On Wednesday, January 24, we went to a Stats Perform Opta office in London, UK. The office was located just off the canal near Paddington station. The company was founded in 1996 to analyze Premier League football matches and was contracted by Sky Sports for their television broadcasts of the 1996–97 season. The following season, Opta became the official statistics provider for the league itself and became sponsored by Carling. Now the company records performance stat data from a variety of different leagues and teams while compiling them in an easy to read manner to sell to any company that wants them. Many football metrics used in national broadcasts, team facilities, and player analyses are compiled by Opta.
Upon arrival we had a great introduction to the company by some of its employees. We spoke with Mike Morrison, the head of marketing for the London branch who explained the services and products that the company provides for its clients. Next we spoke with Danny Dinsdale, a data scientist who works with the raw numbers and derives programs in order to work through the vast amount of data quickly. He explained that the data is compiled mainly by humans but also by AI (player and event recognition). It is then brought to them as raw data, then they use their programs to make that data meaningful. Lastly the data is given to a data analyst like Johnny Whitmore, who gives these random numbers meaning for the company that contracted them. This could be a graphic for a news report, in match fan graphic, a short sheet for an announcer, the list goes on.
Throughout their presentation Mike, Danny, and Jonny gave various interesting insights within their business. One interesting point they made was that the majority stats are still manually imputed by stat keepers onsight on any given match day. These stats are then stored into their humongous database of football stats kept from 1976 when Opta was created. Some may assume that there has been some sort of automation to keeping these stats but this is not the case here. The stat keepers do get help from AI though as it is used to give more precise and detailed statistics at much faster rates on any given play. For example, AI may be used to record the type of pass being made or the distance covered by any given pass. It also allows statititions to make predictive models for games so betting platforms have accurate and timely information for people to bet on. AI technology within football and sports in general is still growing though as there are still many areas that stat keepers still have trouble with even with the help of AI. Some stats such as clears are still difficult to accurately record due to the arbitrary nature of the play. These were great points made in a great discussion facilitated by the Stats Perform Opta team.
Our group found that the most valuable connection to apply to our future endeavors at coaching came towards the end of the presentation— in a somewhat unexpected fashion. Jonny was describing the difficulties in marketing their products, especially to the prototypical “Brexit” coaches that are frequently seen across the British game. Specifically, he mentioned a coach who was incredibly skeptical about adopting statistical analyses into their club’s decision-making process. This skepticism about xG was slowly dissolved as he physically explained the concept to the coach by having him take ten shots from outside of the box. This story was part of a larger argument about the four different types of learning: visual, auditory, kinesthetic, and reading/writing. What we took away from this portion of the presentation was that we all learn with a different combination of these educational typologies. Both as coaches and just as individuals operating together in the world, it is most productive to attempt to meet people where they are. In this way, we can learn from each other and educate those we are tasked with developing much better. Instead of attempting to fit every square peg into a round hole, we must genuinely listen to the person across from us to best adjust to their learning styles, growing ourselves as coaches as we do so.
By Oryon, Dash, Jack, and Bem (January 24)