TXODDS goes all in on data visualisation

David Reed, director of research and editor-in-chief, DataIQ

Since 2014, TXODDS has been offering real-time betting odds to a client base around the globe via streaming APIs, In the same time, it has built a comprehensive archive of sports betting data. Which means that, if one of its betting business operators wants to understand whether a pattern of play is unusual, it has a useful point of reference.

TXODDS sports betting global heat mapDelivering that sort of insight is no simple matter, however. “Traditionally, we have called, processed, shifted and stored data which our customers have used for risk management and customer management,” explained CEO of TXODDS, Einar Knobel, in an interview with DataIQ. “But the data volumes have grown which makes it harder to get a view in real-time.”

The core Oracle and Hadoop environments support analytical queries by experts with programming skills. But Knobel realised that big data risked become as much an obstacle to business-critical insights as it is a support for real-time odds adjustment. As he explained: “How do you explain to the CEO, product manager or trading manager who doesn’t have the time to code?”

The solution has been to go all in with a TIBCO implementation, expanding on the existing real-time StreamBase analytics with the vendor’s Spotfire data visualisation. “It lets our customers get that overview on what is happening in parallel with their expert users,” he said.

Online betting is a 24/7 global industry where operators need to keep a very close eye on a number of factors that represent both opportunities and threats to their business. One is in-live betting which is the biggest slice of the action. “If a client is offering in-live odds on 10,000 football markets, but its rival is offering twice that, why can’t it do the same? Is it about the prices? We allow them to compare themselves to the competition to find out,” said Knobel.

“If you are out of line with the market, you are giving away free money.”

“For the CEO, the challenge is to take something you don’t know about your own company, like whether your offer is competitive, and make those critical decisions,” he said. There is also a space in which customers may arbitrage high and low odds, thereby making money out of misaligned pricing. “If you are out of line with the market, you are giving away free money. So you need to track your prices against the market in real-time.”

To help with this process, the company is pre-packing a range of reports in Spotfire which can be accessed off the shelf. Since all the raw data remains intact, expert users can still build out bespoke views. 

TXODDS started life 15 years ago as an idea Knobel had with a co-founder who had the coding chops to get the database up and running. API-based from the outset, it is now in the enviable position of ingesting data streams from worldwide betting platforms and then repurposing it and selling it back to those same operators.

Fraud detection is proving to be a major growth area on top of the core use areas. “A client can look at volatility in a market compared to historical fluctuations in betting activity. That could be down to differences in price or something more suspicious,” said Knobel.

“If punters suspect a game is bent, they won’t watch it.”

With the phenomenal number of sports activities globally - TXODDS holds data on 1,500 football leagues alone - combined with the range of prices and markets being made, betting operators face a major challenge in checking that all games are fairly contested and that odds are not being manipulated for criminal ends. Typically, this tends to happen in Asian markets and around lower league matches.

“If the operator sees the prices going down on a team in-running, they can trigger a flag. They used to spend a lot of money doing traditional fraud detection without having an insight into whether a movement was suspicious. We have come up with mathematical models using factors like the time and the number of goals or cards that looks at deviation from the standard,” he said.

The outcome of that insight might be to trigger a fraud investigation with all payouts suspended, or it may show that a particular league or fixture is always unpredictable or prices volatile because volumes are low. It is this type of careful policing of sports betting that is critical to the future of the growing sector, as Knobel points out: “If punters suspect a game is bent, they won’t watch it or bet on it and that destroys the whole sport.”

Knowledge and strategy director, DataIQ
David is developing the framework for soft skills and career development among data and analytics practitioners. He continues to be editor-in-chief and research director for DataIQ.