Brad Preston: Alternative Data in Emerging Markets

In this episode, we’re joined by Brad Preston, founder of Beagleworks. Brad spent two decades on the buyside in South Africa before moving into data and research, giving him a rare perspective on how alternative data gets used outside the US and Europe.

We discuss what makes emerging markets different for data providers, why pricing needs to reflect liquidity and stock coverage, and how Brad thinks about the line between raw data and research. A highlight of the conversation is Brad walking through how he would evaluate a hypothetical South African consumer receipt dataset, from the economics of the market to the questions it could help investors answer.

This episode is essential listening for anyone selling alternative data into smaller or less mature markets, or trying to understand what buyside users really need before a dataset becomes useful.

Transcript

SPEAKER_01 0:05

Welcome to Telling Technology, the podcast focused on how businesses actually monetize and sell data. Each episode, we interview an industry insider to hear their experiences and lessons learned.

SPEAKER_02 0:15

This data is powered by Valitis, the company that transforms your data into investment-ready intelligent products.

SPEAKER_01 0:21

If you enjoyed the episode, please subscribe from wherever you get your podcast.

SPEAKER_02 0:26

Today's guest is Brad Preston, founder of Google Works. Brad's had a pretty unique journey across the financial and data space, having spent time both on the buy side using data to make investment decisions, and now building data products themselves focused on emerging markets. That gives them a really interesting perspective on how data actually gets used in practice, especially outside the US and Europe, where the dynamics can look very, very different. In today's episode, we'll get into what alternative data looks like in emerging markets, how funds think about evaluating and paying for data, and some of the lessons Brad's learnt from sitting on both sides of the table. Brad, welcome to the podcast.

SPEAKER_00 1:01

Hi, Eric. Thanks very much. James good to chat to you. Thank you.

SPEAKER_02 1:05

So maybe it makes sense to jump into BeagleWorks. Tell us about what you're working on there.

SPEAKER_00 1:09

Sure. Thanks very much, and thanks for the opportunity to share. So BeagleWorks is an alt data and research business. We really focus at the moment on, I would say, two categories of data sets. So we're based in based out of Cape Town, South Africa. So quite focused on the South African um market and focused on building out traditional alt data products for South African listed equities. So that's one of our offerings. And then the other offering is building out a some special situation or event type data sets that are more globally focused, and thus far with a real focus on short settings, so on the short side, a data product around activist short settings, et cetera. The ambition is to grow the essay side, perhaps broader into emerging markets over time, and the the um the global side into other special citizens types of data sets.

SPEAKER_02 2:13

Awesome. And I mean it's quite rare for uh a buy-sider to move across uh and create a data business. There are some cases where that's happened. But so what drove that decision to make that move?

SPEAKER_00 2:24

So I spent um about 20 years at the um actually at the same uh boutique fund manager, um, joined sort of just after founding of the business, a business called Merchants Investment Managers, and saw that business grow over 20 years, um, sat in various different seats um within the business. And there were a number of reasons that I made the decision to leave the buyerside, um, largely driven by some some some personal reasons um and needed to take a break from running institutional money. And, you know, I started out um as a quant, um, but spent to spent quite a lot of time with my career in various different discretionary roles, but always trying to look at things from a quantitative perspective. I've always had an interest in the alt data, but that hasn't been in a very developed market, especially over the last 20 years in South Africa. So I left the um the buy side, wanting to continue to focus on generating um signals and trade ideas and insight into the investment process, uh, but perhaps spend more time focusing on that than focusing on the uh management of clients, communication with clients, and those that sort of um part of the process, and to look at things from a more data and quantitative driven perspective. And at the same time, I think I looked at the market and it felt like um emerging markets in general and market like South Africa had lagged behind some of the progress globally in alternative data. And I think there are a number of reasons why there's an opportunity for these types of markets to catch up as the cost of acquiring this data drops significantly with uh with AR tools, etc.

SPEAKER_01 4:11

Eric mentioned that um uh that it is quite uh relatively rare for someone to go from the buy side to to becoming a a vendor. It'd be really interesting to hear um from your perspective what alternative what alternative data you were using when you were on the buy side, what that investment process looked like, and then kind of how you're you're thinking about that now that you're on the vendor side.

