The Euro 2024 international football tournament gets underway today, and we're delighted to report that AI has finally been turned into something useful in the form of a virtual pundit for sports fans.

Want to know how tonight's game between Scotland and Germany will go? Bad news for fans north of the border: Luzmo's AI Octopus reckons there's a 76 percent chance it'll end up being two nil to Germany despite a "spirited" Scottish attack and a possible streaker incident.

According to Luzmo's CTO, Haroen Vermylen, the company used something similar to a Monte Carlo model and simulated the Euros a million times. In 22 out of 100 simulations, France won the tournament. Luzmo's native Belgium currently has a 3.4 percent chance of lifting the trophy, while England stands at 3.2 percent, although there is a good possibility the team will make it to the semi-finals.

The octopus is a sly nod to Paul the Octopus, who predicted the results of Germany's Euro 2008 matches and the 2010 FIFA World Cup. Although a cynic might note that similar success could be had with the toss of a coin, the Cephalopod's unnervingly accurate predictions lacked the detail afforded by squirting data into an LLM.

As for how it works, Vermylen told The Register that "The model itself is essentially a match simulator that does a play-by-play simulation of games based on things like offensive ratings, of team's defensive ratings … to essentially make a probability distribution of how many potential targets and expected goals a certain team would be able to achieve versus another team."

Readers of a certain age might remember Kevin Toms' Football Manager on the trusty ZX Spectrum. We certainly do.

The models will be refined as time passes and scores are added. However, at present, it looks very much like a three-horse race between France, Portugal, and Croatia.

It's a neat demo of Luzmo's cloud-first data analytics platform, which pulls data from various sources (online or via a scheduled synchronization) to provide customer insights. As is often the case, AI pixie dust has been sprinkled over the product. According to CEO Karel Callens, the wonder tech has been used for "opening up an audience that previously didn't have access ... because of not knowing what the data is behind it, but still knowing their business."

In other words, it's using an AI platform to come up with new insights without needing to be a data scientist.

The company faces stiff competition - the analytics platform Tableau springs effortlessly to mind. Of his customers, Callens said, "A number of them actually moved away from Tableau. I think the most important thing to note there is that we're purposely built for embedded analytics."

"We're a building block that you can embed in just a couple of days. You don't have to host or maintain any servers or keep them up to date - that's done for you ... You can connect APIs, not just your databases, we don't need to know who your customers are, we can work with your own security system.

"So all of these tiny but very important pieces make it night and day between just internal reporting and better client-facing analytics."

As for the predictor itself, Callens confessed to being an international football fan and used match data to demonstrate analytics dashboards. "I always use the data of my own indoor soccer team," he said.

As for who might win it, Callens said, "It's hard to say Belgium because we had a hard loss against France. I'm hoping that if we confidently win our group, that these chances will grow significantly over the tournament." ®

https://www.theregister.com//2024/06/14/ai_octopus_predicts_euro_2024/