What if IBM Watson Gave Racing Tips? Would Ascot Become a Carpark?

Wed, 06 Dec, 2017

By Will Mace & Michael Franklin

Artificial intelligence is the technology du jour, and while its current applications in the gambling industry are well known, it is a slightly different application that could provide a major, even existential, shock.

As if the current political and media pressure wasn’t threat enough, the gambling industry is coming under even more significant threat from a far less obvious source. So how does IBM Watson giving racing tips lead to Ascot becoming a carpark?  Dramatic it may seem, but in the context of the rapid growth of artificial intelligence (AI), some are predicting that a major shock to the betting industry is on the cards, with an equally significant shock for all the sports that rely upon it.

Artificial intelligence is of course the technology du jour and is becoming widely used to disrupt and reinvent whole industries – from healthcare to humanitarian aid, from retail to recruitment. Its current applications in the gambling industry are also well known; from Netflix style sportsbook recommendations to the identification of problem gamblers. It is a slightly different application however that could provide a major, even existential, shock.

The data explosion

Data has long been the lifeblood of sports bettors and bookmakers alike. Every score, every shot and every pass has been recorded and used to create what were once considered quite sophisticated predictive models. There is however a rapidly increasing wealth of new data, or at least newly processable data, that is fuelling something of a revolution.  A few hundred data points in the predictive models of old are rapidly becoming a few million data points in the AI powered predictive models of today. These new sources of data, and the techniques to process them, allow – for the first time – an algorithm to really read a game, to follow the ebb and flow and to identify shifts in momentum.

One very small example is that we can now – remotely and in real time – measure variables such as a player’s pace, acceleration, and heart rate, identifying any signs of fatigue and accurately predicting the impact it will have on the rest of the game. Millions of data points, per game, per fight or per race can be collected in real time, related to each other, related to known outcomes in every previous game, fight or race, and used to build highly accurate real time predictive models – based on a previously impossible number of correlated data points.

Ok, but where is the shock? How does this exponential improvement of predictive capability provide the shock to the gambling industry?  Models have always existed, albeit usually the domain of fairly secretive betting syndicates, so how does ‘simply’ improving them, even exponentially, really change an industry?

The shock comes when the predictions become freely and easily available to all, when IBM Watson, or Google, or Facebook, or Amazon, or any of a new breed of AI-first companies, decide there is value in openly sharing their predictions – in the way that newspapers and blogs have always done. What better way for Facebook to support their increasing sports focused content and streaming with a value add service such as AI powered predictions? Why would Google not want to add value to a search for the weekends fixture list with an AI-powered outcome predictor?

The logical demise of an industry? 

So. AI-powered predictive models based on millions of data points will become freely and easily available to all, creating a situation of absolute information parity between bookies and punters.  Punters will have immediate access to better information than ever before and so will make better bets – or at least make bets on the outcomes the predictive models are forecasting. In effect, every punter is now a wise guy. 

Punters will start to win more often, as their bets are based on data rather than flawed and limited human opinion. But bookies will have access to these prediction services too – so prices will shorten to reflect them, and shorten again as punter money floods in on those outcomes. Prices will quickly become highly unattractive and punters will, by and large, stop placing bets.

What? Would they? Why?

This is the crux of the hypothesis and rests upon an understanding of the different motivations for betting. Wiseguys, deprived of any pretense of information superiority and faced with perfectly efficient markets, will stop betting as the value opportunities disappear. The self-proclaimed sports expert – who bets in order to demonstrate their superior insight and expertise – will also stop betting, knowing they are not and cannot be more insightful than the Google predictor engine in the long run. As well, those who bet to add a little spice to a match they are watching; for there is very little spice in simply betting on a Watson prediction at incredibly short odds and they would be surely crazy to bet against Watson.

Professor Ian McHale of the University of Liverpool, one of the creators of the Sports Analytics Machine, said that: “the gambling industry has already seen a convergence of odds and mirroring of prices. If more accurate probabilities were made available and information parity existed across bookies and punters, this situation could only intensify, and make it very difficult for professional bettors to make a living as the odds offered would be completely efficient. On the other hand, there will always be casual bettors willing to bet, just like there are always people playing lotteries, despite knowing the probability of success, and the odds offered perfectly”.

So perhaps there is hope for the betting industry after all? There will always be the unfathomable decision or the lucky bounce that can turn all predictions on their head. Some punters would certainly choose to continue to bet – perhaps the Leicester City fan who believes their team can repeat the famous Premier League win against the odds, perhaps the romantic who likes the colour of the silk, or perhaps the diehard, betting on the ‘Hail Mary’ pass actually coming off.

But could such bettors be anything like enough to sustain the industry as we know it today – an industry that has grown used to double-digit year-on-year growth?  These punters will be few and far between, as in the long run, betting against the data, against the always learning, always improving AI-powered predictive models, is only ever going to be a very minority sport; the realm of the reckless and foolhardy.

But what about the carpark?

Other than the odd Leicester City fan, much of the current volume in sports betting could dry up as many of the motivations for betting are rendered obsolete. Turnover and revenues will implode, betting shops will begin to close, and online operators will shrink. High profile shirt sponsorships will be scaled back and it will start to be safe to watch sports on TV again, as the wall-to-wall gambling ad coverage will simply become unaffordable for a much-reduced industry.

The most significant impact, however, would be felt in the sports that are dependent on betting for their very existence. The most obvious example is Racing, which, without betting, would very quickly cease to exist.  The smaller tracks would close first, followed quickly by the more iconic tracks such as Ascot, Cheltenham, and Aintree. Ascot, in particular, sitting in prime commuter belt territory, would be sold off to the highest bidder, much of it becoming a carpark for a Crossrail branch line.

AI – good or bad?

On a macro level, the debate – being driven by philosophers such as Max Tegmark and Calum Chace amongst others – is very much open as to whether AI will be very good for humanity or very bad for humanity. The general consensus is that it could be either, depending on choices we make now and in the coming years.

On a more micro level, there are certainly some very exciting applications of AI in the gambling industry, but there are almost always unforeseen consequences to big new capabilities such as this and they are absolutely worth exploring and debating now, as we have tried to do here. The logic expounded here is very much up for discussion – we have certainly made a few big leaps – and actually, we have no idea if Ascot is in a good location to become a Crossrail carpark or not but hopefully the point remains. many potential impacts are unknown and some of them could significantly change our industry.

I don’t suppose too many people will shed a tear for the gambling industry, after all, many industries change or even die as technologies advance. But to potentially lose the sport of kings is rather more significant – and would be deeply sad.