Like Koerl, Arnesen and Fornaess were interested in sports and computer programming. And gambling. They knew enough about the sports betting market to recognize its inconsistencies. Bookmakers were good at analyzing a match and offering the kinds of odds that would generate action and produce profit for the company. Setting the right odds was the bookie’s special talent. What bookies weren’t good at was paying attention to small details. And these small details could make a big difference.
Arnesen and Fornaess realized that bookies, since they were dealing with so many games, often listed the incorrect starting time of a match. Sometimes match administrators moved a contest from one site to another, a change that could affect the outcome, yet a bookie might fail to report this fact. A bookie could receive the wrong information from a spotter and list the incorrect final score of a game. Also, without noticing it, a bookmaker might list odds that were, say, 15 percent higher or lower than the market average. If a bettor possessed more accurate information about a match than did the bookie with whom he was wagering, then he had the upper hand.
Bookies were taking action on league matches, international friendlies, World Cup qualifiers, youth games—from all over the world and at a constant pace. With so much volume, bookies simply couldn’t keep track of every detail of every match that they listed. Arnesen and Fornaess developed a method by which they could. They created an algorithm that could handle mass data extraction, specific to the online betting market. Using the data supplied by the online books themselves, their program was able to pinpoint inconsistencies in the market. Arnesen and Fornaess could identify the weak link (or links), the outlier in the network of international bookmakers, and then exploit it. When they detected which book was listing the wrong odds on a particular match, they would place a bet on that game with that book. They had figured out a way to game the system, using one book against many others to make a profit.
As exhilarating and profitable as this was, Arnesen and Fornaess understood that as the market became more developed, which was an inevitability, bookmakers would gradually address and eliminate these inefficiencies. Eventually, Arnesen and Fornaess would use their algorithm to squeeze out smaller and smaller profits, until their tool became virtually worthless. They believed that there was a better, more sustainable way to utilize their program. They decided to develop an instrument to aid bookmakers in correcting their offerings.
They pitched the idea to Bwin. But they weren’t businessmen. They were programmers. Their product and pitch were raw. All the same, Koerl recognized an opportunity, for he was the entrepreneur. And after Bwin’s perfectly timed IPO, he had the abundant capital to finance a start-up.
In 2004, Koerl, Arnesen, and Fornaess established a new company, calling it Sportradar. The company’s product, which was called Betradar, expanded upon the technology that Arnesen and Fornaess had been using to beat the bookies, but with Koerl’s new market-specific modifications. “From my time at Bwin, I knew what a bookmaker needed, because I was a bookmaker,” Koerl says. “I understood risk management, because I learned about it in a painful way, from clients cheating on me. I had precise knowledge of what I needed. That was our advantage. This was something that the bookmaker needed.” For a monthly fee, which a bookie could build into the back end of his business, like insurance, Betradar would handle the tedious tasks of verification and data collection for which bookies traditionally displayed little talent or patience. Koerl was back in the bookmaking game, yet in a decidedly different role. It wasn’t long before that role would change once more.
In January 2005, German soccer officials announced the suspension of referee Robert Hoyzer. During an August 2004 German Cup match between Paderborn and Hamburger SV, Hoyzer had awarded underdog Paderborn two questionable penalty kicks. He had also dispatched a Hamburger player by red card. Paderborn won by two goals. An investigation determined that Hoyzer had bet on the match.
As the German soccer federation combed through Hoyzer’s match record, the German prosecutor’s office initiated its own investigation. It turned out that Croatian organized crime had been controlling Hoyzer, who had fixed a number of matches. And he wasn’t the only one. Revelations implicated dozens of players and referees, even hinting that someone from inside UEFA was leaking the identities of those who would referee Champions League matches, allowing the Croatians to compromise them. Hoyzer was sentenced to two years and five months in prison. The leader of the betting ring, Ante Sapina, received two years and eleven months. The case was closed, but Koerl knew that this was not an isolated incident.
From years working in the betting business, Koerl knew that fixing was happening with regularity. Every bookie knew. They could usually tell when something untoward was happening with a match. It was one of the challenges of the business. Not only was a bookmaker concerned with establishing the proper odds to offer his clients, but he also had to worry about fraud in the marketplace upending all of his calculations and ultimately cleaning him out. Everyone in the business had general knowledge of fixing. What made Koerl different was his specific knowledge. He just didn’t know that he had it.
Shortly after the Hoyzer case broke, Koerl received a phone call from a journalist. The reporter informed him that several bookies had revealed that they had recognized these fixed German matches through the Betradar interface. One component of the software was called “removed lines.” This displayed bookmakers that had offered betting on a match, but then subsequently removed it. This had always been a curious practice, for if a bookie removed a match after first listing it, he must know that the match was tainted. Combining this element with other factors of the Sportradar software, Koerl approached the German Football Association.
He explained that his technology could alert the federation to the likelihood of a fixed match. And it could do it in advance of a game, so that the federation could take action to protect its organizational integrity. When the German Football Association, mired in scandal, enthusiastically signed a contract, Koerl and his team set about modifying the technology. The new product had to be designed not for bookies, but for soccer administrators.
