VISION 2026: Innovation and the next era of investing - Neil Constable

Glen Davidson: [00:00:00] Welcome back from break. Neil, it's good to be here with you again. We've had a few of these chats with an audience like this. Quantitative Research and Investing, which is the division you're part of, is so interesting for me to hear about. I hope our audience feels the same way. I think you're still very interested in it. I'd love you to give the audience a reminder of what QRI is all about. You started with the group in 2020. When you think of it at Fidelity, we're made up of fundamental, technical and quantitative investing, bit of a Venn diagram. As a portfolio manager you can extract from each of those. But the Q part was quite small in the past. Not so since you started in 2020. Please take us through the evolution of QRI.

Neil Constable: [00:00:44] Before I do that, good to be here, good to see everyone. It's always good to back in Canada. Being based in Boston for the last 25 years it's nice to be on home ground, as it were. You're absolutely right. Fidelity has always, as you accurately point out, and it would be impossible not to, really, as an active manager to have some aspect of quantitative analysis, quantitative research working alongside the portfolio managers, fundamental analysts and so on. That's been true in equity, fixed income, the multi-asset class work we do as well. What's changed in the last, I guess five and a half years now since I joined Fidelity in 2020 to lead up this effort, it's basically gone from 50 or 55 quantitative analysts, some data science types that were spread across the different asset classes, it sort of started as a centralization effort, and now it is, I don't know the exact numbers but it's gotten to be about 220 quantitative researchers and portfolio managers and another dedicated staff of about 250 technologists, data scientists, software engineers and so on, and then a bunch of other operations, data quality type folks. So call it a 500-person organization. 10X in five years is no small feat. As you might guess, it's because it's a strategic priority for Fidelity to be a world leader in all things quantitative and sort of the big umbrella definition of quant, data science, quantitative analysis, risk management, trading. We're going to talk about what all those things are here now, I think.

Glen Davidson: [00:02:32] I would like to but I'd like to talk about the catalyst. I imagine Fidelity being private, and what Abby Johnson's vision was for looking towards the future of development of the firm, allowed for the group, you said 500 people now from 50 in 2020. Talk about the benefits of the private ownership and the allowance for you to build out a team the way it is.

Neil Constable: [00:02:55] I think the goal at the outset, as it was articulated to me, was never to be, okay, you know what, we haven't had a massive quant effort in the past, we need a massive quant effort. That wasn't how the conversation went at all. It was actually much more organic in the sense of, look, Fidelity's DNA is as a world class active manager across all the different asset classes. The questions being asked when I had my initial conversations with Abby herself and my boss now, Bart Grenier, who runs all of asset management, were more along the lines of how do we make sure that Fidelity is as relevant and as good as an active manager 20 years from now, that was five years ago, 20 years from then, as it has been for the last 30? I now know they were asking that question on a lot of different fronts. The answer to that is not just build a quant team, there's many other things you have to do. One of the big things that came out of the conversations I had with them was this growing need to incorporate the ability to onboard, analyze and make useful massive quantities of data that are being generated by every aspect of our lives today.

[00:04:09] I've talked at this event and others before so many of you have likely heard me talk about these. It's kind of unsettling, I suppose, but we can track peoples' cell phones around shopping centres. I mean, I don't do it, there's companies that do it. I don't have your name or anything like that, no one does. From the cell phone tower companies you know how many cell phones were going in and out of various retailers at any given time, and then you can get real-time data on Monday morning about which retailers did well on that weekend shopping session by geolocation, different cities, different malls and whatnot. In a world where that's possible, I'm not saying all that data is useful all the time, but in a word where that's possible it became clear that it needed to be imported and made available to our portfolio managers and our analysts so that they could use it when they needed it.

[00:04:57] If you think through all that, the people that you hire, just recency bias here, Darren was on the stage before me, he's got unparalleled, as do all the portfolio managers at Fidelity, corporate access, sitting across the table, interviewing CEOs and CFOs about their business models, their prospects for growth, the risks they see and so on and so forth. That is a skill that takes years to develop and train and Fidelity has always excelled at hiring people that are good at that and training them and making them world-class at that effort. Those are not the same people that you're going to hire to comb through terabytes of cell phone tower data to figure out the foot traffic in shopping centres.

[00:05:37] Just like I wouldn't hire me to interview a CEO and CFO but in the modern world you very much need both. That's how the conversation is going. What do we need to maintain our edge as an active manager? What does that look like in a world awash with data and technology? That precipitated this aha moment where they came back to me and said, we need to seriously invest in all things quantitative data science, the technology to go with it, the talent to go it because it's a different skill set. That's how it got started.

