Applying AI to Harness racing, Said Aspen from Visigon were able to predict the outcome of a horse harness race with approximately 37% predictability. Based on Swedish statistics going back to 1995 – 37% was also the chance of the favourite horse winning the race. So, from the very beginning of the project, the goal was to at least be on par with these odds.
Said Aspen is one of the founders of Visigon and has worked there for the last 10 years. Said has a technical background – he is educated in Computer Science and Engineering and has over 13 years of experience in IT, software development and Capital Markets.
“Prior to this project, I didn’t have any actual experience working with AI and Machine Learning. While I didn’t have much experience, I was curious to learn more about the concepts – so I began to educate myself through online courses and research.” – Said Aspen, Co-founder & Consultant, Visigon
The way Said really began to work with AI and Machine Learning was by applying the concepts to Harness racing.
Harness racing is one of the largest sports in Sweden and Finland. It is a kind of horse racing, yet different to regular horse racing. In harness racing, the driver does not sit on top of the horse. Instead, the driver sits on a cart which is attached to the horse.
“Harness racing was the perfect way for me to try and apply AI and Machine Learning, because every detail about every race was recorded and accessible. Everything you could ever want to know about the horses, drivers, trainers, tracks, weather conditions, and records is available online – it’s heaven for a data scientist.”
One of the most important things to keep in mind when working with AI and Machine learning is the data. Not only the accessibility, but also the accuracy of the data. In this case, having access to this amount of data was the perfect satiation to try and apply AI and Machine Learning.
“Even though Harness racing wasn’t directly related to the work I do at Visigon, the technology is similar in any field, and we were already seeing that technology being discussed on the client side. Beyond that, the concept of betting on horse races and trades in the Capital Market can also be compared, so when I spoke on this with my colleagues – people could relate.”
For Said, the first step was of course to begin gathering the data and from there on it was all about trying. One of the things he quickly discovered was that one could not expect to avoid failing when working with AI and Machine learning, without almost any experience. Said found that when he ran into a problem, he had to learn it – it was a do it before learning it type of process.
“I believe that generally with undiscovered technologies, one of the biggest challenges is that there is no framework. There is no set of rules of how use the technology. No defined strategy. This is both a blessing and a curse. For one it does not make for the same set of boundaries, but it does make for a challenge.”
Another challenge when working with AI and Machine learning is putting it to use. During the experience Said has had as a company advisor for several Master Thesis students, one of the things he has discovered is that AI is often only considered a solution to grand problems, and not minor issues – where it would be easy to implement and see results.
“One of my foundings, working on this project, was that it was actually quite easy to do. Considering how little experience I had prior to the project, I believe the results of only a few months effort, were significant.”
This is something Said is also seeing in the Capital Market industry. The goal should not always be to solve the biggest problem, or all of the problems at the same time – using AI technology. If someone decided to start implementing AI and Machine learning in order to solve smaller problems – great results would show.
“While my Harness racing and AI adventure was never actually put into use, it taught me a great deal. It taught me that even though AI and Machine learning seems extremely complicated and seems like it requires a lot of software developing and statistics skills – it is not that difficult.”
AI and Machine learning is open for exploration. With increasing capabilities for machine learning algorithms, and a constant pressure for higher efficiency, financial institutions are demanding to leverage their platforms with AI and machine learning technologies – Visigon are ready to meet this demand.