Koerl’s new system analyzed the numerous characteristics of a given match—its location, the relative strengths of the opponents, player injuries, the weather, and many more data points—in order to produce the most logical betting line. This was the work of a bookie, establishing a baseline. Then Koerl’s system compared this ideal, baseline bet to the actual offers of the sportsbooks. When the market offers deviated wildly from the established baseline, this was a signal that a match was questionable. Koerl knew that there were only two reasons why a bookie would move his numbers drastically from what they should be. Either the bookie had inside information, or he had taken big action that he didn’t want—that is, bets on a fixed match. For example, if a team’s chance of winning is initially 45 percent, and then a bookie pegs the chances at 70 percent, the game is undoubtedly fixed.
The program served a function similar to the analysis of insider trading on a stock market. If the price of a share suddenly collapses for no apparent reason—the CEO hasn’t been fired, the quarterly earnings report contains no surprises—there is likely some nefarious explanation for the drop in value. Koerl’s system was able to review the share price of a soccer game, then investigate why it had collapsed. He recognized that he could see the marketplace deteriorate ahead of time.
Koerl named his new product the Early Warning System (EWS). Beginning in 2005, with its first client, the German Football Association, EWS monitored five levels of German soccer, from the Bundesliga on down, roughly five thousand matches a year. Koerl then forged deals with the football federations of the Czech Republic, Estonia, and France. When UEFA signed on as a client shortly thereafter, Koerl’s new venture attained unquestioned credibility. EWS was now overlooking every professional soccer match in Europe, from fifty-four member states. This arrangement enabled EWS to ana
lyze not just the activity in one country, but the ways in which fixing in one country influenced fixing in another, and how betting was being placed in Asia. By the time FIFA signed on, in 2007, EWS was the recognized authority in its field, with Koerl supporting governments and police across Europe. It had taken a few years, but soccer administrators had come to realize that the EWS was filling a necessary gap in their security.
“Manipulation is damaging to the commercial possibilities of sport,” Koerl says. “Federations are very aware of the connection. It’s a massive commercial industry.” Altruism aside, protecting the integrity of the games with EWS supported Sportradar’s core product, Betradar. “Do you think sports betting would exist if people knew the outcomes of the game before it starts?” Koerl asks. “I think not.” There was just one complication. Koerl’s clients didn’t know it, but by 2007, his EWS was obsolete.
Only insiders understood that by 2007, technological innovation had fundamentally transformed the betting market. Criminal syndicates were now able to manipulate bookmakers with impunity, hitting this corner candy store over and over. And there was no way to detect their crime. Koerl understood that he would have to adapt his technology to the new reality. When FIFA developed its own in-house monitoring algorithm and called it the Early Warning System, Koerl let them have the name. In sports betting, there was no such thing as an early warning anymore. “The name was not really appropriate for the kind of manipulation that had developed based on how popular live betting became,” Koerl says.
Live betting, or in-game betting, existed before 2005. But 2005 was the year that it began to affect the marketplace in a significant way. Traditionally, a bettor placed a wager on a game before it started. And he was stuck with the bet. If the game’s events began to affect his chances adversely, there was nothing he could do. He couldn’t relinquish his bet. He couldn’t make a new one, since the game was already under way. He might yell at the TV screen, but that was the extent of his ability to express himself.
By 2005, technology was beginning to enable the bettor to monetize the thoughts he was having about the flow of play in a live match. Broadband Internet had penetrated deeply into homes in Europe and Asia. Smartphones were taking over outside the home. Instead of having to walk down to the bookmaker’s shop to place a bet, a bettor could bet on his computer at home, or on his phone while at the bar watching a game. Online portals like the one that Koerl had pioneered at Bwin began taking abundant action.
And the kind of action they were offering changed in accordance with the new form of betting. Bookmakers realized that online betting allowed them to provide a new array of offers, stimulating the betting market and driving liquidity.
The types of bets didn’t change. You could still bet the Asian handicap, the over-under, and other wagers familiar to the traditional client. What in-game betting changed was how frequently the dynamic of these bets changed, as a match progressed. Any event in a match—a goal, a red card, or simply expiring time—affected the game itself, and thus altered the probability of the bets related to its outcome. As a match played out, a bookmaker would now change the odds he offered on a bet depending on the shifting factors of the game action. If a bettor realized that his chances of winning a wager that he had already placed were decreasing, he could now bet against himself, or he could place another bet entirely. He was no longer stuck with the bet he had made before a game. And he could place bets throughout a match. Suddenly, sports betting changed from a static, passive experience—a fan’s experience of watching the game—into a frenetic opportunity for arbitrage, stimulating the adrenaline that one might feel by playing the game. Bettors converted to the new way in droves, fundamentally altering the betting landscape. In 2003, 10 percent of Bwin’s online sportsbook business was in-game betting. By 2010, in-game betting represented 70 percent of Bwin’s business.
