Trading Insights: Data in the discretionary investment process, with AKO Capital’s CEO

[Music]

Patrick Hargreaves: Generally a company that is good quality doesn't immediately go bad overnight. It tends to be a gradual process of erosion. If a company is pulling accounting levers in a way that it hasn't done historically, that is not a good signal, if we have a poor company meeting followed by a poor market research survey, followed by poor behavioral analysis, that is a much, much more powerful signal for us than one of those in isolation.

Eloise Goulder: Hi, I'm Eloise Goulder, head of the Data Assets and Alpha Group here at J.P. Morgan and today I'm delighted to be sitting here with Patrick Hargreaves, who is CEO of AKO Capital and PM of the AKO Capital Global Funds. AKO Capital being a London-based investment management firm specializing in long-term investments and quality listed companies. So Patrick, thank you so much for joining us here today.

Patrick Hargreaves: Thanks Eloise, it's great to be here.

Eloise Goulder: So Patrick, could you start by introducing yourself and your background?

Patrick Hargreaves: Of course. I am, as you said, CEO of AKO Capital. I've been here for the past 15 years. I joined from Goldman Sachs, where I worked for about nine years. And then I had, prior to that, stints at both Casanova and Pricewaterhouse, where I qualified as a Chartered Accountant. And my degree, maybe a bit incongruously, was in English literature before I started my career in finance.

Eloise Goulder: And turning to AKO Capital, can you describe what your investment philosophy really is?

Patrick Hargreaves: AKO Capital is a bottom-up quality investment fund. We have been around for 20 years. By quality, I mean companies that can generate and sustain supernormally high returns on invested capital over many years. And we try and align ourselves with those companies for the long term, by which we mean at least three years, but often much longer.

Eloise Goulder: And why do you believe that investing in quality companies is so important in terms of alpha generation?

Patrick Hargreaves: The starting point is that we care about industry structure. We like industries that are either stable duopolies, stable oligopolies, or even what we call mini monopolies, where a specific situation enables a company to have an insurmountable, insuperable position in industry. We look for those industries and then we look for companies within those industries who demonstrate what we call  patterns of quality. We wrote a book nearly a decade ago now which sets those out. There are 12 of them. Some of them are really obvious like pricing power, super important at a time when inflation is rising. Companies tend to have pricing power because they have super valuable products where they can price up almost whatever the environment. There are some more esoteric patterns. There's one that we call the friendly middleman. That's a qualified or a trusted intermediary that is selling you a product or recommending a product to a consumer. Generally what they care about is efficacy, efficiency. They're not so bothered about price. What we see in those situations often is good pricing power and high margins for those companies. We look for these different patterns. Great companies have typically more than one of them, so they tend to be overlapping. And then we align ourselves with these companies for the long term. And by long term we typically mean at least three years, but in practice it tends to be much longer than that. So many of the companies we own, we've owned for seven, eight, ten years plus. And the old Buffett aphorism of our favorite holding period being forever applies to us as well. The reason that long term nature of holding is important is that in any given year, you don't necessarily get paid out for earnings growth. So somewhere between 70 and 80 percent of a share price move in any one-year period can be explained away by just whether the valuation has gone up or down. And that's something we just don't feel we have the ability to call on a short-term basis. By the time you get to three years, 70 to 80 percent of that share price move is dictated by whether earnings have grown. So the earnings compounding effect over that period. And the further you go out, the more earnings compounding is critical. At the heart of our investing philosophy is the idea that if you can find companies which compound earnings at a stable rate over time and align yourselves with those companies for the long term, you ultimately get paid out for that as an investor.

Eloise Goulder: That's so interesting, the fact that you've got such a long-term investing horizon and the fact that, as you say, the valuation moves in the shorter term is perhaps where you don't feel you have the edge. But if you have the edge on forecasting the earnings growth, then that plays out in the longer term and you want to capture that over the longer term. Presumably the challenge is where a quality, high margin business starts to lose those characteristics.

