Neudata

Dealing with alt data: what can finance learn from Formula 1?

McLaren is a racing team that has long been defined by its approach to data analysis, so it was a crowd-puller when the company’s Head of Modelling and Data Sciences Dr Karl Surmacz presented at the Neudata London Summit 2019 https://www.neudata.co/events/london-conference-2019

Predictive analytics have been used in the sport since the turn of this century, enabling technicians to make strategic decisions on vehicle and driver performance during a race, across the season and multiple seasons. Dr Surmacz told delegates: “Getting insight out of streams of data is now second nature to us… our competitive edge depends on it.”

But selecting what out of a mass of data will produce key insights without overloading the car can prove the real issue for McLaren’s race team. Analysts and engineers have to walk a fine line between micro-data that can produce almost limitless insights into the performance of vehicle and driver, and simply having overwhelming amounts of information that are expensive to generate and time-consuming to analyse. It's a situation that is familiar to those working in the world of financial data analysis.

“Data is a first-class citizen. We treat our models the same way – they generate data. We look after them. The teams that perform best are the ones that have got this down to a T,” Dr Surmacz explained. “But we also have to understand the uncertainty associated with the data we gather. There is a lot of room for error.”

Working out what data will provide real value to the racing team and what is simply TMI (Too Much Information) is key, as is working out what external information and analysis can give McLaren the edge. Alt data for this industry includes understanding the behaviour of both the team’s own cars and that of their competitors.

“We do everything we can to try and get every snippet of data we can,” said Dr Surmacz. “Everything is happening on the edge.”

Similar applies to the world of financial alt data: “Choose the right data sources, look after your models and understand your risk: F1 best practices have much in common with asset management.”

 

Date

7 Nov 2019