A Database Of Dreams
ESPN The Magazine, Dec. 13, 2010
WHAT WOULD YOU DO IF YOU INVENTED THE PERFECT SPORTS-BETTING TOOL? (KEEP IT BETWEEN FRENDS, FOR STARTERS.)
By Chad Millman
Tom Stryker lives on a quiet block in Granger, Ind., a suburb of South Bend. He has a buzz cut, smiles easily, coaches his 10-year-old son's baseball team and sits with his 7-year-old daughter when she has strep throat. A baseball player at Indiana in the early 1980s, he's now in his mid-40s and needs reading glasses when he stares at his computer, which he does often. Because contrary to appearances, Stryker is keymaster for what may be the world's most valuable sports handicapping database.
It's a database so prized it costs $15,000 to buy, is owned by just nine people and hasn't been sold since 2007; it's so complicated that learning to use it requires several days of training; and it's so deep that filling it took three years. Why? It provides statistical information for every NFL and college football game dating back to 1980, and every NBA and college basketball game to 1990. "The man who invented this was a perfectionist," Stryker says. "And he had to be, because this database is perfect."
It's called the Buckeye Database and it was conceived of and bankrolled by Tom Scott, a Cornell grad, Ohio native and huge Ohio State fan. Scott, who was 67 when he died in May from kidney cancer, was not a big gambler but loved sports-the way each game presented a mystery that needed solving. He'd once been the safety director for a motor-parts company, moving throughout the Midwest before settling in Granger, where he spent off hours nose-deep in newspapers, transferring stats onto a legal pad and trying to come up with predictive formulas. "He was constantly tinkering, trying to figure out how things worked," says his widow, Barbara.
On the side, Scott-a pseudonym-started to write for a newsletter called Steam Sheet that offered betting advice. By the early 1980s he had turned to handicapping full-time; before the end of the decade he'd become one of the country's biggest touts-guys who sell betting picks-with a client list that reached into the thousands. "We called him The Innovator," says Stryker, who picked up his own pseudonym when he began working for Scott in 1984. "He was brilliant. But you really had to meet him to enjoy him. He could be, well, abrasive."
A round, 5'9", 240-pound man, Scott was irascible, generous and loud, with a booming voice that had no trouble navigating the cigarette that usually dangled from his mouth. "He was just very exacting," says his daughter, Andrea. "When we worked in the garden, he had a precise way the shovel should be planted. If we did it wrong, we had to do it again."
Handicapping is an odd choice of vocation for someone who considers every mistake to be a part of his permanent record. Perfection in this industry is a myth, as unattainable as 100% cell phone coverage. Most professionals will tell you if they win 54% of the games they bet on in a season, they've had a phenomenally good year. It's not just because bettors are subject to the whims of teenagers and temperamental coaches; it's also because of the overwhelming amount of data to process when analyzing a game. Professional bettors consider how teams have fared off byes and on the road. They worry about how many yards per rush teams allow and how many yards per pass attempt they gain. For every number, there are stats extending like roots that need to be pulled and examined.
The handicapping research Scott and Stryker performed was labor-intensive and included late nights that turned into overnights and early mornings. They combed through four file cabinets filled with newsletters like The Gold Sheet-The New York Times of the handicapping world-jotting down against-the-spread stats for various scenarios, trying to find trends. "We'd ask, ‘How do dogs do on the road after a big win?' " Stryker says. "Then we'd spend three days researching years' worth of Gold Sheets only to find the record was 36-36. We always joked about needing a computerized database, but we weren't using machines for anything other than holding a mailing list. We didn't know any better at the time."
AT LEAST since Christians began battling the lions, sports fans have had opinions about winners and losers. And then they bet. If you wanted the favorite-the lions-you were given shorter odds (2-1, 3-1, etc.). If you liked the underdog Christians, betting the higher risk meant longer odds. Either way, for the first 1,900 years or so of the Common Era, making a bet meant choosing a straight-up winner. But in the 1940s a University of Chicago-educated banking analyst named Charles McNeil, an avid gambler who supplemented his salary by betting baseball with bleacher bums, introduced an idea to the world.