Patrick Hargreaves: That is something we watch for very closely. No industry is immutable. What we're looking for are industry structures where we believe that there is a lower risk of change, of dislocation. That is critical for a couple of reasons. If you have an industry with high competition, high propensity to alter over time, the probability of permanent capital erosion or losses is much, much higher. So avoiding those industries is critical. Likewise, if you have a company where these moats, these competitive advantages, are starting to degrade, that is a critical point for us. We have a system that we call the boiling frog catcher. It's based after the apocryphal story. If you put a frog in cold water, slowly boil it, the frog doesn't jump out and it dies. It's actually not true, it does jump out. But the point is that we have a system that tries to capture these things where a hitherto high-quality company is seeing erosion around these markers of quality. And so there we absolutely would change our mind if those factors are altering.

Eloise Goulder: So in terms of identifying these quality companies in practice, what do you do? How do you go about that process?

Patrick Hargreaves: The identification of new ideas can happen in a number of different ways. We run screens, for factors like high return on invested capital, which is in our mind the best financial shorthand for quality. Clearly that in itself is not particularly a value added process. Lots of people can do that. The value lies in us trying to understand the reasons for that return on invested capital and understanding why it might be persistent in the future. So that's the work the analyst team are doing. But one of the things that I think differentiates AKO is some of these what we call our specialist teams that we've built up over the years. These are teams of experts in their field that help us to assess investment hypotheses or investment questions from lots of different perspectives. We have five of these teams. The first of them is market research. We have over 20 people speaking nearly 20 languages. And what they will do is go out and survey companies that we are investing in or considering investing in. Some of those might be surveys of competitors, of peers, suppliers, of customers, of franchisees. That is a team that we've had in place for over 15 years and super valuable, particularly on the long side. The second oldest team we have in place is our team of forensic accountants. They have two primary functions for us. The first is acting as a safety net on all of our longs. Before we invest and after every one of our longs reports, we will get a report back from our forensic accounting team, which rates the accounts from one to five, one being clean, prudent accounting, five being legal but highly aggressive accounting techniques, which we think flatter numbers and which we'll need to reverse at some unknowable points in the future. That's a hugely valuable tool for us in terms of risk management. Third team we have is the behavioral analysis unit whose expertise is in language. All of the people in that team come from either a psychology or linguistics background and their job is primarily to observe interactions between ourselves and management teams and also to analyze transcripts. And what they're looking for are indications of discomfort, of avoidance, of obfuscation in the responses of management teams to the questions that we're asking or the questions that are being asked on the conference call. And again, like the forensic accountants, they will rate an interaction on a one to five scale, one being clean, direct responses to questions and four or five being responses from a linguistic perspective that are full of those indicators of avoidance or obfuscation that I've just talked about. And that for us is another big red flag that we look at. So useful on the longs and the shorts. Our fourth and fifth teams are the more recent teams that we have, but still we've had those in place for many years as well. Data science, which has a huge remit ranging from simple automation of pricing trackers to running algorithms on different price prediction tools that we have. And sustainability, which is our newest team and is trying really to look at companies through the lens of all of the stakeholders that are affected by it. So what are the impacts? Is this company generating positive or negative externalities in the long term? Can we dimension those in any meaningful way? And could it have a material impact on the business's financials over the long term?

Eloise Goulder: Absolutely fascinating to learn about these five different distinct groups, all of whom are focusing on a specific function and all of whom presumably working with the analysts. Do you believe all of these functions have quite unique edge versus what's going on in the wider investing community? And can you speak to the power of one versus another?

Patrick Hargreaves: Over the years, we've trialed lots of different things, some of which have worked well, basically the five teams that we have in place, and some of which have worked less well. We had a project with academia many years ago where we were trying to assess and measure corporate culture from the outside. It looked very interesting initially. As we developed that program, it became clear that it wasn't going to be a meaningful signal for us. So we've only invested behind the teams that have added value over the time and we measure their input closely to make sure that that is the case. But different teams have different value at different stages for different companies. What I mean by that in practice is if we have a company that has been a long idea for us for many years and the accountants give it a clean bill of health, that really for us is an absence of a negative. It's not a meaningful signal if the accountants give that company a poor rating, that is an immediate red flag. We would probably at least sell half of that position and the likelihood is we would exit entirely because we know the odds are meaningfully stacked against us. So that immediately becomes the most important data point that we have. The same for market research. There are some companies where a non-material change in a data series is not going to cause us to do anything with the position. But if we see a clear erosion of market share, if we see a new competitor arising that we hadn't previously considered, if we see a material degradation in pricing that we hadn't forecast, that then could become a very meaningful drive of what we decide to do with the stock.

