AI and the Sporting Edge
- By Lachlan Colquhoun
- April 20, 2023
At the 2021 Tokyo Olympics, Australian swimmers used a system called “Sparta Two,” which automatically tracked the performance of every swimmer in a race.
The system delivered details on stroke rates, turn time and breathing rates and was the culmination of four years of work at the government-funded Australian Institute of Sport.
The Tokyo Olympics was the most successful for Australia’s swim team, the Dolphins, which collected nine gold, three silver and nine bronze medals.
The contribution that the “Sparta Two” data project made to this success is difficult to quantify, but Australian officials know one thing: by the time the Olympics return to Paris in 2024, all the other teams will have developed their equivalent systems as part of the ongoing technology war in sport.
Predicting outcomes
GPS devices have been collecting data from athletes for some years. But it is only with the application of artificial intelligence (AI) to this data that sporting coaches are moving from recording and analyzing the data to predicting what might happen next or what tactics or strategies might be successful. In short, AI is emerging as the new edge in sporting performance.
There have been several landmark moments along the way. One was not technically in the area of sports but in gaming. When AlphaGo, a program developed by the London company DeepMind Technologies, defeated South Korean world champion Lee Sedol at the ancient Chinese game of Go in 2016, many people in sports realized that technology could be successfully applied to strategy and tactics.
“The global player tracking market is expected to grow at a compound 24.9% over the next 5 years, with teams in even minor leagues using solutions”
Four years ago, a regional English football team called Wingate & Finchley used a rebadged Amazon Alexa device plugged into a computer program to analyze data points from recent matches.
Human coaches asked the AI coach for tactical suggestions, and it even delivered some inspirational lines for the coaches to use in talking to the players. The result, with the team facing relegation if they lost, was a draw.
From there, things have developed fast in the world of football. The US Soccer Federation has built a neural network to understand why some attacking moves succeed and others fail.
The AI coach
There are new innovative start-ups. SportsPower AI has created an AI assistant coach who offers expertise in several sports and can deliver coaching suggestions in real-time.
There are other applications too. Recruiting and selection have long been considered a “dark art” and have given rise to theories such as Moneyball, which uses digital analysis to understand the right blend of skills and personalities in a team.
British football Brentford used computer analysis in its recruitment of striker Ollie Watkins. The club paid GBP2 million for Watkins, who was instrumental in their rise to the Premier League, and then later sold him for GBP35 million, and he now represents England.
These theories have progressed as digital tools have evolved, and data science now empowers coaches to complete exhaustive analyses of the parameters of potential team selections.
Machine learning applied to data collected on the field delivers a complete analysis of an athlete’s performance and skill sets, and technology is getting ever closer to developing accurate programs which can determine the likely performance of individual players and a team's combined performance.
There’s also an industry developing. The global player tracking market is expected to grow at a compound 24.9% over the next 5 years, with teams in even minor leagues using solutions to understand better player movements, who is performing and who isn’t.
Then there is coaching itself, where AI seeks to eliminate some of the flaws in conventional coaching. Coaches are human, after all, and often forget things or develop biases. AI delivers a stream of accurate analysis, which keeps coaches on point.
At the same time, analysis of a bowler’s action in cricket, for example, can help develop better training modules and identify points of stress which could worsen through repetition. In this way, AI can contribute to keeping athletes injury free and be a valuable addition to the medical coaching staff.
Leveling the playing field
Another area is in umpiring and adjudication. Almost all professional sports now rely on video replays to decide contentious decisions. Did the batter hit the ball? Was that goal touched on the line? Was that player off-side?
Through storing banks of accumulated data of similar past incidents, AI will learn to make faster and more accurate decisions based on video evidence, and rule infringements will become easier to identify.
As fans know, these decisions can be crucial. In the US, studies of the NBA between March 2015 to June 2018 viewed 26,822 plays from 1476 games. In those 4297 minutes of action, officials missed or incorrectly called 2197 plays, or about 8.2% of all calls reviewed. This amounted to 1.49% of wrong decisions in the final minutes of each close game.
Eliminating these errors might take some of the unpredictability out of being a sports fan, but introducing more certainty into umpiring will make for a fairer and more even competition. Of all the dangers of introducing AI, this would not appear to be one of them.
Lachlan Colquhoun is the Australia and New Zealand correspondent for CDOTrends and the NextGenConnectivity editor. He remains fascinated with how businesses reinvent themselves through digital technology to solve existing issues and change their entire business models. You can reach him at [email protected].
Image credit: iStockphoto/anton5146