Racing predictions made by computer science experts at the University of Lincoln and featured on BBC Radio 4 successfully beat the bookies, picking out the programme’s first winner in 15 weeks.
Distinguished Professor of Image Engineering, Professor Nigel Allinson, and Principal Lecturer in the School of Computer Science, Dr John Murray, worked together to predict the winners of two horse races in Pontefract and Doncaster on Thursday 17th September.
Their prediction for the 2:10pm at Pontefract, a two year old horse called Spike, came in as the winner. The race at Doncaster was called off due to bad weather, but BBC producers were quick to congratulate Professor Allinson and Dr Murray on their success.
The academics’ method was based on a research paper by Professor Allinson on Machine Learning and Neural Networks. Dr Murray gathered race data and using the algorithm from Professor Allinson’s paper they ‘trained’ a neural network to ‘learn’ patterns from previous racing data, such as combinations of jockeys, trainers and horses as well as ‘going’, odds and value of race, to establish the best combinations.
Dr Murray entered the data from the relevant race card into their system the night before the race, and the system ‘predicted’ the winner of the race, which was broadcast on BBC Radio 4.
Dr Murray said: “I’m not a gambling man so I didn’t have a flutter, however a number of my colleagues in the university have stopped me today to thank me for the tip for which they did place a bet and won some money. We could produce more tips and try to improve the system, but the one problem we have is the availability of the data. It is ‘out there’ but it is very expensive.”
The feature was part of the BBC’s ‘Intelligent Machines’ series, which explored the ways in which advancements in technology and robotics have changed our world.
The team’s success was picked up this week by The Racing Post, which featured their winning methods as part of the paper’s news round-up this week.