SPEAKER_00 4:32

Certainly, um James. So so the my experience of actually as a user of alternative data um is probably quite limited, especially relative to to other people in the space for a number of reasons. Um, you know, I largely ran discretionary processes. Um the the quantitative processes that are that I was involved in were more traditional multi-factor um quant type of processes. And also just due to over the last 20 years the limitation of these types of products and data sets being available in in SN. And I think it's more where some of the inspiration has come from. This was something I was interested in, attempted to do internally um myself a number of times. And it's just a difficult sort of a boutique fund manager, difficult to build out um, you know, all of the infrastructure and capability to perhaps produce some of these data sets yourselves when they aren't being being sold by vendors and the rest of the market. Um some of the places where we did, you know, there certainly have been over the number of over the last few years more and more of these types of products that have been started to be offered in South Africa. So there were some instances where where we did um use some of this, but I think it has often been something where I looked at the potential opportunity, but wasn't really able to execute on it. And that's been some of the inspiration to go and say, okay, well, I think that that is a gap in in the SA market, and let's provide some of those data sets and some of that capability into fund managers in this market.

SPEAKER_01 6:06

Okay, so if we were to kind of take uh a real world or it's a fake a made-up example, but um, we're currently working with a uh a Canadian consumer receipt uh transaction data set provider. Um and obviously I know that your your knowledge is much more of the South African market. So so let's say um I came to you and said, I've got um a South African consumer receipt panel, the way it's generated is you know, South African consumers, when they go shopping, they take a photo of their receipt, upload it to us, and and we're able to turn that into a kind of transaction level panel that we want to make available to investment funds. How would you think about uh the research process you would do uh on the buy side or even even now in your your um uh uh work at BeagleWorks, uh, into thinking that there might be signal there? How would you analyze that? But it would be really interesting to hear that the process you'd go through.

SPEAKER_00 6:58

Certainly. So I think the first question, and maybe to look at this from two different perspectives, is to um with an analyst hat on, what sort of research would one want to do on that type of data set? And then from a vendor or a data buyer perspective, you know, the question of do the economics work, which I think is often a uh a much bigger question in our markets. So I would say from you know, to tackle that at first, I think, because there are these data sets do exist in South Africa, um, I think often the use case is more um on the corporate side than the investment management side. And I think that's because often the uh the economics are quite different. Um and if you if you look at um both liquidity and sort of breadth of coverage, to get the cost of acquiring that data to match up with the opportunity to monetize that data in the market, I think that's you know, that's always a question that needs to be answered. And it's perhaps one of the driving reasons why smaller markets haven't um seen large penetration of these types of data sets into investment management market, because you know, where a data set like that may have application across a large number of liquid opportunities in a developed market, you know, that data set may only be applicable across eight stocks, um, some of them not necessarily that liquid. Um, you know, another practical consideration is that um, you know, SA companies report on a six-month basis rather than on a quarterly basis. Um, maybe joining the US. Well the US may be joining us in that. But again, that you know, that cuts down your your sort of trading opportunities if you're trading on the the release, cuts that down in half, also cuts down your historic data points to try and build correlations, etc. So I think a starting point would be to understand the economics of acquiring that data versus the economics of able to monetize it. Um I do think though that I mean, you know, obviously there's there's really interesting insights, and particularly at the moment, one of the questions um I think that's top of mind is uh the impact of energy costs uh everywhere. You know, globally we're seeing massive um impact on energy costs. We've seen very significant um fuel price increases across South Africa, and there's lots of questions as to how is that impacting on the cost side for um across across routine um retailers in South Africa, and then are they able to pass these these on? So I would say, you know, as a research question right now, a top of mind research question from that type of data set would be to look at first of all, can you see um at a product level um where which retailers and which products are passing on inflation? Um and then secondly, um to look if you can on a consumer level, and I suppose that's dependent on you know the sort of nature of the panel data, can you see um if we're seeing trading down from consumers in response, you know, are sort of average basket sizes perhaps dropping? Are consumers migrating to perhaps um cheaper, cheaper options in response to a very significant increase in their fuel spend?

SPEAKER_02 10:06

And that uh passing the cost through to the consumer, is that to try and think about the the um maybe the business has decided to eat into their own margins? Is that uh or they've decided obviously to pass that cost on to the consumer and how that might affect how consumers engage with that brand. What's the sort of driver? Yeah, I guess what's that the why of I want to understand whether that's been passed through? What's that next step?