Glen Davidson: [00:06:08] It makes me think of a quote, you may dislike change but you'll dislike irrelevance more. It's about Fidelity having the edge and the portfolio managers and what you'll find, what our viewers will find today is that this discussion is about the assist that you provide portfolio managers but also the solutions that you provide for investors. We're going to go into two -pronged. I also wanted to touch on hiring because you talked about data scientists and so on. If I'm not mistaken, your PhD, you may have two but I can't remember, your PhD was in quantum field theory and its relation to gravity. That's something we've all got. What's fascinating is you made a pivot into financial services many years ago. Other people from your division that I've talked to have PhDs in mathematics, engineering, mathematical engineering, it's been amazing. How do you find the folks that want to move into financial services and how do you get them to understand that pivot with their skill set makes a lot of sense?

Neil Constable: [00:07:13] I'll try and give a brief answer to that. It's actually not as uncommon as many people probably assume, for the simple reason that the financial industry, not uniquely, but the financial industry is one of the places where there is massive quantities of data being generated at all times. There's markets open everywhere, prices are changing, all this geolocation data I just mentioned now, that's all now part of the mix. There's massive clients of data and there's no shortage of questions that need to be asked and answered to help people come up with mental frameworks to think about companies or industries or macro economies.

[00:07:50] The best way to answer questions, or one of the best ways to answer these types of questions, is to apply a data-driven approach to getting quantitative answers to things. There's qualitative aspects to it as well but the quantitative — when you think about physicists or mathematicians or engineers, what are we? Ignore quantum field theory and black holes, it's hard for me to do that because I love it, but either way we are, at the core, highly trained logical problem solvers. There is data, there is a mental model you have, there's equations, whatever, the variables. What do I know, what don't I know? What can I go get data about and what do I have to do some math to derive a result? That's a very generic thing to do and it pertains in the financial industry even more and more and more because more and more and more data is available to help you ask quantitative questions. A lot of the whole quant industry is full of people like myself.

Glen Davidson: [00:08:46] Amazing, and we'll talk about gravity another time. What differentiates, in a nutshell, Fidelity's QRI division from the competition?

Neil Constable: [00:08:59] There's four things to talk about. These four things don't make us different. I want to talk about how we're different on each of the four. It starts with data, unsurprisingly. I've said the word 15 times already, right? But data, technology, integration of quant and fundamental, which is part of the strategic aspect of fidelity which makes it different, and then the talent. In all these things I think we're different and, frankly, better, on all of these than almost any other shop in the world. What really makes it differentiated and, I think, durable and sustainable and how does it answer that 20-year question that was initially posed to me by Abby, when you bring all of my answers to all four of these things together.

[00:09:41] First on the data, all this data I talk about, it's fun to talk about cell phone data, it's fun to talk about credit card data, it's fun to talk the fact we have the contents of every container that's onloaded or offloaded at every container port in the world within seven days. We know all that, that's fun, but if we can get it so can others. Now, there's an inherent budgetary constraint to get all this, which we'll talk about in a second. We have dozens and dozens and dozens of really interesting data sets like that. You have to be able to not just collect that but build the technology to host it and make it usable. Like I said, in theory anybody else can get that but there's the aspect, what Fidelity brings to it, is our proprietary data.

[00:10:29] This is data that will never, ever leave the four walls of Fidelity. It amounts to things like ... in the equity division alone in Boston we have about 150 or 160 equity analysts who are meeting companies, writing research notes, putting ratings, strong buy, strong sell, EPS estimates, so on and so forth all the time. We have every note they've ever written, every rating they've ever made, every price target they've ever put on a stock going back 25 years. It's all digitized. We have every trade every portfolio manager ever made, time stamped when it landed on the desk, how long it took to execute, how much it cost to trade. We have every one of those going back at least 15 years. That's proprietary data and it's metadata, i.e. it's data about our own activities. It can be mined for all sorts of insights.

[00:11:21] When you combine it, what do our analysts say about oil companies when there's an inflationary spike, every time there's been an inflationary spike in the last 30 years what have they typically said about it? What are they saying now, how is that different from the street? Where's our edge in these things? Now what data can we get that everyone else has that we can apply to what our analysts care about? It's bringing those things together, this massive amount of publicly available data plus all this private data that really creates a differentiated edge for us.