Eloise Goulder: So there are points in time where these teams are not necessarily coming out with a critical signal for the stock, but there are points where they are and it's at those points that you're really going to pay attention to those respective teams.

Patrick Hargreaves: It's often the combination that is important. When I talked about our boiling frog catcher system, what I can see there is chronologically every data point that is generated by all of our teams. What it shows is that generally a company that is good quality doesn't immediately go bad overnight. It tends to be a gradual process of erosion. If a company is pulling accounting levers in a way that it hasn't done historically, that is not a good signal. It tends to suggest there's something else going on under the hood that we're not aware of. And so when you get an aggregation of negative data points, if we have a poor company meeting followed by a poor market research survey, followed by poor behavioral analysis, that is a much, much more powerful signal for us than one of those in isolation.

Eloise Goulder: And in terms of understanding the efficacy of these different teams on a standalone basis, do you measure the sort of implied alpha of each of these teams?

Patrick Hargreaves: Absolutely. So obviously, as you said at the outset, we're a fundamental bottom-up organization. But of course, we use data. So what do we track? We track things like every time we trade a stock as a PM, we have to annotate with a reason code. Over time, we track those reason codes, what are good, what are bad. One very obvious one, for many years, we're selling down a position for valuation reasons. For 15 years, that was a horrendous reason to trade. Consistently, we lost money by not running our winners for long enough. Then in 2022, it flipped. Clearly, rates rising, different environments. So the core message within that is that we use data, we try to learn from it, but we try not to over-learn from it. And there's an important overlay of rationality and sense to any lessons you're learning, and I think understanding the reasons why an outcome is what it is are as important as the data itself. But for the specialist teams in particular, we track everything they do. So every time we get a positive or a negative read from any of those teams, we'll track the stock price performance following that over a number of different time horizons. And we will then aggregate those up. What we can see over time is that had we traded behind every single positive market research, every single positive behavioral and accounting, those positive scores would have outperformed and the negative ones would have underperformed. We're not a quant fund, but what we're trying to do is use all the data from these different processes to stack every decision in our favor or to understand where we're running risk. The measurement is to try and work out how strong a signal each of these things are, what's potentially signal versus what's potentially noise.

Eloise Goulder: And on the accounting side, why don't more investors do this, do you think? Is it extremely difficult work to identify these accounting issues?

Patrick Hargreaves: We're seeing more people using forensic accounting tools. We're seeing more people analyse language. We're seeing more within the industry do data science. Our view is that it takes a long time to get these things right. So we have had a behavioral team in place now for nearly 15 years. What they do has evolved over time. The same with accounting. We can assess what works, what doesn't work, what accounting concerns are most likely to yield a good result for us on the short side, or most likely to yield a problem for us on the long side. We can break the data down again, without overlearning from it. Some of the things that have worked for us really well over time are things like percentage of completion accounting concerns. If a company is aggressively recognizing revenue on a percentage of completion basis, that is a huge red flag for us. And we would tend to be very, very careful about those. If there is a big inventory cycle coming, well, there are different fundamental reasons why that might be good and that might be bad. It's understanding those nuances that is key.

Eloise Goulder: Well, all of this data and all of this learning, it creates such an incredible feedback loop, to the analysts and to the portfolio managers at the end. And it's so interesting hearing you talk about the trading signals and the fact that you want to learn, but you don't want to overlearn. And I guess you're marrying two things, because on the one hand, there are classic behavioral biases which probably always exist and people are always fallible to, maybe like taking profits too early, which you were referencing earlier. On the other hand, there are sort of long-duration cycles, like the fact that we've gone through a growth versus value boom over the last 15 years or so. And you need your analysts to remain aware of that, that that won't last forever, albeit it might last for quite a long time. So fascinating to me that you should be running all of this data, and ultimately providing such a rich feedback loop to your analysts.

