Can Data Science predict the Premier League?

La Fosse’s Data & Analytics practice set out to answer just this.

Six Data Science teams from a range of leading companies gathered together on 7 September at Stamford Bridge Stadium to crunch through 3 years’ worth of English Premier League Data. We challenged them to predict a range of results, from final score down to number of yellow cards, and then asked them to put their money where their mouth was by betting 10,000 virtual dollars based on their predictions.

Predicting football results is a notoriously difficult task with ‘expert’ pundits pre-empting their predictions with enough get out clauses to make Apple’s tax lawyers pause for thought.

So how did our teams fair? Well not too badly as it turns out. Although final score predictions were only marginally more accurate than your ‘expert’ pundit, the teams hit their stride when it came to predicting corners, shots on target, number of fouls and yellow cards. With some favourable odds selected and bets placed, our teams are profitable by over $70k (virtual unfortunately)!

The true test of course is whether this is replicable, perhaps next time we’ll scrap the virtual dollars for something a bit more tangible.


​Data Science vs Data Engineering

Whilst all this was going on a group of leaders from the Data & Analytics space (Chief Data Officers, Heads of Data etc.) were gathered discussing the opportunities and challenges in working with such large quantities of unstructured data.

One analogy in particular stood out, in countries suffering from poor water infrastructure, 90% of the time is spent worrying about finding, then transporting and cleansing water, with only 10% of the time spent actually using the water.

In data terms, data engineering is needed first to identify, migrate and cleanse data before the data science begins. Our teams of data pundits were no different. Before they could get to the challenge of crunching numbers to make their predictions, our teams had to deal with very rich but complex datasets in a variety of formats.

This highlighted the importance of having a healthy mix of data engineering and data science skills in any team taking on such a challenge. Beyond this the role of data engineer, which typically steals little limelight on the Big Data stage, seems to be gaining fans and attention at a high level. What will this mean come the transfer window?

La Fosse advises and works with a variety of organisations in assisting them with their Data & Analytics talent acquisition needs. If you would like to find out more about or unique approach to market then please do get in touch.

Learn more about our data and analytics specialism.