Since he was profiled in 2014, Thesis Asset Management’s Matthew Hoggarth has been promoted to head of research and has embraced machine learning.
Back then he was a senior investment analyst, but he took on his new role following the retirement of another former Wealth Manager cover star, Michael Lally, last year.
He discusses how and why Thesis took the decision to centralise its fund research function and offers his views on why the growth of big data prompted the firm to look at the ways machine learning can be used to digest and interpret this.
How has your role changed in the years since your profile?
‘I was made head of research last year. It is similar to the responsibilities I had before, but we have centralised more of the research into Chichester.
‘So my team is now looking after fund selection which was previously based up in our Guildford office. It is good to have everything under one roof.
‘The change was partly for a bit more centralisation, but also our model portfolio service has been quite successful and Steven Richards, who manages that, was also doing the fund selection and both of those alongside each other were going to be quite a big load for his team.
‘Centralising the research frees Richards’ team up to concentrate on the MPS and puts all of the research in one place.
‘The centralisation we have been through, putting everything into a research team, allows the investment managers more time to focus on client relationships and the sort of things that really matter to clients. It also gives us the consistency and the ability to take a bit more of a data-driven view.’
How has the greater use of data changed your research process?
‘Taking a data-driven view is something we found very successful. You look at others in the industry it seems to be a trend that has become bigger over the last few years with more firms seeing the efficiency that you can get out of that.
‘Also, with the regulators increased focus on suitability it is important to be able to justify the investment decisions you are making and show they have gone through this sort of scrutiny rather than the expertise of the individual investment manager—sometimes you need a centralised process to make sure that everything is properly justified.
Has your investment approach changed at all?
‘Data is becoming a much bigger part of the story here too. Partly just the increased availability of data, but also the new ways to utilise it.
‘Our equity stock screen, which has done well, for UK stock selection is something we have expanded over time.
‘We launched an AIM [Alternative Investment Market] portfolio service in the autumn, which is based on the same test screen [as the UK stock selection] which is then applied to a universe of AIM stocks we have pre-screened.
‘Also my teams been looking at using more of a screening approach to asset allocation. It is a little bit more work, but it is potentially has a lot more input on the macroeconomic side for asset allocation decisions compared to just looking at equity fundamentals.
How do you tackle the sheer volume of data out there?
‘One of the guys on my team has been applying some machine learning techniques to that and has come up with some quite interesting results. We are currently testing it parallel to our existing process to see whether that might be a way we can improve things going forward.
‘Just the bigger volume of data we need to sift through means we need potentially cleverer ways of looking for the insights that that data can generate.
‘I think there is a big “buzz word” element around topics like AI [artificial intelligence] and machine learning, but I think applied in the right way it is definitely going to be a powerful tool for the future for the industry.
‘The other side of that though is the more data you have actually analysed, the more need there is for people, because the interpretation and presentation of the data becomes more important to show the insights.
‘This is pretty important at the moment with all of the extra data that is being reported to clients on costs and charges, for example.’