Welcome to Part 3 of our ‘Ask the AI Expert’ series. If you missed them, click here to read Part 1 and click here for Part 2.
Artificial Intelligence (AI) and Machine Learning (ML) are hot topics at the moment. Not just in finance but across all areas of work, politics and society, they seem pivotal to our future.
We were keen to learn more about the opportunities that AI and ML will bring to the financial advice profession and the broader implications the application of this technology will have for our lives and society. I had the incredible opportunity to speak with Chris Cormack, Co-Founder and Managing Partner of The Quant Foundry, to explore the issue in more detail.
The Quant Foundry is a boutique consultancy offering bespoke quantitative solutions for all areas of financial risk including credit, market and operational risk. Chris Cormack is involved in all aspects of running the company and as a client engagement partner oversees multiple client engagements. Chris is also head of the methodology and models research team that builds both conventional stochastic models and Machine Learning and AI models.
Chris Cormack started his career as an academic, a lecturer in physics at Queen Mary University of London. He has an MA in Physics from Oxford, a PhD in Particle Physics from Liverpool and a Master’s in Mathematical Finance from Oxford. Chris has travelled the world doing research, including California, CERN in Geneva, and Japan. In 2004, he co-founded The Quant Foundry.
Part 3: Risk, Regulation and Integrated Services
AV: Looking at financial planning and wealth management, AI is already having an impact on asset allocation, portfolio modelling and risk profiling. Some of the smarter stuff now is around nudging people in terms of their financial behaviour. What sort of things do you see developing in wealth management going forward?
CC: It’s an interesting space. One of the opportunities is access to new ideas or increased understanding of risk profiles. And it’s the idea of ‘am I missing out on something?’ and ‘does my wealth adviser have access to the depth of information to advise me on opportunities?’. It is also being able to ask the right questions about things and get a response and be able to keep up with that curve in terms of structures and opportunities.
I think another of the big areas is the improvement of that engagement exercise. Whether it is through improving an understanding of a client’s risk profile or also helping them to understand their risk preferences better in terms of a series of questions. Then being able to match similarities in terms of clients’ risk preferences in different asset classes with products under different market circumstances and being able to enhance that overlap between clients.
For example, it’s noticing that client 1 is a little bit like client 7 in some respects, and then letting the client engage a bit more with some of the other products out there in terms of different risk profiles and performance aspects to find different recommendations for them and think a bit more about what they really want out of the portfolio in terms of risk and returns profile. It’s that refinement and enhancement in process in wealth management.
AV: Another element that’s really interesting is the assumption – I’ve probably made it – that a lot of this stuff is going to enable the service provider to provide better services. But the bit we haven’t focused on is the ability for the consumer to ask more searching questions and hold their adviser or wealth manager to account better.
CC: Exactly, that’s the good thing about AI – its ability to assimilate considerable amounts of information. The early adopters classified risk profiles of products or markets or segments and then were able to say that they have this data and collect this information. With the right assistance, they were able to have better conversations with clients. It could highlight areas that may be outside the expertise of certain wealth managers in terms of opportunities, but the data had been collected and shared so that information was available to them which is an enhancement of what they would usually be able to provide, if you like. This enables a continuity of knowledge in an organisation too.
AV: Looking at compliance, one of the challenges we have in financial planning is around regulation. You mentioned reduced risk profiling when we get older, but there is a lot of data that says actually clients should increase their risk profile as they get older because they need the higher return from equities to make sure their portfolio lasts as long as possible. That is based on data, and a machine learning tool could make that work, but if I said to the regulator ‘this is my client, they are 80 years old and I have put them 100% in small cap US stocks because that’s what they need to make sure they are going to be ok’ I would probably get my licence taken away from me! This will have some really broad implications for our regulators in terms of how they engage with it and take it seriously.
Regulators are taking advancements in AI in finance very seriously. If a model is making a recommendation, it’s vital that we are able to understand what that recommendation is based on.
CC: Yes, and the regulators are taking it very seriously. If a model is making a recommendation, what is that recommendation based on? It’s about fairness and transparency, which will always crop up whether in the human domain or in the machine learning domain. As long as you can explain why you made your recommendation then that is the first step in building trust in the regulatory space.
Another side to the regulation question is that the regulators are in quite a unique position in many respects, as they can see a huge amount of data across the whole industry. If you take the banking sector for example, the regulators are very keen to get involved in helping control money laundering which is a big growth area for the application of big data. They are actively seeking out a consortium to start pooling the data and enhancing the oversight. They are one of the big adopters of big data and machine learning and will be looking to make sure that people start integrating across these regimes in the coming decade or so.
AV: Would you go as far as to say that the responsibility is with the data owner to use that data in a beneficial way and if they don’t, they are doing everyone a disservice?
Those who are late to adopt AI will miss out on huge market share and opportunities in the next five years or so.
CC: Yes, and the demonstration of what deep learning can provide with the increased data shows that not only does it work, and not only does it enhance the service the more data that you have, but it’s economically beneficial. Those who are late adopters will miss out on a potentially huge amount of market share and opportunities in the next five years or so.
AV: Generally speaking, what you are your thoughts on how AI and ML will affect the markets as a whole? If we look at evidence-based investing, do you feel that this is going to lead to higher market volatility, or do you think it is going to stabilise it?
