Toronto Raptors Digital War Room

Toronto Raptors Digital War Room

By Team CodeBusters

The NBA Draft is a nerve-wracking lottery, and not just in how the first few picks are awarded. Under time pressure, teams are making multi-million dollar bets on which players will become high performing professionals. The hurdles and pitfalls of the wrong decision are numerous, but the potential rewards for picking right are huge. How should a team choose the right player? Historically, teams used highly manual, slow processes prone to human biases and errors including Excel spreadsheets of player metrics, scouting, interviews, and exercise drills. At best, these factors offer a partial snapshot of a player’s future performance at the NBA level.

In February 2016, the Toronto Raptors unveiled their digital war room that captures data from advanced cognitive technology from Watson. IBM’s Sports Insights Central solution utilizes real-time data to assess a player’s organization fit and advise the Raptors on which player to choose. The system pulls in unstructured data from a variety of sources, including the players’ statistics, medical records, and social media profile, and then quickly assesses how well the player would meet the organization’s specific needs. The tool is geared towards getting a more holistic view of the team: Watson is able to quickly analyze a player not only in isolation, but how he would perform in combination with other draft picks and players already on the team. The Raptors are even able to assess cultural fit in a more rigorous manner, using profiles created from social media habits using the Watson Personality Insights tool. The war room is meant to complement, not replace, the decision making skills of coaches and managers. The war room features visualizations and other dashboards that present data to managers in a digestible way. Humans are still making the final decisions, but are doing so with much more relevant and timely information.

We have three primary concerns about the effectiveness of the Raptors’ current program: 1) difficulty measuring impact, 2) long, unclear commercial payoff, and 3) the efficacy of existing tools for critical choices. Since the implementation of the program, the Raptors have demonstrated marginal improvement in their win-loss record – illustrated in the table below. Beyond this, it is difficult to construct a useful framework to measure impact, especially from a commercial perspective. Often, commercial success is dominated by non-performance factors. Despite many losing seasons, Forbes rates the Knicks the NBA’s most valuable franchise simply because they play in New York city. Star players are drafted at the ages of 18-19 and tend to be 7-8 years away from their prime. The length of this horizon contributes to measurement challenges and diminishes the investment’s attractiveness since payoffs are many years away. Finally, basketball is uniquely dominated by the superstar. Unlike other team sports, single players can dominate their teams and the league, overwhelming draft room improvements. These talents tend to be well-identified and are selected at the top of the draft. Even the smartest draft room has no chance of selecting Kevin Durant, Russell Westbrook, Anthony Davis, etc. unless the franchise is bad enough to earn a top draft slot. In contrast, the most basic draft room would have made the right choice on the most important draft selection of the last 15 years: LeBron James. It is unclear whether superstars who have emerged from later in the draft (Kawhi Leonard, Jimmy Butler) were cases of prescient player selection or fantastic coaching and player development.

While determining a player’s “fit” with the raptors based on Watson’s personality Insights is a nice first step, the team could go further and solicit feedback from current raptor’s players and personnel to crowd source a fit score. The team could also target players based on their publicity and sponsorship potential, and target players who will put the most fans in the seats, increasing the revenue and profit of the team. The existing draft process is very competitive, and in basketball the game-changing recruits are generally well known to all teams, so there may be limited value and delta in applying this model to drafting, but it may be useful in other areas teams haven’t solved as much. For instance, using the model to target veterans that can help develop your younger players develop and teach them good habits. The competitive focus around the draft may limit the edge the model provides the raptors, so finding novel data to utilize or creative applications of the model may give them more of an advantage.

Appendix I (Win Statistics)