Amid the excitement of March Madness, the insatiable appetite by fans for these games continues to grow.
Turner Sports has revealed that a record 54 million live video streams have been watched by fans during the NCAA Tournament’s first weekend. 11 million hours of which derived from its March Madness Live property. The uptick in live video streams and numbers of hours consumed has risen year-over-year by seven percent and eight percent, respectively.
Of course, these statistics still come from users having to prove that they’re linear TV subscribers. This model for NCAA Tournament games has been around since 2011, with first sample-free, streaming testing dating back to 2006 by CBS. Opening the door for users to utilize their devices led to instant exponential opt-in: a 158 percent jump between 2012 to 2013. In the next annual interval, though, the increase totaled 40 percent. Negating over-the-top viewership could correlate to the stall in rapid gross figures, given the prerequisites policies in order to access.
Over the course of the opening weekend, though, there’s been a handful of games compelling enough that fans should tune-in, including the Notre Dame versus Northeastern matchup that drew nearly four million live video streams.
Deciding which March Madness games are important or of interest to the end-user depends on their respective preferences. Thinking generally along the lines of what specific game should be deemed attractive for the masses doesn’t directly dovetail into dividends, especially in a hyper-connected climate.
Thuuz, a Silicon Valley-based startup, takes, seemingly, the opposite approach for determining what’s must-watch TV.
They are able to receive a significant volume of sports data that’s produced, then convert it into a subjective output that’s user-friendly to be digestible to consume. Thuuz has developed its own algorithms that enables them to gauge the relative excitement level of any given game, relaying it back to the user’s personal interests. A couple of the data providers where they leverage this intel from are Opta Sports and STATS. This excitement meter, so to speak, is a compilation of anything from play-by-play data, time of possession, positioning within the playing field, and score. These analytical points become calculated to a 0-100-based scoring formula, which should help fans see what games are, indeed, intriguing to watch.
With this set-up, Thuuz has secured deals with the likes of TiVo and Dish. In these partnerships, their work serve as integrated features to guide users on what’s the best game on at the moment, all based on its excitement metric. For Thuuz’s own apps available in iOS and Android, users can quickly identify same point of interests, but specifically for their favorite teams, fantasy players, or sports leagues. A notification informs the users about the aforementioned details as well as what’s the channel to flip to based on their respective cable provider.
In light of the nature of the tournament, with so many games on close to each other and during work hours, Thuuz, naturally, can be a useful tool to keep track of all the action. Not to mention, its services benefit those users that want to receive notifications about non-March Madness sporting events to tune-in instead or in addition to.
Each day, Thuuz measures the excitement level of games in real-time–no different than any other point in the sports calendar year. They factor in historical data of these moments of high interest to form this predictor of games with the most potential watchability. Their goal on a nightly basis is to anticipate and calibrate excitement, it isn’t to accurately project what teams win or lose.
“Why do we find certain games exciting; and what are the factors that we are intuitively measuring in our minds when we determine a particular game is exciting?”
That’s the underlying question asked by Thuuz prior to developing its core algorithm.
“This led us to understand the specific factors that we intuitively sense when watching a sport,” Warren Packard, Thuuz’s Co-Founder and Chief Executive Officer, tells SportTechie.
“We then generalized these sentiments based on what most of us look for when it comes to excitement across all athletic competitions: high energy, close scores, rare performances, extreme suspense, uncertainty, comebacks, upsets, and meaningful implications (playoff impact or historical milestones),” continued Packard.
Accordingly, there’s six key factors that came as a byproduct of these generalized sentiments. By abstracting these six different perspectives as one, that’s where the dynamic excitement metric presents itself, with the average game rating being 50–for every single moment of each game on.
Pace is one of the factors considered, which specifies a game’s energy level. Parity is a second one, where the balance between two teams or a group of competitors are configured. Novelty stands as a third, which accounts for the uniqueness and entertainment level of specific plays throughout the game. Momentum measures the dominance of one team over another from one interval of time to another. Context covers the rivalry between teams or players, impact of game towards the playoff picture or overall leaderboard, or just the a reflection of the widely accepted belief that a certain game is that important. Social buzz, lastly, dovetails the amount of social conversation taking place around a game in comparison to the standard level of buzz that’s consistent with the average game within its respective sport’s season.
Bearing in mind the context of a given’s game variability, different data points do get weighted more heavily than others within the composition of Thuuz’s excitement meter.
Packard offers this example: “Early in a season, pace and novelty are more important since the playoff picture hasn’t been established yet. However, as a season progresses, parity and context become more important since the objective is to win championships; and that storyline begins to dominate as teams start vying for the final spots in the postseason.”
That said, they develop all of their own algorithms in order to carry out such analyses, be it before a game starts or during the process that it’s live. Various, different mathematical modeling techniques are utilized that produce these analyses, including machine learning. Thuuz believes that they are breaking ground by using algorithms to measure individual sentiment related to sports events.
“Given the subjective nature of human sentiment, we’ll never be 100 percent correct. We start out with rather simple algorithms and we add to them to constantly, as we detect nuances that are not being handled optimally. Over time, our algorithms get better and better, but we’ll never be done refining our algorithms; and incorporating additional personal details from sports fans to make our models better,” says Packard about their process.
So, as the Big Dance enters the Final Four, Thuuz’s predictor has served as a notable indicator for those teams that could provide interesting games. While certainly not a predictor pertinent to which teams will win out, the teams and players it highlights all season long are likely to be just as relevant during March Madness viewing.
“Thuuz is predicting and measuring sentiment, rather than performance. Of course, performance has an impact on overall sentiment, but we’re not trying to determine who is going to win or lose; we’re determining whether you’re going to be thoroughly entertained or not,” Packard states.
Thus, March Madness represents the ultimate event for Thuuz, with multiple games going on at once and fervor surrounding it round to round. Generating sentiment data–in its own fashion compared to the likes of Kaggle or Bing–from the memorable plays that take place in college basketball’s postseason presents the ideal vehicle to showcase its excitement meter. Pay-TV users are enabled to make better choices from what they’re purchasing insofar as being more satisfied with their respective provider. More viewership of the madness equals more advertising viewership, too.