Role profile: Working in machine learning and AI
Mark Ainsworth, Head of Data Insights, Schroders
Mark joined Schroders in 2014 where he has formed the Data Insights Unit, a team of over 20 data scientists. The team focuses on using big data, analytics and machine learning for investment and research, to enhance and complement the existing skills of our fund managers and analysts, and to give Schroders an ‘information edge’ over competitors.
After his degree in experimental psychology from Oxford and masters in operational research, in 1999 he was recruited by the McLaren F1 team as a Race Strategy Analyst. In 2002, he moved to Tesco, initially reporting on trade performance and analysing customer spending behaviour and then heading up the analytics team in the site location planning function. In 2007, he became Chief Technology Officer of Talent Innovations, a start-up selling online 360 degree feedback software. In 2012, he joined Telefonica Digital, as Head of Analytics of Smart Steps, their big data monetisation division.
What is your current role and what are your main responsibilities?
I head up a team of 20+ data scientists who provide insights from various sources of data to help our colleagues in investment make better decisions. In particular, we build tools that provide behavioural science insights to portfolio managers about their trading decisions, and we harness alternative data (big data sets too big or messy for investment professionals to be able to analyse using Excel) to give insights into the companies they’re investing in.
As you can imagine, many investment firms use data to improve their short-term trading, such as predicting companies’ quarterly results so they can trade around earnings surprises. However, Schroders’ investment teams are much more focused on the long-term so my team are focused on insights that give a deeper understanding of those companies and how they might perform in future years.
What do you enjoy the most about your role?
There’s an amazing amount of variety - our job is to do analysis of any dataset that might help analysts and PMs understand the things affecting the companies we invest in, which means pretty much anything in the whole world! Luckily my own background is almost as varied - I’ve worked as a data scientist in other industries including airlines, telecoms, retail and even motorsport - and this comes in surprisingly useful. Often we’re doing the kind of analysis happening inside the companies we invest in - for example we do analysis of store locations and population data to forecast store roll-outs or identify good potential merger pairings. This is exactly the kind of thing I was responsible for when I worked at Tesco - this is data analysis that the board of the company use to ‘steer the ship’, so can be just as useful for an investor to judge whether they think the ship is going in the right direction!
What are the most important skills for success in this role?
At the heart of the role is expertise in data science, in particular skill in mathematics and statistics (in order to draw valid conclusions from data) combined with the ability to fully harness the power of modern computers and so the ability to think algorithmically and write code. However, we deliver value by helping investment professionals make better decisions - this is ‘human in the loop’ data science. That’s why soft skills - the ability to build relationships, ask the right questions, and clearly communicate your answers - are the most important skills. Unless it’s genuinely useful and makes sense to another person, it’s not an insight, it’s just some data.
How did you get to this position and what would you advise CFA charterholders who would like to do a similar role?
Follow your passions - do what you love and keep growing yourself.
No-one should emulate my route to my current role, it’s very random! Joining Schroders was pure chance really - there was an initiative to create a new data science function in Schroders and I was lucky enough to be persuaded to apply for it by a friend who was involved. Up until then I’d never really considered working in financial services. I’ve always done whatever role sounds interesting and which grows me as an insight professional. For example, when I had the chance to move from British Airways to become the Race Strategy Analyst and McLaren F1, I leaped at the chance. It was an amazing industry to work in and it’s very satisfying to see the visual displays that I created almost 20 years ago still visible on the pit wall on TV. However, after I’d been there three years there was nowhere for my career to go - McLaren is an engineering organisation, and I’m not an engineer. So I went looking and ended up at Tesco for 4 years, a company I absolutely loved, and where I learned lots of very different things, many of which are useful to me now and which I wouldn’t have learned doing one job in one industry.
By always doing a job that grew me as an insight professional, it set me up brilliantly for the role I’m in now. So if your passion is data science, find opportunities to do it in projects at work or in your spare time, and keep your eye out for a role that can take you forward.
How has the role changes over the past 1/3/5/10 years?
Some people joke that ‘data scientist’ is just a ‘statistician with a Mac’. There’s a bit of truth to that - lots of people who might previously have called themselves a statistician are now called a data scientist, and a lot of them do use a Mac! However, there is a real change too, in the use of software tools and coding. You simply can’t analyse the huge datasets available these days without learning a proper programming language like Python or R.
Another big change in the Investment industry specifically is the presence of data scientists along with the adoption of alternative data. Teams like mine just didn’t exist in the buy-side five years ago, but there are now hundreds of firms (mostly hedge funds) with data specialists doing this kind of thing.
How do you see your role changing in the next ten years?
Fundamentally, I think the future looks like investment teams consisting of a mix of investment professionals and data professionals. Being a good investor requires effective use of modern tools, whether that’s behavioural science, big data or artificial intelligence. It’s simply not possible to learn all the traditional CFA skills of company valuation and portfolio construction AND data science skills like data wrangling, machine learning and writing code. It make much more sense to have a clear division of labour, working as a team. As well as having a central team, we are increasingly embedding data professionals in investment teams, who sit with them and tailor our tools to that team’s processes. So my role will increasingly be focused on managing this blended structure, where their day-to-day priorities are guided by the local team but their long-term career goal is to get my job one day!