Patrick Hargreaves: The feedback loop is super important, but the analyst is the expert. So if I talk to somebody within the behavioral team, what they're telling me is detail on the language specifically. They are not stock experts. It's really the analyst's job to understand, to weigh up these inputs, to calibrate signal versus noise, and to determine the ranking importance of these different inputs. And when I talked before about the relative importance of some of these inputs, there are times when these are a simple non-event, and there are other times when these are critical and suddenly we need to change our perspective very meaningfully. And the analyst is the person who's best placed to determine that.

Eloise Goulder: Well, let's turn to the analyst then. We've spoken so much about the processes and the input teams that you have, but where does the value of the analyst lie? And perhaps I can turn one of our age-old questions on its head. We typically ask our investing community, what's more important, the data or the model? But I think in your case, a more interesting question is what's more important, the process or the human? And can you also speak to how the human and how the analyst really optimizes all of the information that's available to them and can deliver the alpha?

Patrick Hargreaves: What we are trying to achieve at AKO is a process that is replicable, that is to some extent industrializable. It's something that should be followable by an intelligent, trained person. But there is always an element of judgment to everything. When I talked before about not overlearning from data, that is where the analysts' knowledge, their pattern matching from history will come to the fore. And I still firmly believe that that human ability to determine these changes, the old Mark Twain adage about history not repeating but rhyming, I think that is where an analyst can really add value to these processes. So distinguishing between signal and noise, if we're looking over a 10-year time horizon, we have to be very mindful that we're not being swung around by short-term noise. A short decline in terms of demand for a company that we're very confident will deliver good double-digit growth over the long term is not something that should be aerating us too much. But if there's some material change in a competitive dynamic, well, that requires a different mindset.

Eloise Goulder: So if you had to choose one or the other process versus human, where would you go, Patrick?

Patrick Hargreaves: I would choose both! I think you have to have the combination. And I think one without the other doesn't work as well. The process is there to stop us making big mistakes. So you alluded to biases earlier on. Biases is something we care a lot about. We spend a lot of time trying to identify biases, trying to mitigate them. And all of these specialist teams are to some extent there to try and prevent us falling into these bias traps. I might think a company is wonderful. But if our accounting team comes and tells me, you know what, they've got a massive buildup of deferred. So I'm like, well, OK, maybe I'm wrong. Whatever I think, if the accounting is aggressive and problematic, I would change my view and I would sell that position. And that means I can't let my biases play a big role.

Eloise Goulder: Yeah it's incredibly powerful. It sounds like a real privilege from a learning perspective to be an analyst in an environment where you have all of these tools because for an analyst to grow, they need to learn where to be tempering their biases as well.

Patrick Hargreaves: I think that's right. What we encourage analysts to do is to leverage these teams as best they can. There's never just one way to answer a question. And so if you have a breadth of data science tools available to you, market research tools available to you, how are you going to go about answering this one specific question? How do we get to the nub of the investment debate by leveraging these tools as best we can? If you were doing a market research project yourself as an analyst, you'd make a few calls, you would have massive availability bias. Whereas somebody else is doing this for you at scale with a questionnaire that's been sanitized by a professional market researcher. The difference in the quality of the data that you get from that versus just doing a few calls yourself is enormous. Hugely valuable.

Eloise Goulder: Yeah. So we've touched on processes, but obviously the data and the analytical landscape is changing so rapidly at the moment with LLMs and with new AI related tools. Patrick, how much are you looking at and looking to utilize these tools?

Patrick Hargreaves: I think one has to look at these tools at the moment. The landscape, as you say, is changing unbelievably quickly. We've used old school AI, machine learning, in a number of different ways over the years. We have a tool that we developed to screen for potential short ideas, which is predicated on machine learning. We have some price prediction models in industries that are predicated on the same. We are using LLMs in a number of different ways to try to improve analyst team productivity, to try and improve the robustness or the reach of our due diligence. And that cuts across a number of different vectors. If we're going to own a stock for say 10 years, there's a fundamental question of precision versus speed that we need to try and get to the bottom of. And sometimes it can be as nuanced if I'm reading a full transcript, then it's not until I reach a specific point that I understand that that is the key thing from this transcript. If I've had a summary presented to me, I don't know whether the LLM has picked that up. And prompt engineering can potentially help and that's something we've absolutely learned over the course of the last few months is the importance of prompt engineering to getting good output from the models as they currently stand. But we're still not at the point where we can trust them entirely. We actually have an external project with some professors of AI at a top university to try and work out whether there are more insights we can elicit from publicly available data. And in time, potentially our own data that we've generated over so many years as well

Eloise Goulder: I think that precision versus speed debate is so important. But then as you say, this space is evolving so rapidly. And the use of LLMs is not just in providing decisions or summaries, but it's also in efficiency. So no surprise that you're using LLMs in the efficiency process.