CC: There are two aspects to the application of ML and AI, especially in terms of their influence on the markets. One thing that people know a little bit about is the concept of what’s happened with flash crashes driven by a lot of the strap option traders. That is something that will be endemic and there will be these events driven by machine learning algorithms working out that there is an opportunity and everyone else jumping on that same opportunity. Anyone that we have talked to with experience of this knows that you end up with these very crowded trade positions and liquidity gaps, and they just happen at higher speed. That is a concern. The only way to deal with that is to regulate at the exchanges, which is something that’s already happened with random time ordering.
AI will lead to far greater transparency in businesses, which will have a huge impact for ESG investing.
The other area in terms of investing is around ESG (Environmental, Social and Corporate Governance) investment, which has come to the fore and grown in popularity recently. A lot of people are trying to build a better picture of which companies really are governed well, with a social conscience, and really do have concerns about the environment rather than those that just say they do.
At the moment that is something which is a little bit fluffy in terms of information content and impact, but in an era of machine learning and big data, we are able to make statements that are quantifiable and definitive. Rather than producing fluffy reports where you’ve hired a novelist to write for you, people don’t have time to wade through that. There’s a big push to try to standardise data formats to make them more objective and more machine-interpretable. In a way, that could help that investment landscape by identifying those who do genuinely want to generate some value from an ESG perspective, rather than the green-washing we have all heard of. This is something that could be very positive to come out of machine learning – the ability for people to make quantifiable statements and judgments on these topics.
AV: One of the easy throwaway lines in financial planning, or something that we advisers like to use as a comfort blanket, is that the machine will never do emotions and so financial planners like us will always be valued. However, if computers are starting to be creative (as discussed in part 1), should we be anticipating a future where machines can have intelligent conversations about emotion and pick up on feelings, and so forth?
CC: I think it’s already here in a way. If we look at natural language processing, the text interpreters that we have are able to pick up on sentiment and can tell if it is a positive statement or negative, happy or unhappy. They are able to interpret text and various tools can pick up on emotional responses in faces and recognise happy faces, smiling faces, sad, confused etc. Tech already exists to interpret emotions and there is some very interesting AI out there already that can do this considerably faster than the best human beings.
It’s only a matter of time before people start having interactive conversations with robots that are emotionally enriching and responsive to our emotions.
It’s applied across a range of different areas. We’ve been talking to a machine learning scientist who’s been developing interactive billboards (for those familiar with the film Minority Report). They already exist in this world, it’s already here. The world is changing rapidly and it’s only a matter of time before people start having interactive conversations with robots that are emotionally enriching and responsive to our emotions. I think it’s just a couple of years away from that becoming mainstream. We’re not sure how good they will be yet, but it’s likely they’ll be very convincing.
AV: The point is that this technology is here, and it is only going to get better. Earlier you mentioned some of the AI that wealth management teams were using to support interaction with clients or new clients – could you give some examples of things that they are doing?
CC: Most are around enhancing customer communications, by which I mean having and generating more meaningful conversations with customers. Customers are not being spammed or cold-called for no reason. A lot of investment, certainly in the financial space, has been around better engagement with clients – more targeted and meaningful conversations and marketing campaigns. People have spent lots of money on understanding their client base better in terms of the segmentation of their client base, the demographics, the likes and dislikes, and also the types of products they have engaged with successfully.
A lot of investment in AI, certainly in the financial space, has been around better engagement with clients – more targeted and meaningful conversations and marketing campaigns.
It’s about a better understanding of the client, especially on the retail side where mass marketing campaigns can become a bit more targeted. Also, on the wealth side where you can really begin to tailor solutions and identify classifications around risk preferences, product preferences, how people would like to engage with their portfolio or their private wealth manager, for example.
There are some really compelling indications that people investing in this technology are developing higher success rates in terms of campaigns, and clients are getting the advice that they believe is pertinent to them, certainly from the feedback to these processes. It’s something that’s been happening for a while in the industry. People are familiar with Amazon and Netflix recommendations; essentially, it’s an evolution in that space in terms of learning. Specifically, it’s tailoring that solution to the domain of asset management, and for this it requires a combination of skills. It has shown a lot of growth and interest and there are strong indications of its success, whether in retail or the more refined private wealth space.
AV: We have become much better at giving clients advice. Thinking about your portfolio – what are the things you need to do? We know about keeping down costs and the right asset allocation. We know about good financial behaviour and populating your portfolio with equities. That’s how to be successful. An issue around a lot of financial planning is that it is subjective. For example, is it better based on your health and life expectancy to buy an annuity or to take a draw down? There’s no reason why the model couldn’t evolve to give you a number and a firmer indication. ‘Based on everything we know about your health, postcode, income and so on, we believe that the annuity would be the best option rather than taking the draw-down’. We would be giving advice based on the data, rather than what the client is telling the planner.
CC: I think with enough client information about characteristics and taking the knowledge you have built over the years, you can start building these more refined pictures of how people’s preferences change or what the best advice is over time, for example, from more risky advice in your younger years to more stable advice as you get older. As you build this picture across your thousands of clients, you will be able to make better decisions. Your machine learning recommendation tool will give you better and better recommendations over time and that’s the great thing about it. The beauty of the latest tech in AI is that they just get more and more refined, and you make better decisions the more data you collect. Which is more than most human beings can store in their head.
This document is marketing material for a retail audience and does not constitute advice or recommendations. Past performance is not a guide to future performance and may not be repeated. The value of investments and the income from them may go down as well as up and investors may not get back the amount originally invested.