Online Sports Betting on Cusp of Recommendation Engine Revolution
Today’s guest columnist is Lloyd Danzig, the founder and managing partner of Sharp Alpha Advisors.
In a future that’s nearer than many think, recommendation engines will have a transformative impact on people’s behaviors in all walks of life—including wagering on sports.
More from Sportico.com
Genius Sports Ready for MACtion With First NCAA Conference Data Deal
Ridley Betting Suspension Tests NFL Standards as Sportsbooks Expand
NHL MVP Connor McDavid Signs BetMGM Deal in U.S. Sportsbook First
Smart alarm clocks will analyze a user’s wearable fitness tracker and recent sleeping schedule to recommend the ideal time to wake up. Smart closets will cross reference a user’s work calendar and the day’s weather with each piece of clean clothing to recommend the optimal outfit. And, smart sportsbooks will leverage historical user data to recommend the exact bet that a user wants to place at precisely the moment they are interacting with the platform.
Recommendation engines are typically built using one of three methods—collaborative filtering, content-based filtering, or a hybrid of those two approaches.
Collaborative filtering is the process of analyzing past behavior to make predictions based on the exhibited preferences of statistically similar users.
For example, if Alice enjoys betting Same Game Parlays, Monday Night Football moneylines, and Anytime Touchdown Scorer props, while Bob enjoys Same Game Parlays, Monday Night Football moneylines, and First Basket Scored props, it is likely that Alice would like First Basket Scored props and Bob would enjoy Anytime Touchdown Scorer bets.
Content-based filtering identifies the core elements common to products or media that a user has previously shown a preference for in order to identify other products or media that contain those same elements.
For example, if Charlie exhibits a preference for wagering on baseball games that take place on Tuesdays when a heavy home team favorite playing in the same state in which he is geolocated, he can be served a pre-populated bet slip whenever those criteria are satisfied.
The third type of recommendation system is a hybrid of collaborative filtering and content-based filtering. Common to all of these methods is the process of factoring in users’ acceptance or rejection of the recommendations into future suggestions.
Of course, salespeople have been learning their customers’ needs and factoring in purchasing observations in order to make recommendations for as long as there have been salespeople. However, the vast array of available data points and corresponding non-linear relationships require machine processing to compute efficiently.
Recommendation engines will help sports bettors discover new bet types, new players and teams to root for, and new people to socialize and engage with. They also minimize the decision fatigue and anxiety that can degrade user experiences, thereby helping to maximize utility derived per dollar wagered, especially for the entertainment-seeking customer.
As with any algorithmically implemented business process, the quality of the output is directly correlated to the quality of the input. Garbage in, garbage out. Sports betting operators today are often constrained by siloed databases, technical debt and disparate work flows. For example, several sportsbooks have been slow to introduce Same Game Parlays because their third-party bet builder is incompatible with their risk and trading infrastructure or player account management system. Also, if not monitored properly, recommendation engines can reinforce unsustainable user behaviors and problem gambling tendencies. Operators with greater degrees of vertical integration—for instance, DraftKings, FanDuel and PointsBet, will be better positioned to streamline the customer journey.
Virtually every major industry is now dominated by large companies that sit atop mountains of first-party data, which is leveraged to build and calibrate finely tuned algorithms that enable the company to maximize profitability. Netflix and YouTube utilize recommendation engines to suggest content that a user is most likely to want to consume. Amazon leverages them to suggest products that a user is most likely to want to purchase. Even the auto-complete suggestions on search engines and email clients typically rely on these systems.
Sportsbook operators are no different. They are already deploying considerable resources to make subtle enhancements to their user interfaces that will drive retention and maximize ARPU. For example, the DraftKings app is now serving users with a small banner message that might say “Parlay AFC East Teams To Cover Today.” When the banner is pressed, it populates a bet slip offering a parlay that fits this description and only requires entry of the amount risked to place the bet. Eventually, these prompts will grow more contextually relevant and tailored to specific user behaviors. It is only a matter of time before recommendation systems are widely relied upon to deliver each customer with an optimally customized experience.
Sharp Alpha Advisors is a venture capital firm specializing in sports betting and online gaming.
Best of Sportico.com