[00:11:52] The last thing I wanted to mention on this whole point is the talent because that brings it all together. When I go out and start recruiting this team that you mentioned, yes, most of them have a PhD or something in engineering or something. That's great but why do they want to come work with me or work with Fidelity, more importantly? The answer is, by the way, I've recruited most of them from places like these hedge funds you've heard about, Citadel and D.E. Shaw and Point72, also our traditional asset management competitors, but why are they willing to come here because it's not traditionally been a destination. The answer every one of them gives me is that, well, there's never been a traditional asset manager that let me work on these problems and play with these toys, the degree of technological firepower that's been thrown at it.

[00:12:37] They're saying, well, I can solve the same problems, work on the same problems at the same calibre with the same calibre of people at Fidelity that I really had no choice but to go work at these hedge funds to do this before. Now we're doing it here in a much longer term investment-oriented firm, a more client-focused firm. For an awful lot of people who want to focus on long term investing with the client interests mostly in mind, that becomes a very valuable thing. Again, put all this together, the data, the tech stack that we're  building to host it all, the fact that it's deeply integrated across quant and fundamental, and then the fact that the talent wants to come work on it, that together makes us very differentiated.

Glen Davidson: [00:13:15] You talked earlier about understanding what's in every container on every ship in the entire world. That data is not available for me and everybody in the audience. I imagine that comes with a big ticket. Can you give us a sense of what we're talking?

Neil Constable: [00:13:31] This is something I very deliberately modelled after a couple of the hedge funds who I think have done, over the last 15 or 20 years done the best job at data acquisition. There's what we call regular data. This is the stuff you'd get from a Bloomberg terminal, numbers about companies, the usual kind of stuff, earnings calls and all that kind of stuff. Then there's what we call, I hate the phrase, alternative data which is shipping transaction data, geolocation data and so on and so forth. Every big quant shop, or even non-quant shop, who needs all the Bloomberg style data is going to be spending hundreds of millions of dollars a year on that data. We're doing that, we have to. In addition, for all the so-called alternative data, it's in the tens of millions of additional dollars per year.

[00:14:18] Some of these data sets that cover very narrow things like just credit card transactions across all the US, that's a $2 million a year data set. I subscribe right now, I think, to 45 — they're not all that expensive — but 45 of these data sets and we have to have a pipeline because there's always new data sets becoming available. We don't use the phrase anymore but the Internet of Things really meant that every single part of our lives is throwing off data all the time, not all it useful, of course, but the abilities of always be aware of what's available out there. So we have to build a pipeline. I have a team of people whose job it is just to evaluate what data's out there, what it's good for, if anything. If it is good for something how do you catalogue what it's good for, bring it in, ingest it, create analytics, and then deliver those analytics to portfolio managers so they can use them.

Glen Davidson: [00:15:08] Which takes me to the so what. IF somebody in the audience stood up, don't do this, stood up and shouted out, so what, it's really about an assist for the fundamental PM that needs additional information to have an edge but there's also about people that want to invest in specific products provided by QRI. Let's start with the assist it provides to a fundamental portfolio manager. Shilpa Mehra was here earlier and we were in the back talking afterwards and I said I was going to be talking to you. I don't know who knows, I said, you know who Neil is? She said, oh yeah, part of the quantitative group. I said, do you use the information provided or do you have the need to? She said, absolutely. It gives her an edge for confirming a thesis and so on. How do you convince or how do you market this to fundamental portfolio managers within the firm?

Neil Constable: [00:16:01] There's a lot of different things we do. There's a push and a pull aspect of it. One of the things that we've always done at Fidelity, this predates me, the quants that were there before me, we always call them embedded quants because the ones that worked with the fundamental teams sit with the fundamental teams, talk to the portfolio managers, talk to the analysts, are in the sort of information flow. What those quants end up doing is they learn the lingo. They learn what the portfolio managers care about, but they also speak quant. They translate that into a question that can be asked and answered via some of this data. If our quants are figuring out that the PMs are coalescing around certain types of questions then they can go away and figure out how to answer and go back and say, hey, by the way, it sounds like you care about X, here is a view on X based on these data sources. If it's useful that push then creates a natural pull. The problem managers are busy. If you're not providing them useful information they got a company meeting to go to. To the extent they start paying attention to it, and they are increasingly, it's because it's incremental value they can't easily get elsewhere.