Patrick Hargreaves: An LLM can search at a scale that no individual person can search at. That to us is valuable, even if the precision is not there, because it's potentially identifying data points, new sources that we wouldn't otherwise have captured. And that in itself is a valuable.

Eloise Goulder: Patrick, it's been such a fascinating conversation. We always ask our guests, what's next? Patrick, what are you really looking at the moment? Is it more of the same or are there changes on the horizon?

Patrick Hargreaves: We are always looking to improve what we do. That is a constant state for us. We've just talked about LLMs. A lot of time and effort is going into understanding how we can best leverage those and the extent to which they can improve what we do. We have a number of different work streams across the various specialist teams. One small example in a behavioral team is whether we can use the pitch of a speaker's voice to understand whether they're being truthful or not. That's proving challenging, but it's still an interesting work stream. Another area that we've been spending a lot of time on is improving our recruitment process, the processes by which we get the human hardware in to understand and interpret what we're doing and so about two years ago, we started what we call the AKO Talent Academy, where we hired three relatively junior analysts in for a two-year training program. That program has just come to an end and made offers to three more who will start later in the year. And in putting together that program, in understanding how we can improve what we're looking for, what are the personality traits, what are the behaviors, what are the skills that people need, just beyond investing where we think we can train people to have the skills that we want. It's been a fascinating journey.

Eloise Goulder: And out of interest, what are the traits that you really look for?

Patrick Hargreaves: There are lots. And things like an ability to own and learn from mistakes. It sounds simple, but it's remarkable how rarely we come across people who are truly reflective. Genuine intellectual curiosity is another thing that is incredibly hard to teach somebody. If you're not curious yourself, then you're probably not going to change. So we need those people who have that ability to be intellectually curious and to think differently. I know that is quite nebulous, and we try and test for that in a number of different ways, but that is something that we're searching for.

Eloise Goulder: Yes. Patrick, this has been such an insightful conversation. I think from my perspective, very refreshing to hear about the long-term nature of your investment process, but all of these very complex and honed processes that you use to put into your investment decision, and the feedback loops that you are providing to your analysts to make them better, to put the odds in their favor. It's really inspiring work. So thank you very much, Patrick, for taking the time to speak with us today.

Patrick Hargreaves: Thanks, Eloise. It's been fun.

Eloise Goulder: Thank you also to our listeners for tuning into this bi-weekly podcast from our group. If you'd like to learn more about AKO Capital, then please do go to the link in the show notes. Otherwise, if you'd like to be in touch with our team, then please do go to our website at jpmorgan.com/market-data-intelligence. And with that, we'll close. Thank you.

[Music]

Voiceover: Thanks for listening to Market Matters. If you've enjoyed this conversation, we hope you'll review, rate and subscribe to J.P. Morgan's Making Sense, to stay on top of the latest industry news and trends. Available on Apple Podcasts, Spotify, and YouTube.

The views expressed in this podcast may not necessarily reflect the views of J.P. Morgan Chase & Co and its affiliates (together “J.P. Morgan”), they are not the product of J.P. Morgan’s Research Department and do not constitute a recommendation, advice, or an offer or a solicitation to buy or sell any security or financial instrument. This podcast is intended for institutional and professional investors only and is not intended for retail investor use, it is provided for information purposes only. Referenced products and services in this podcast may not be suitable for you and may not be available in all jurisdictions.  J.P. Morgan may make markets and trade as principal in securities and other asset classes and financial products that may have been discussed.  For additional disclaimers and regulatory disclosures, please visit: www.jpmorgan.com/disclosures/salesandtradingdisclaimer. For the avoidance of doubt, opinions expressed by any external speakers are the personal views of those speakers and do not represent the views of J.P. Morgan. © 2025 JPMorgan Chase & Company. All rights reserved.

[End of episode]