Could machine learning help players to bring down the house?

Could machine learning help players to bring down the house?

Thursday, January 18, 2018 Posted by Luke Massey
No alternative text provided

With sportsbook prices changing over 5 million times per day, it’s unsurprising to find online gambling operators leveraging machine learning as part of their calculating process. But what about the players? Could machine learning, with its ability to allow AIs to learn from and adapt to new data without human input, be useful to the gaming consumer?

In its newly published report, Playing a Smarter Game: How Machine Learning is Changing Online Gambling, data-driven payment solution provider Acapture puts ML and gaming under the microscope. And, amongst its findings, it offers some fascinating glimpses of how the process could alter the gaming experience. 

Sports predictions get a new level of accuracy

Using machine learning to better predict sports results has long been a point of fascination for hackers and data geeks. Back in 2016, a Hackernoon blog post entitled How to Predict the NBA with a Machine Learning System Written in Python, concluded, ‘If you build your own machine learning models you will find that you can correctly predict winners at a rate of around 70%,’ which, he proudly adds, is ‘better than ESPN experts and a lot of academic papers.’ At the World Cup two years previously, Google used data analytics to get 14 out of 16 match predictions correct. 

One company looking to corner the market in this field is Stratagem. A UK startup with the stated aim of ‘revolutionizing sports trading by fusing expert insights with sports technology’, it’s developing an AI that it hopes will be able to analyses data from sports events as they happen to make real-time predictions and offer instant, reliable betting advice. 

Its current version monitors football matches as they are played. Loaded with information from previous events, it can identify players, the ball, the kit colors and the pitch markings. As events happen on the pitch, it can measure how these deviate from previous data patterns and make potentially money-making judgments such as the time to the next goal or the final result itself. 

While that remains in development, it seems certain that the role machine learning plays in sports betting will only become more important in the coming years.

No alternative text provided

Machine learning wins big at the poker table

Held at Pittsburgh’s River Casino in January 2017, Brains Vs. AI: Upping the Ante was a poker tournament with a difference. It pitted famous players such as Jimmy Chou, Dong Kim and Daniel McAuley against Libratus, an AI developed by a team from Carnegie Mellon University.

Over 20 days and 120,000 hands of Heads Up Texas Hold’em, Libratus consistently defeated the pros, with its winnings totaling USD1.7 million.

While chess playing AIs have routinely been defeating human grandmasters for decades now, poker AIs have never had the same success, as poker requires a much more ‘human’ decision-making process. As well as the mathematical probability of winning with their current hand based on the limited shared information available, players also need to include factors like their opponent’s betting style and previous behavior to consistently make the right play. 

And it is machine learning that allows Libratus to think like a human. 

Libratus runs constant simulations in order to work out the best move in each situation. Included in its calculations are not only the odds of winning based on the visible cards but also the changing behavior of its opponents. Also, Libratus is smart enough to detect when an opponent has noticed a weakness in the AI’s own game and can adjust its strategy to defend itself. 

This raises the possibility of bots being programmed to defeat human players, which would massively unsettle the online poker landscape. 

These are just some of the ways in which machine learning is affecting your sector. Want a more in-depth look? Then download Acapture’s free report today. From improved consumer engagement to increased conversion rates and higher revenues, it covers all the possible outcomes of the application of machine learning techniques in the online gambling industry. 

The Acapture team will be at ICE 2018, with fraud expert Aldrin Mangalabal providing 1on1 demonstrations showing how their ACEprotect solution uses machine learning to fight fraud and improve conversion rates for online gambling operators. Want to get a head start on your competitors in 2018? Book your 1on1 demo with Aldrin here.

No alternative text provided

No alternative text provided

RISQ's Julian Borg-Barthet - The evolving jackpot landscape

No alternative text provided

QTech Games CEO Jonas Alm: The challenges and complexities of the Asian market

No alternative text provided

Number Crunching - John Asher, Cherry AB and GIVEMESPORT

No alternative text provided

Number Crunching - STS, DrückGlück and Etoro

Gaming Products & Services Directory

The essential directory for the gaming industry