SPEAKER_00 10:28

Yeah, certainly. So that's that's really looking at um margins at both food producers and and food retailers. So the um the the fuel effect is is immediate. And you know, we've been doing some work around uh trying to understand using various different disclosures to try and understand the exact energy impact and fuel um cost at these businesses. Um diesel prices are close on doubled um in South Africa over over three months. And for a business with a significant distribution network, you know, that's a double in your doubling your distribution costs. Um food prices haven't reacted um at this stage nearly as much as you would you would have thought. And so, you know, and some of our our sort of research seems to suggest that there's about a nine-month lag to see that full um pass through into food inflation. And so a a food retailer is really going to try and offset a lot of that through taking some margin themselves, but I think just generally in in revenue increase, um, if they get if they can pass through the food inflation, um, you know, they should they should pick uh gain back some of that margin. But if that's on a nine-month lag, how much margin pain are they taking in the interim before that's um that is passed through? There certainly are different leads and lags in terms of certain products, so certain product sectors on the food side are able to pass through inflation before others. Um, particularly if if you know in a market like South Africa, um, import substitution is is quite important. So suddenly import costs are a lot higher. So an industry that is able to now has got a bit of defense from from imports because those imports are so much more expensive, they're able to pass through pricing quite quickly, as perhaps a more competitive domestic sector might not be able to push through pricing um as quickly. I mean, another another interesting dynamic in South Africa right now is that um Walmart has just entered the country. Um, and there's only really the uh a handful of stores. Um but they're um competing quite aggressively on price. And they're only having to do it across uh, I'm not sure where the store numbers are right now, but it was three stores in the country a little while ago. Um so they can they can take some um some margin pressure across only three stores, but just create the impression of being a lot a lot cheaper than peers. And so, you know, into this dynamic, you've got a big global um sort of value player coming in and competing quite aggressively on price. So a number of uh interesting dynamics around um those sectors and um and and cost in in South Africa.

SPEAKER_02 13:09

And have you seen any interest from maybe the uh the bigger funds in the US because Walmart has sort of expanded into the South African region, that they're interested in seeing the success of that, or is the expansion still quite small? The funds have have cared less about that outside of South Africa?

SPEAKER_00 13:26

Yeah, so so so we haven't um seen that um ourselves chatting to some other players and the market in general, and that's not not particularly particular to this question, but in general, the feedback is you know for a global fund or a US-based fund to look at just the South African market um for a you know multinational player. What is South African what what are Coke sales in South Africa or McDonald's in South Africa doing? Generally the feedback is that's you know, it's it's too small to move the needle. And so they are looking for much broader data sets than you know, to try and aggregate single market data sets.

SPEAKER_02 14:09

That's really interesting. I I don't know how true this. I saw uh I read an article a while ago about the uh Coca-Cola sales disproportionately come from Scandinavian countries. That may not be true, someone told me that in passing. Um, but therefore funds look at that quite aggressively because they have a disproportionate amount of sales. Is there any sort of like product in in South Africa or brand where they see a disproportionate amount of spend in that realm where you see this unique interest from from other more global investors?

SPEAKER_00 14:45

Not that I can think off the off the top of my head at the moment. I do think you know those those types of questions of um, you know, perhaps you're seeing a company expand. Can we track their expansion to this one market? It's it's not meaningful enough to move the needle on the overall, but because it's an isolated case of an expansion and how well it's it's progressing. I feel that often those types of questions are perhaps you know constrained by just bandwidth of the analysts, the peer, et cetera. And I do think that it that is potentially an interesting theme, that as the you know, as AI gives all of us more analytical bandwidth, the ability to then consume more data, even if it is only perhaps a small portion of um of your sales, but perhaps still tells a story, um, may change the dynamics of some of these types of questions.

SPEAKER_02 15:40

Interesting. And if we think about the funds in South Africa, uh are they investing more globally or are they sort of hyper focused on South Africa as well?