McNeil had spent years using stats and probability theory to examine football teams. His logic was simple: Apply what he knew about gambling on the stock market to gambling on sports. That meant examining factors that contributed to winning and losing, then rating them according to importance. He was able to calculate, in his opinion,what team would win a game and by how much, which helped him find value in the odds bookmakers offered. If, for example, he determined the Bears would beat the Packers by two points, but bookmakers listed Green Bay as a 10-1 underdog, he liked those odds and bet the Packers. And he cleaned up-so much that many Chicago bookies stopped taking his bets.
Feeling shunned, McNeil started his own operation. In addition to offering odds on every game, he gave clients the choice to make bets based on his system. If he determined the Bears were only two points better than the Packers, he let people bet that the Bears would win by two or more. He called it "wholesaling odds," but it soon became known as the point spread. And that's when handicapping got really interesting.
McNeil's invention turned sports betting into an insider's game. It was no longer just about rooting for one team to win. Now losers could actually be winners and winners could wind up losers. Of course, beating the spread spawned new angles. Contrarian theory, which popped up in the 1960s, argued that bettors would find success wagering against popular public opinion. Another tactic, espoused in the 1970s by Huey Mahl, a Las Vegas gambling writer, relied on math to examine probabilities and percentages. In the '90s bettors started using computer models to play thousands of game simulations. No matter the idea, all were designed to do the same things: eliminate bad choices, pinpoint opportunities and provide certainty where there was none.
In many ways, handicapping today is a highly sophisticated science, steeped in PhD-level statistical analysis. In 2009, at the 14th International Conference on Gambling and Risk Taking, presentations included Exploiting Inefficiencies in Financial and Sports Gambling Markets. At the M Casino in Las Vegas, bookmakers create lines with the help of a supercomputer nicknamed Midas, which runs algorithms that determine game outcomes. This April, a London-based investment firm introduced a sports-betting hedge fund in which analysts use mathematical models to make thousands of bets.
Even the bettors tend to be more nerdy than naughty these days. Big Al McMordie, a top handicapper, is a former corporate lawyer with a law degree from the University of Michigan. Dr. Bob, another high-profile 'capper, studied stats and economics at Cal. But in the late '80s most handicappers were like Scott and Stryker- still in the hash marks and scratch-paper phase-who could only dream of Jetsons-like computing efficiency. Then in 1991, after a particularly flush year, Scott happened to be seated next to a computer scientist on a flight out of South Bend. "I'd like to build a computer database that can help me with handicapping," Scott told him. "Is something like that possible?"
"It is," the man answered. "I know a computer programmer who can help."
FOR YEARS, Stryker, Scott and a Cleveland handicapper named Marc Lawrence had traded ideas about what this database would look like. So when the programmer arrived at Scott's office, Scott was ready with a neatly typed fivepage document titled "The Buckeye Database." Scott had listed 37 fields he wanted searchable, as well as definitions for each one: "SURF" for surface, "LINE" for point spread and "RATS" for a team's current win or losing streak against the spread. The mandate, Scott wrote, was "to ask hypothetical questions that do not require several pots of coffee to answer."
It ultimately took three years to finish and included 600,000-plus bits of initial data, most of it tediously entered by Stryker. And it worked. Suddenly the record for any situation a team had ever played in-at home off a loss against a team off a win; a team on the road as a favorite after two straight wins of 10 or more points- could be determined immediately. Says Stryker: "What had taken 23 hours now took 23 seconds."
Every gambler, from penny players to hardened Vegas vets, is constantly on the lookout for that sure thing, the angle that comes from that secret place where only wiseguys seem to roam. Well, Scott and Stryker now had the answer to every angle they could conceive of at their fingertips. This was a eureka moment, the handicapping equivalent of mapping the genome- nothing with its statistical depth and breadth existed in the gambling world. And they were both in disbelief that it actually functioned. Looking at, say, a slate of Saturday college football games, the duo scanned each team's previous results for ideas about angles. When inspiration struck, they opened up the database and typed in a query like: "How do home favorites of 20 or more that lost by 20 or more the previous week perform?" Within seconds, they got the answer (80-49-2 the past 30 years, by the way). There was no more guessing if home dogs always covered on Monday nights (answer: 87-71 against the spread since 1980). "For days we just played with it, testing it, running different angles, trying to understand its power," Stryker says. "We loved handicapping and wanted to find answers."
