Building the Next Generation of Sports Betting

Jeevan, where is the real competitive battleground now – offering more markets and combinations, better pricing, easier UX or making BetBuilders more relevant to individual players?

The market has matured and we are now seeing the first and, arguably, second phase of BetBuilder move aside for the latest iterations. The key differentiator, now that the tech and the pricing is solid, is UX and the ease at which bets can be discovered and placed.

Discovery is a critical element to a client placing a wager and returning, or being so daunted that they leave and do not come back.

The latest generation of BetBuilders have the capabilities to price a vast plethora of markets that have limited appeal as singles and can cause overload on a front end; navigating around them and finding interesting parlays can be overwhelming. Nailing an intuitive bet discovery function is, for me, where the next stage of the journey needs to head.

With rapidly evolving LLMs, perhaps the next stage should be punters giving voice or text demands and prompts and letting the algorithm source a relevant BetBuilder. During the SBC Lisbon event, I listened to a presentation by AlgoSport around exactly this. They are building a product they call AlgoSport Zero, which is designed to “supercharge sportsbook navigation to events and markets” and “create bespoke markets on the fly.” This is all done using the latest LLM tech and their black box BetBuilder solution.

As BetBuilders become more complex, how difficult is it to price correlations accurately while still giving players an attractive proposition?

Provided you have reliable core and prop market inputs then the actual calculation engines aren’t overly complex for those with such skills. In fact, there are many more market types that could be offered but which aren’t to avoid clogging the UI. This is where the idea of using an LLM is interesting; helping to unlock more focused parlays that punters simply don’t have the patience to source via navigation alone.
The real challenge, as I see it, is how to better manage any correlated risk that this next, open-world, stage of BetBuilders bring.

During the 2026 World Cup we saw exactly how risky correlated parlays can be for operators. Just two heavily recreational parlays wiped out player prop profits for the entire month of June.

How important is live BetBuilder to the next stage of the product, and what additional demands does it place on data, modelling and trading technology?

Live is critical to any BetBuilder and gone are the days where a pre-match only offer would suffice. That said, we still see much higher volumes pre-match, but this does disservice to the inherent value in being able to offer a cash-out function.

Live pricing is much more complicated, not necessarily mathematically, but from a data source(s) point-of-view and trying to map a number of different providers. A (preferably) stadium-based scout feed needs to align with your core market pricing and your off-tube player stats feed. Scout feeds will ensure you are suspended swiftly but won’t provide the rich data that is required to understand which player has, for example, taken a shot.

Trying to sync all these external data sources into a free-flowing product with high uptime and readily available cash-out is a challenge.

Where does BetBuilder go next? Are we heading towards a sportsbook where almost any combination a customer can imagine can be priced instantly?

If you believe the AlgoSport guys, then yes, this is exactly where we are headed. I’m still unsure as to what cohort that approach appeals best to though. I’m a big fan of being presented with a narrative for a match along with a relevant BetBuilder to capture that narration. bet365 has added “steam indicators” to its pre-canned builders, allowing bettors to see the most popular bets.

Looking at the Prediction Market scene (hard to miss it) we have seen huge growth in Request for Quote (RFQ). This is allowing traders (not bettors…) to come up with complex or unique parlays that market makers are racing each other to quote.

In order to price these accurately the market makers also need solid core and prop inputs and accurate algorithmic solutions. Sportsbook BetBuilders currently offer a nicer UI, but RFQs are essentially allowing users to request prices for whatever they might have in their head; similar to the LLM approach we have spoken about.