SPEAKER_00 15:50

So a combination of both. Um South Africa has um quite a strong home bias. Um, a lot of that is regulatory, so pension fund regulation um uh limits the amount of exposure that um pension funds can take outside of South Africa. That limit that that limit has been increased, and so you know that regulation has been relaxed over time, but it's still quite a large. Um there's a large focus on South Africa. And I mean, and and and SA has a very developed um pension fund industry. So sort of relative to the size of the economy, the pension fund and saving industry is very developed. So there is a large um home bias. Uh but any of the larger hedge funds or larger fund managers are very focused um globally as well. And I think you know, certain funds, when they get to certain size, the opportunity for growth comes comes outside of SA. It then though becomes a challenging question because those are businesses that often have competed very strongly um with Intel Africa, have a strong argument around their competitive advantage understanding this market, and then go and you know, need to argue that they are um able to compete at running um you know uh develop market um money from the top of Africa, and that's a hard argument to make sometimes.

SPEAKER_02 17:15

And then I guess if you think about the pricing argument, and I think you touched on the economics earlier, but if the majority of money managed in South Africa is sort of managed internally, or um there are providers coming outside to try and sell into those investors that are investing more globally, how should providers be thinking about their pricing structure versus selling to a US fund?

SPEAKER_00 17:38

Certainly. So I think I think that is a big challenge. And if you think about the economics of alt data, you know, there's the cost of acquisition of the data, and then there's the ability to monetize, and that ability to monetize is, I suppose, you know, signal multiplied by liquidity multiplied by number of stocks covered. And in this open market, you know, number of stocks and liquidity are often a lot lower. And so just that ability for um funds to be able to practically monetize those signals, even if they are very strong, um, is is a large constraint. So I think from that perspective, um, data sets need to be priced differently around being able to take advantage of that. I do think, you know, the like a lot of markets, a lot of emerging markets, the cypher market is quite um concentrated as well. So that question is very different with if you are in, you know, the top 10 or 20 names as opposed to sort of the long tail of the rest of the markets. Um, you know, the other practical type of considerations are um, you know, a lot of funds don't necessarily have a separate line item budget for alternative data. They will they will think about um acquiring you know data, so like Bloomberg, Terminal, et cetera, acquiring market data, and then they'll think about having a a research budget, um, which is typically traditional written research. And you know as opposed to a large multi-strat with a data sourcing team and a particular old data or data um uh budget line item, that often doesn't exist. And so you need to decide are you selling and competing for the sort of market data um budget line item or are you competing against research providers?

SPEAKER_01 19:26

Yeah, that that was um I was actually thinking earlier when you were walking through the retail example, your what you were thinking about was very South Africa market specific. Um and then even on the research call, I think we walked through a few other examples. One one was to do with mines and water uh and then rainfall, I believe. Um and or maybe I got that wrong, but it was clear that kind of across the board, your understanding of everything that's going on in the investment space across different industries, specifically within South Africa, was critical to your assessment of the utility of an alternative data, but also like what the kind of the juice you were able to squeeze out of that and how you would go about doing that. So I get I guess you you you were touching on it, but like to what degree do you consider your the Beagle works to be a data provider versus uh a research provider? Are you and have you found that you you know maybe started out thinking one way and moved over time to the uh to be something else?

SPEAKER_00 20:22

Yeah, so I think there has been some some shifting um of that. And it's also so I think you know, to some extent I have reacted to to market demand. Um uh and and to the other extent, it's also about trying to understand, you know, what are um my competitive advantages versus disadvantages. So, you know, I'm not going to um at this stage as a startup business build out um the infrastructure and the scale to compete with a um large data provider in a very, very large data set. But the competitive advantage that I do have, uh particularly in the South African market, is having spent 20 years on the buy side, um, yeah, having been a portfolio manager and a CIO in this market, understand the market well, understand the questions that need to be answered, et cetera. So I think both of those things have driven me in that direction. The um the feedback from clients as well, and you know, that's also supposed depending on the type of client that you're focusing on. A boutique or sort of a discretionary hedge fund, perhaps with an investment team of 10 who are both looking at you know, essay and global stocks, um, you know, do have significant bandwidth constraints in terms of the individuals in the team. And so a lot of the time those analysts are asking for insights as opposed to to data. They want to see the data. And um it's also a question, I think, of you know, marketing that data point or that data series out to clients is that clients don't know. Know what data is out there and whether it might provide signal or not. And so often it needs to lead with research. So this is a research piece. This is the insights that we've seen, and this is how we think it is actionable. And then that data series may be something that they would then ask to be refreshed multiple times afterwards because they've seen that that has been relevant for a stock or a trade, et cetera. But the short answer is that certainly there is a lot of research laid on top of the data as the way to get that in front of people, but also to add value to it.