Scott called Lawrence, telling him, "We finally got the boat in the water." And from there, word about the new tool leaked out. The businessman in Scott quickly realized the potential market for selling it was lucrative. But the handicapper in him knew how powerful an edge the database could be if he kept it locked in his office. So he made a decision: He'd offer it only to friends he trusted or to those whose work he admired. And the asking price was a steep $15,000. "He just wanted to throw out a big number to see if anyone would take," says Stryker. "He was surprised people did."
Over the years, Scott chose nine lucky winners, yielding approximately $115,000 in profit on his investment. With their purchase, buyers were invited to Scott's house in Indiana, where he made a meal fit for a wiseguy: The vegetables came from his garden, the wine from his cellar. Stryker trained them to use the database, and each left with a copy on a jump drive. Even in the Internet age, Scott refused to e-mail the information. If you wanted it, you had to come get it. "I remember flying to Indiana to check it out a year after it was done," says Dr. Bob, one of the chosen ones. "It was magic. I was like, Wow, it's all here."
The database became legendary around the handicapping community. Not only because of the price, but because of its exclusivity. "There was something mythological about it in the forums," says Sal Selvaggio, president of madduxsports. com and the last of Scott's buyers. "People think you get the database and automatically become a winner."
It certainly seems that way. There's a reason why Lawrence, a 35-year industry vet who's relocated to Florida, is only half-joking when he says, "It's become my closest friend." Stryker opened Team Stryker Sports-his own tout service-in 1999 after buying the database. Big Al went from an apartment in Bethesda, Md., to a house in the Hollywood Hills. Sal became one of the most trusted touts on the web, and The Wall Street Journal called Dr. Bob "one of the world's most influential sports handicappers."
But those who used it also quickly learned this: The database is a living, breathing entity. It requires daily updates of scores, spreads and statistical minutiae that can take hours. It requires the user to be curious, because the database cannot answer a question it isn't asked. Dr. Bob has worked up more than 1,000 queries combined over the years for college and pro football and hoops, and he estimates he's thrown away 15% of those because they no longer work. Big Al has 2,000 queries and adds 50 to 100 new ones every year for each sport.
Mostly though, it requires the user to remember that, in betting, there is no sure thing, despite what the computer says. On college football Dr. Bob hit more than 70% in 2005, then just 46% two years later. Selvaggio and Big Al aren't strangers to up-and-down years either. They know past performance does not guarantee future results. "It's a valuable tool, but there are also dangers with it," says Big Al. "It provides objective analysis, but not subjective analysis."
It may be able to tell you that defending Super Bowl champs on the road off a loss, and favored by less than seven against a .670 (or better) foe off a win, are 13-0-1 since 1980 (see sidebar). But it can't tell you there's friction in Minnesota's locker room, or that the Panthers started a quarterback in Week 11 who had been out of the league in Week 10. So when Stryker says this database is perfect, he really means it's as perfect as it can be. In the same way Charles McNeil weighed several statistical factors when he created the point spread, handicappers use the database as one of several decision-making tools to beat it. It helps confirm their theories, helps disprove their theories, but it doesn't create theories. The database can be only as wise as its owner.
SHORTLY AFTER Scott died, Barbara was going through his office and found binders filled with systems and the original Buckeye Database memo. She asked Stryker to come over and offered up the keepsakes; they were tokens only he could appreciate. As he stood in the basement holding his old boss's things, he remembered the times they sat in Scott's garden, riffing on systems, discussing how to use this database they'd created. Then he noticed Scott's old computer sitting on his desk. It was a 1995 Macintosh, the one he bought right after they had finished loading college basketball and the NBA. He had never upgraded. Stryker asked Barbara if he could have that, too. When he got it home, he turned it on and sifted through Scott's databases, discovering even more systems and analysis-a treasure chest of handicapping data. He looked through them for several minutes before turning it off. Just as the screen went black, Scott's voice, recorded on the computer, screamed, "I'm done!"
His database, though, is far from it.