While Data Gathering In Sports Becomes Ubiquitous, This Startup Aims To Add Meaning To It All
While San Francisco is gearing up for Super Bowl 50, four teams are still in contention for the Vince Lombardi Trophy. Those four teams have something in common – other than making it to their conference title game. The Carolina Panthers, Denver Broncos, Arizona Cardinals, and New England Patriots rank one through four in strongest defenses in 1st and 10 situations this season.
Now, this sounds like one of those interesting, but kind of random, stats that an analyst throws out during a game broadcast – a stat that he clearly got from a researcher that the average fan would not have access to.
Tom Covington and Jesse Paquette are changing the game of data analysis with Tag.bio, launching today out of San Francisco.
Covington, a former engineer, and Paquette, a former computational biologist in the Cancer Center at University of California, San Francisco (UCSF) met playing soccer six years ago. While at UCSF, Paquette developed a software application known as Exploratory Gene Association Networks (EGAN), which visually interprets assay results and allows scientists to analyze the significance of patterns in the data. Essentially, EGAN allows scientists to focus on interpreting the meaning of the data instead of relying solely on the insights the computer generates from analyzing data.
The two were intrigued with how this software could impact other fields and began exploring how they could help anyone become a data scientist. They started with sportsbecause the data is clean and curated, and they knew what they could do with the information they found. They began by asking the question ‘is EGAN useful for teams?’ The answer is yes.
They found that soccer clubs didn’t have the same information they were generating, and more importantly, the data they could provide helped the teams. Football teams also didn’t have the same information while baseball teams had access to the data, but this new software could provide it faster. So two years ago they developed a spinout from UCSF and founded Tag.bio, a mobile-first data analytics platform.
Their current partner teams create private instances to analyze data on their own performance as well as their opponents’ performances. While detailed results of team’s success are confidential, Covington told me there have been ‘things that shocked them about themselves and upcoming opponents.’ Paquette added, ‘in other words, it’s working.’
There is also great value beyond the coaching and training staff. Using the public datasets, the media can provide unique analysis and show interesting data visualizations on screen. Fantasy enthusiasts can dominate their leagues. The average player can now run analyses and instantly find insights about teams and players, helping to level the playing field in fantasy sports. A routine set of stats can help you determine whether the player who’s only given you two points the last four weeks is primed against a certain defense to give you 10 or 20 points.
Using a hypervisor that sits on top of any dataset, Tag.bio performs sophisticated analysis to find patterns. Users can ask any question of existing or custom analysis modules, known as protocols, to easily find insights and determine which insight is the most impactful. A tag score indicates the confidence of the likelihood of a certain protocol; the higher the tag score, the more confidence you can have. Each hypervisor turns the dataset into an individual API, allowing you to build a report to share or keep private.
Each report is posted as a blog post with easy to share social links. Reports are enriched with data visualizations including charts and graphs in addition to presenting information in the context of the field or strike zone.
Going back to the four playoff teams remaining and their defenses in 1st and situations this season – that’s analysis that Tag.bio generated. The protocol in question is ‘Which defenses allow more success?’ The scenario is a 1st down and 10 yards-to-go situation and success is defined as their opponent gaining five or more yards. The enrichment ranking is then derived from the insights pulled. On average, teams have a 42.5% success rate against teams at that down and distance. The four playoff teams hold opponents to lower success rates than the average. At number one, the Panthers defense only allows opponents to gain five or more yards on 1st and 10 34.5% of the time.
Tag.bio has hundreds of protocols you can analyze for football, both at the professional and collegiate level. In addition to protocols for baseball, they are also developing an analytics product with a real-time predictive data stream. They also hope to expand beyond sportsinto non-profits and eventually back into the academic and research space.
Now anyone can find insights about their favorite teams or players instantly and you can finally dominate your fantasy league. Teams have an easier and faster way to discover patterns in their performance, and their opponent’s tendencies, that they can analyze to prepare for competition.
It will be interesting to watch the impact of using the power of computers to augment human insight will have on sports analysis, and ultimately performance. But for now, go check out this new platform and try your hand at being a data scientist. You’ll definitely find a nugget or two you can use to impress your friends at your Super Bowl party.
The author of this story was Lindsey Ann von Thron, permission to re-publish from SportTechie; link to the original.