[00:17:07] Examples always help with this so I'll give you two. One in our high yield bond group, I think I saw Adam Kramer here somewhere, one of his colleagues who runs a big high yield fund, Ben Harrison, there's a Ritz-Carlton Cruise Lines. I don't know if many of you knew that Ritz- Carlton, the hotel chain, started a luxury cruise line, but they did so and they built these very, very expensive multi-billion dollar boats. They borrowed a lot of money in the high yield bond market to do this. These bonds were trading at about 100, 102 cents on the dollar on the premise that they would have no problem paying this back, of course. They were going to sell these luxury giant yachts, in effect, and cruise around the world with billionaires or whatever they were going to do. The street bought this story. What we were able to do is figure out that the bookings were less than half what they'd advertised when they were raising this debt. You could verify this via credit card transaction. It was a big triangulation exercise. There's credit card stuff, there was actual geolocation data from some of the boats that had already sailed,  how many cell phones were on the boat, well, that doesn't make sense, collecting data from travel agencies. It wasn't any clean thing but you stitched it all together it's like, this is not a good story. The portfolio manager exited the entire position at 101 cents on the dollar and three weeks later it was trading at 75 cents. He gave me permission to tell that story.

[00:18:38] There's an example of collecting all this data, you get a leg up. That's a very dramatic example. Other examples would be something like — I'm not going to talk about our use of AI but our equity team, obviously, and you've heard from multiple PMs today about the generative AI technology and everything around it, how important that is to the stock market, investing opportunities. Is it a bubble, is it not a bubble? Is it going to change the world in the way people say, whatever. What we can do is ... we have every app in the world that's on your phone or on your computer or desktop, we get usage data from all this. We can monitor the ChatGPT apps versus usage of Gemini versus usage of the new Claude stuff and provide lots and lots of detailed feedback to the portfolio managers and the analysts who are covering this space, because they want to know who are going to be the winners, who are going to be the losers. Which customer segments are they winning with? Is Chat GPT doing better with the average user on the street and Gemini doing better with corporations, or vice versa. This is a rapidly moving space, what should they be paying attention to, what questions should they ask? When these are private companies that might IPO how interested should they be? There's no dramatic massive sell-off of the bond in this case but it really is part of the conversation on a daily basis and the data's being collected by that quant team.

Glen Davidson: [00:19:58] Those are great examples about how a portfolio manager benefited from your use. Actually, a question's just come in from the audience. How do you distinguish meaningful signals from noise across data sets?

Neil Constable: [00:20:12] This is what I was talking about, this big pipeline that we have to engineer. We have a whole team working on the data acquisition thing. It's not just about going out and finding the data sets. Actually, what it's about is first finding data sets that are out there. Sometimes big companies sell them, sometimes small startups have figured out how to get their hands on it and they sell it, whatever it is. The question then becomes what did that data forecast, if anything? Based on the source of the data you've got some hypotheses. Then we go and check. This team is checking to see if this foot traffic data is actually correlated with same-store sales in the lead-up to quarterly numbers or whatever it is. What's another good example, this app usage stuff that I just mentioned for, say, OpenAI's apps, is that actually indicative of what comes out publicly in terms of usage and whatnot a quarter or two later.

[00:21:11] We go through this testing period to see is it forecasting a KPI, a key performance indicator, for the company that anyone cares about. It happens quite frequently, you can forecast something about a company and no one cares. Like, yeah, great, so what? You can forecast that, it doesn't matter, the stock price isn't going to move so you don't want to buy that data set. There's a whole process of figuring out what [indecipherable]. That's iterative with our analysts, actually, because the analysts are the ones that are the best place to tell us what's going to move the needle for a company. We don't go say, hey, we've got this data set, it forecasts X, you should care. That's a waste of everyone's time. What we want to do is say, hey, we think we can forecast X, Y or Z, do you care about any of these things? Oh, my God, yes. If you can forecast Y for this company, absolutely I'd care about that. That helps narrow the types of data sets we care about. It's a very good question because none of this stuff is a smoking gun all the time. It's all just helping paint the picture for the PMs and the analysts.

Glen Davidson: [00:22:09] Have you had occasion where companies come to visit Fidelity to talk about their capital structure and so on and then say, so what do you guys know about us? Because they're aware of the research, or maybe they're not aware of the depth.

Neil Constable: [00:22:22] I doubt they're aware of that. I'm not actually in the company meetings.

Glen Davidson: [00:22:27] And it's proprietary so...

Neil Constable: [00:22:27] I'm not in the company meetings so I don't actually know.

Glen Davidson: [00:22:30] I wonder if that happens.

