Lies, damned lies and xG
Why improved data analytics will generate higher profits for operators …if they can find a way to unlock it.
The 2024/25 Premier League is upon us, and we’re about to find out how Chelsea fare in the Champions League, whether Everton can repeat last season’s feat of qualifying for European competition and if Man Utd’s flirt with relegation was a blip or a trend.
Confused? Not if you reach for the 23/24 league table that’s based on points gained from Expected Goals (xG) rather than points gained on the pitch. On that basis, Chelsea and Everton would be 15 and 16 points better off and finished 4 and 6th respectively (before deductions) with Man Utd a whopping 20 points worse off in 17th place.
The xG figure is now a staple football statistic. To recap, xG is a measure of shot/chance quality and how likely that shot is to produce a goal when compared with many other shots taken from that position on the pitch – a shot with an xG of 0.4 being a 40% chance of scoring, for example.
While xG is a useful metric for deciphering team and player performance, the above examples show some of the problems associated with relying too heavily on historical data when the context of that data is not fully understood.
Similarly, while statistical outputs such as xG can give us an interpretation of past behaviour they don’t always give a better indication of what is coming next. Using historical data to model predictions of future performance is an imperfect and extremely difficult proposition.
Well, what did you expect?
To a greater or lesser degree, the major B2B sports betting suppliers use some form of pricing and trading automation, incorporating algorithms based on modelling the same type of historical data that produces the xG metric, such as shot quality for example. For these major suppliers, this type of predictive tooling represents the frontier of their current algorithmic development.
Undoubtedly, these types of “maths models” have facilitated the growth of live betting over the last 15 years; but they, and the type of analytical process that underpin them, are ill-equipped to support the industry moving forward.
The crippling cost of operations – for official data rights in particular – increasingly mean the revenues generated from these types of models are not high enough to pay for the data that is required to drive them, increasingly a “Catch 22” situation for licensed operators in regulated markets.
So, what’s the solution for sportsbooks looking to compete in this high-cost environment while improving the profitability of their operations and distinguishing themselves from the competition?
It's all in the maths
Fortunately, the answer lies in the use of data that every sportsbook already possesses but doesn’t currently use effectively. Unfortunately, it is also a more complex set of data to contextualise and a much harder proposition to model. It is also one that the current supply chain isn’t able to utilise because it can’t be incorporated into their existing set of simplified algorithmic maths models.
What we are referring to is the live integration of customer sharpness analytics into the underlying pricing, trading and risk management function. Modelling customers according to their betting behaviour and then incorporating the real-time information they are giving you through aspects of their betting patterns – what they are betting on, when they are betting on it, what has happened immediately before and after they place their bets for example. This enables operators to build a unique picture of how their customer base is trading against them, and how their prices should be optimised according to the type and amount of risk they are generating.
Moving forward with confidence
The good news is that understanding customer behaviour in this way allows operators to change prices confidently based on information they can understand, in line with their own risk profiles. It also enables a very granular understanding of where underlying models (or trading teams) may be weak or strong, which in turn ripples through to increased confidence in generating improved turnover, as well as differentiating effectively for marketing campaigns for example.
By looking beyond the current sectoral norms for pricing and trading that are being driven by a duopolistic supply chain, operators can embrace the types of innovations that have underpinned market-leading growth in areas such as exchange market making and financial trading. All operators have rich seams of liability and customer data that can unlock market-leading returns, improve turnover and enable brand differentiation, and by combining the best maths models with both real-time liability and customer behaviour data they can reinvigorate their bottom line.
