A COMPUTERISED tipster that regularly outperforms its human counterparts has been invented by an Australian scientist.

The software, which was developed using Australian rugby league results, predicts the result correctly up to 66 per cent of the time, using artificial intelligence to “learn” how likely a team is to win.

Alan McCabe, of James Cook University, Queensland, used neural network technology originally designed to recognise handwriting to develop the McCabe Artificially Intelligent Tipster, or Mait, New Scientist magazine reports.

A neural network is a computer system that can learn to recognise the combinations of inputs that produce a particular output. Each input factor can be weighted, so that the most important carry more influence on the program’s result.

Instead of feeding in information on pen strokes and the shape of letters, Dr McCabe used the system to study factors such as team performance, and points scored and conceded. Over the course of a rugby league season, Mait performed 25 per cent better than human tipsters, even though it does not consider factors such as the availability of star players for particular games.

“A conscious effort was made to ensure that no subjective features were included,” Dr McCabe said. He added that the system should work just as well for other sports, such as football and cricket.

Explore UK News

• Crime News

• Education News

• Health News

• Science News

• Scotland News

Times Recommends

• Swine flu vaccine for all as drug rushed through
• Artificial sperm and the race to build a baby
• BBC jets out stars to tout for prizes