Neil Constable: [00:22:31] I do know our analysts will sometimes they'll ask the team, hey, I'm going to meet this company later this week, they've been making a big deal for the next couple quarters about this new product launch, do we have data that helps them figure out what questions they should ask in those meetings. That happens quite a lot.

Glen Davidson: [00:22:49] Very, very good. Now, QR in the QRI means quantitative research. The I means investing so let's talk about some investments because you're not only helping portfolio managers you've also got a team of portfolio managers who are running some solutions. Advanced US Equity Fund is one of them.

Neil Constable: [00:23:09] This is where, at least in the equity space and we should definitely talk about some fixed income stuff so I'll be brief, that's sort where it all sort of comes together. I mentioned earlier we have all this data from our analysts, from our portfolio managers, I've talked about the alternative data as well. Think of Advanced US equity as a situation where we've basically, we've been quants about it. We broke up the investment process into its independent components. There's data collection, which we've talked at length. There's idea generation, when you worry about time horizons on quarters and years fundamental analysts are the best place to figure out what companies are going to excel at, which companies have good products and all that. What we do in the Advanced US Equity thing is we actually take all the data from our analysts and our portfolio managers, all their price target changes, EPS estimate changes, they all manage portfolios against their coverage universe, paper portfolios that they're compensated on, we get all that data together, including the portfolio manager's positioning which is sort of longer term stuff, we can take all that and we can also say, okay, now our analysts, and the street as well, are writing notes about what's going to matter for the company in this quarter.

[00:24:20] Then we can build a quant model based on all the data I just mentioned. Well, same-store sales is all that matters, or inventory management, or supply chains is all that matters. Let's figure out which companies are going to do well on those metrics and not. You can, basically, put together the alternative data forecasting, sort of shorter term, call it the quarter, if you will, based on what everyone says matters, we can use the data to forecast that. We've got our analysts worrying about 12 to 24 months. We've got our PMs making three to five-year calls. We have all that information, we bring it all together and build a portfolio that's basically representing the best Fidelity thinks it can do with all that information. That's what lives inside the Advanced US Equity. What are the quants bringing to the table here? The idea generation's all coming from our fundamental investors for the most part. We're figuring out how to consolidate that down into some sort of best ideas, if you will, but also do the risk construction, portfolio management, trading, minimize trading costs aspects, things that quants are naturally suited for.

[00:25:14] It's very much of a division of labour, best of all worlds type product, Advanced US Equity is the product that we have available in Canada for that. It's something that we make available to the biggest institutional investors in the world too. Massive sovereign wealth funds are invested in a version of this product, large university endowments in the US. I actually just got a good email this morning saying an investment committee just approved an investment in it. This is really an institutional grade product that we're making available to anybody who's a Fidelity customer, $500 million sovereign wealth fund investment all the way down to a $50,000 retail investor. We want to make all this available and some of the scale that comes with the quant platform allows that.

Glen Davidson: [00:25:57] Wow, that's amazing. We just have a couple of minutes left but let's go to fixed income as you had asked. Let's go to Absolute Income Fund.

Neil Constable: [00:26:03] The fixed income markets, as everyone knows, are very, very different. The bond people in the room are going to kill me, but in some ways they're much, much, much more complicated than equities. In other ways it's much, much less sophisticated than equites, by which I mean because of the complexity in the bond markets everything's OTC, everything's hyper-fragmented, the trading is very illiquid. A lot of the applications of technology, particularly on trading, that revolutionized the equity markets 20, even 30 years ago has never really penetrated the bond markets. That's changing today very rapidly. In addition to all the information I just mentioned to help make investment decisions, that's all there but now you can do a lot more on the trading, the portfolio management, risk management, execution side of things. Ss that becomes possible what we started doing is running systematic bond strategies.

[00:26:53] The Absolute Income one is an example where we get to use quant tools to target levels of income that are commensurate with high yield indices but take the same risk profile as a more core investment strategy. For investors who want lots of income from the product but don't want to take the risk that usually comes with high yield products, or don't want the risk embedded in like some of our competitors products that use lots of derivatives in unclear ways, we can get this high level of income by tweaking how the portfolio construction works to get the right risk profile with the right level of income. That product is available in the Canadian market as well today and it's something we're really excited about.

Glen Davidson: [00:27:29] There's so much we could talk about but we're at time and we have Dan Dupont and Kat Black coming up next. Neil Constable, always fascinating to talk to you. Thank you very much.