In an interview with European Gaming, OpticOdds founder and Gambling.com Group SVP of Sports Data Services Matt Restivo explained what watching live odds from nearly 300 books at once reveals before any single operator can see it, where AI betting agents are actually heading, and why the World Cup final is the one match where automation still needs a human in the room.

Key takeaways

• On the market-wide vantage point: ‘When nearly 300 books are adjusting to the same underlying event at once, the signal in how they diverge is often more valuable than the consensus line itself.’
• On autonomous betting agents: ‘AI agents that advise and surface opportunities will arrive faster than most people expect. It’s already happening.’
• On keeping a human in the loop: ‘It isn’t about second-guessing the model, it’s about being in the room if you need to make a call that the model isn’t designed to make.’
• On World Cup model risk: ‘When you’re aggregating live signals across nearly 300 books in real time, the market itself tells you when it’s uncertain, and we surface that signal instantly.’
• On the durable edge: ‘The companies that win over the next five years will be the ones whose data is easiest for AI systems to actually use.’

Why the whole market sees more than any single book 

European Gaming (EG): You watch live odds from 200+ books at once, a view no single operator has. Going into the World Cup, what does that vantage point show you that the books themselves can’t see?

Matt Restivo: Еvery operator manages their own exposure, and we supplement that by analysing the entire market simultaneously. This allows us to see consensus form in real time, identify outliers before they become liabilities, and detect uncertainty before any single book recognises it.

During the World Cup, this becomes even more critical. Football markets, especially in the group stage, are thin and reactive.

When nearly 300 books are adjusting to the same underlying event at once, the signal in how they diverge is often more valuable than the consensus line itself.

We enable operators to move beyond a single-book perspective and develop a deeper, more comprehensive view of the market.

How close we are to AI agents that bet on their own 

EG: You’ve built a connector that lets AI models pull your live odds directly. Honestly, where does this go? Are we heading toward AI agents that price, advise or place bets on a World Cup match with no human in the loop, and what breaks first?

Restivo: AI agents that advise and surface opportunities will arrive faster than most people expect. It’s already happening. What we shipped with Anthropic in May and with Perplexity Computer in June represents an early version of it. 

Fully autonomous agents placing bets on something like a World Cup final with no human review is a more complex question.

‘The first constraint will be the technology, and following that it will be the trust. ‘

The one World Cup moment automation can’t handle alone 

EG: Copilot automates pricing, risk and settlement. Would you tell an operator to run it untouched through a World Cup final, or is there a moment in the biggest game on earth where a human still has to have a hand on it?

Restivo: I’d tell them to trust the automation and stay close to it. Those are two distinct things.

The automation does its job, that’s what it’s built for. However, a World Cup final is also the one event where a VAR decision or an injury in extra time creates the kind of volume spike your risk team should be watching in real time, not reviewing after the fact.

Keeping a human in the loop isn’t about second-guessing the model, it’s about being in the room if you need to make a call that the model isn’t designed to make.

Pricing lopsided group games when the data is thin 

EG: US books will be drowning in World Cup money on a sport with thin data behind it, including lopsided group games no model has really seen. Is the data good enough yet to price those matches well, or is there still genuine model risk there?

Restivo: Football does have real historical data limitations, but the operators running our infrastructure aren’t relying solely on a pre-match model to carry the full weight. 

When you’re aggregating live signals across nearly 300 books in real time, the market itself tells you when it’s uncertain, and we surface that signal instantly. 

‘Operators on our platform can see how the entire global market is pricing a lopsided group game, adjust in real time, and identify moments when positions are shifting fastest before they become liabilities. ‘

The data risk in the World Cup isn’t evenly distributed. That’s the structural advantage of leveraging a market-wide feed instead of a single or even multi-sourced feed, and it’s why we built OpticOdds.

Where the edge goes once fast feeds are everywhere 

EG: You’re now part of Gambling.com Group and surrounded by Sportradar, Genius and the rest. If everyone has fast feeds and similar AI, what’s actually left as your edge once the data itself is a commodity?

Restivo: Speed and coverage become table stakes fast. The real edge shifts to how data is structured and delivered; that’s where we’ve focused. 

‘Providing an operator’s AI model with clean, normalised, instantly queryable odds from nearly 300 books through a single connection is more than a data advantage and becomes an infrastructure advantage.’

And infrastructure advantages tend to be more durable.

The companies that win over the next five years will be the ones whose data is easiest for AI systems to actually use.

About OpticOdds

OpticOdds is a sports betting data provider that aggregates real-time odds and market signals from hundreds of sportsbooks worldwide, delivering normalised feeds and risk tooling to operators, prediction markets, market makers and media companies. Founded in 2023, the company was acquired by Gambling.com Group (Nasdaq: GAMB) on 1 January 2025 as part of Odds Holdings, alongside sister brand OddsJam, and now operates as the group’s sports data services arm.

The post OpticOdds’s Matt Restivo on World Cup betting: ‘Trust the automation’ appeared first on European Gaming Industry News.