. Soon, Brad Budzinski, the young scout who had followed Aiken since he was 15, was accepting congratulations—"That's your guy, Budz!"—and handshakes. "A lot of seasoned scouts have never even had a first-rounder, let alone a one-one," Budzinski would say.
Luhnow tried to call Aiken on his cellphone, but Budzinski had given him the wrong number. "How well do you really know this guy?" a smiling Elias teased the scout. Then Luhnow appeared to connect. "Hey, Brady, it's Jeff Luhnow with the Astros," he said, as everyone listened in expectantly. Luhnow paused for dramatic effect. "Give me a call back when you get this." Laughter reverberated off the room's metal walls.
The decision to select Aiken over Kolek, Rodon and Jackson—who would be picked second, third and sixth, respectively—had not been a last-minute one. "We decided the morning of the draft," Elias says. "The mere fact that we were willing to take a high school pitcher one-one for the third time in history, even though the first two didn't pan out, showed us how strongly we agreed. We feel good enough about our farm system, that there's enough coming, that we don't want to look back in 10 years and say, 'We passed on the best high school lefty ever just to get something a little quicker.' "
Years of scouting reports, regressed in Mejdal's system, all suggested that Aiken was the draft's best player. Picking someone else simply because he was not a high school pitcher would have been the equivalent of staying on 16 against a dealer's seven. That is not something Luhnow's Astros do.
Luhnow knows there is a chance that Aiken—and, indeed, his own venture in Houston—might not work out. "There are injuries and declines in performance," says Luhnow. "Then there's the luck of playing games. Still, with all those unpredictable variables, I feel pretty good that we're putting ourselves in a situation where if we were to do this a million times, the odds would be in our favor to succeed."
Luhnow, however, does not discount the value of simple fate. "A memorabilia collector gave me the SPORTS ILLUSTRATED from my birth week in 1966," he says. "The issue came out on June 6. My birthdate is June 8. You know who's on the cover? The Houston Astros. ASTROS IN ORBIT, it says. Unbelievable."
In the draft room there were more immediate matters at hand. Not only did the Astros have 40 more picks to make, but they also were at the moment playing against Albert Pujols and the Angels. "We're losing 1--0," Craig Biggio announced, holding his smartphone aloft.
"Already?" said Luhnow. "How'd that happen?"
"Albert hit a sac fly."
It wasn't long before the Astros started scoring themselves—three RBIs came off the bat of Springer—and were on their way to their ninth win in their last 12 games.
"Oh, good, more points!" Mejdal deadpanned, glancing up at a TV.
"They're not points, Sig," said Kevin Goldstein.
Mejdal, like Luhnow, knows that even a long string of correct, intricately considered decisions might not turn out favorably. "What if we don't have good results?" he says. "I love my job in baseball. It would be terribly disappointing. But all we can control is the process, and I'm confident we're creating good processes and making good decisions.
"The rest," Mejdal says, "is hope."