Sophisticated criminals are quick to take advantage of innovation. The match-fixing syndicates found enormous potential in the advent of in-game betting. Previously, they had to guard against placing pregame bets so high that they would alert bookmakers to an upcoming fix. Now they could wait until a match began, then place their bets during the action. Once a game had started, a league or federation wasn’t going to stop it. And if the syndicates placed their bets judiciously, it would be too late before the bookie could detect them, if ever.
Live betting also placed a fixer in the stands at a game, smartphone in hand, monitoring the shifting odds of numerous bookies, communicating with his players and referee. These constant line and odds changes stimulated betting, invigorating the market. There was no innovation that could have better enhanced the business of fixing.
Carsten Koerl alerted his clients to this new reality, to the dangers of live betting. He also told them that he had developed the tool to combat it.
Sportradar currently employs 650 people worldwide. Two hundred IT experts code, develop, and maintain the infrastructure. Another couple of hundred operatives manually collect data from matches. There are roughly 150 journalists on stringer agreements to provide the minutiae from remote locales and clubs. To the games themselves, Sportradar dispatches roughly two thousand scouts, who for a nominal fee transfer match info directly by touch-tone phone keypad. All of this data goes into the system.
In 2009, Koerl launched Sportradar’s new product, the Fraud Detection System (FDS), which analyzes the shifting data from a game in order to identify a fixed match, real-time. The system originates its baseline odds based on three parameters: time elapsed and remaining in a game, the number of goals for each team and total, and the number of red cards. This allows the system to calculate the true probability of a team’s winning a match. The FDS then overlays this against the offers of 350 bookmakers, measuring the differences. When the deviation between the baseline offer and the actual offers grows too great, the system issues an alert to Sportradar’s analysts. If a match is considered suspect, analysts assign a warning level to it, yellow for suspicion, red for certainty. Everything happens instantaneously.
“Bookmaking is not a genius sitting in a dark chamber coming up with odds,” Koerl says. “That doesn’t happen anymore. It’s driven by algorithms, mathematicians, massive progression models. It’s much more related to financial markets.”
And as the FDS has progressed away from the individual, it has become more personal. Prosecutors across Europe frequently rely on Sportradar’s data in fixing cases. Through these experiences, Koerl has realized that what prosecutors and judges value is not raw data or endless analysis, but evidence that implicates players, refs, and coaches. “We learned that the system needs to deliver more information about individuals,” he says.
Sportradar developers added features that enabled the FDS to identify and track suspicious actors. If one club had participated in manipulated matches, and one or more of its players ultimately moved to another team, the FDS would make a notation that allowed a system user to collate related information. The FDS goes into great detail, pinpointing even a player’s performance in a game, a season, or a career. If a defender causes two penalty kicks in a match, for example, the likelihood of his being involved in a fix increases, a fact duly noted in the system. With 190,000 personal profiles in its database, the FDS is the Big Brother of match-fixing.
But what good is all of this data? Some soccer industry professionals complain that Koerl and his people at Sportradar are a mop-up crew. Once the fixing syndicate strikes, it is gone, it has won its bets, and the game is over. The FDS is not a system of prevention. It is retrospective, even as it clocks real-time progressions in the market. Some federation officials complain about Sportradar’s voluminous reports about fixed matches, the paperwork filled with data and figures that mean little to anyone but an IT professional.
Conversely, Koerl claims that he has provided some of his clients with information that points unquestionably to the fraudulent ac
tivity of players in their leagues, to no avail. “Sometimes I’m not happy with some federations, how they integrate our information, how they go about using it,” he says. “I tell them the right thing to do, without political intentions.” The recognized authority in his field, Koerl has been searching for the landmark criminal case that would prove Sportradar’s prescriptive value.
CHAPTER 19
ANTALYA, TURKEY, FEBRUARY 2011
Sportradar was scrutinizing events in Turkey. The Mediterranean resort town of Antalya was hosting two friendlies: Latvia versus Bolivia, and Estonia versus Bulgaria. This was another round in the endless series of exhibitions that held relatively little significance in any standings. This inconsequence made matches of this nature particularly vulnerable to manipulation, though that didn’t stop bookmakers from offering odds on them. Invariably, any game between national teams drew mass liquidity in the global market. Recognizing the fact that large numbers of bettors were eager to wager on a game for which the players themselves didn’t care, was elemental to understanding how the fixing syndicate was able to strangle the sport. That was the criminal genius behind Perumal’s fake Togo match. He realized that it didn’t matter who wore the national team jersey. People were going to bet regardless.
Footy Media International had arranged the Antalya games (in the name of Blapp Johnsen), and in a way that immediately aroused the suspicions of Janis Mezeckis, the general secretary of the Latvian Football Federation. On behalf of Footy Media, Anthony Raj Santia had traveled to Riga and offered the federation €30,000. He said that his company would arrange travel and accommodation for the Latvians, as well as the rental of the stadium and the selection of match referees. Mezeckis signed a contract with Footy Media on December 14, 2010. But he and his colleagues questioned how Footy Media planned to profit from the matches, as TV and sponsorship rights were not included in the proposal. Santia claimed that his company was eager to build its profile in Europe and so viewed this deal as an investment in future business.
The Big Fix Page 11