SPEAKER_02 22:34

I mean, it sounds like given the liquidity, given the the narrow amount of information, sort of quote unquote alternative data that's that's available in in South Africa, that it seems like there would be some desire for exclusivity. How is that have you had any of those conversations and how's that perceived as you think about the compliance concerns the US market has around exclusivity? Has that become up in conversations with you?

SPEAKER_00 22:59

So that has come up in conversations. It's not something I've practically executed on at this stage. Um so there's certainly, you know, the South African market is concentrated in terms of stocks, but it's also concentrated in terms of fund managers. And so, you know, the the the the top five hedge funds in South Africa um, you know, have a big, big share of AUM. And so, yeah, with liquidity constraints, they would they would naturally be interested in um exclusivity, um, really around you know time of delivery of the data. Um, there is a um a sort of a natural dovetail in that some of the larger um sort of traditional long-owned institutional managers of Africa's just got a very strong tradition of sort of long-only value investing. Um Adam Gray, uh, which then went on to sort of birth Orbis, is um you know the large, very um respected global um value house, been very successful in South Africa, etc. So, you know, those types of investors are much more interested in in longer term insights. They don't mind if data delay buy a month and then perhaps do a chance are interested in the more shorter term. You raised the the question of compliance. Um I do think that that you know is an interesting question. I don't I haven't really seen any evidence that that has been um discussed by the regulator in in South Africa. Generally, you know, South African market regulator follows um sort of UK and US best practice. But I do think where that potentially becomes an issue is um as more data from um corporates shifts into the SA market, because again, the market is so um concentrated. So, you know, South Africa has really five large banks, um, so sort of a you know a very concentrated banking sector, very concentrated food retail sector. If any of um those institutions were to sell um alternative data feeds into the market, I think there'd be real questions um just around, you know, can that be provided um exclusively? At what point does um do you need to start answering regulatory compliance questions around around those types of data sets? I think, you know, more publicly available retail web scraped, um, you know, or you know, web scraped type of data sets, um, you know, regulatory data sets that are made publicly available, but perhaps um you know collected and collated. I think those, you know the the regulatory question is is maybe not as as relevant there. But I think you know the question is to when an actual listed player is providing data, that at what point does that cross over into material non-public information is potentially a question.

SPEAKER_02 25:46

Yeah, and to your point about uh the South African authorities kind of or regulators looking more towards the UK and US, I don't think they're necessarily buttoned up on what uh alternative uh what the sort of the line should be drawn on sort of non-public information, uh what's too much. Uh I mean in my time in the industry I've come across some providers that capture a significant amount of spends on you know there's this thesis in the industry that you shouldn't capture more than 5% spend of a given business. But as these big corporates come into market, you think like uh payment procurement processes, they are capturing way, way more than that. And it does raise the question of what is too much information for for an investor where it becomes unfair and how should that be distributed across the market? So, like to the point that your regulator is looking towards the US. I don't think the US is very buttoned up on that either.

SPEAKER_00 26:44

Certainly. And um if you add to that, we now need to try and answer these types of questions that are on um prediction markets and various other different challenges. Yeah, I'm not sure anyone's got too many questions.

SPEAKER_01 26:56

Um actually we've spoken about prediction markets on uh on a previous on the episode we just uh one of the episodes we just released. Um I'd be interested to hear a bit more about the types of data sets you see these long value uh investors finding particularly interesting. I think often when um what people kind of know that from a systematic or a quant perspective, one uh other than signal, one of the most important things that they need to have is a large coverage. Um they need to basically have numbers, maybe ideally thousands of stocks covered within the data set. So thinking transaction data, clickstream data, that kind of thing. Um you mentioned that these long value funds are um more focused on data sets which obviously have implications over the maybe the one year, three-year, even five-year time horizon, I suppose. And so I'd be interested to know what types of data sets in your mind um lend themselves to that. Because obviously, like the five-year predictions are incredibly difficult to do, if not impossible. Um, but that that is kind of you know most discounted cash flow models, et cetera, that these uh long value shops will be using. Um, they do have those sorts of time horizons. And so yeah, I'd be fascinated to know what sorts of things you're seeing them be interested in.

SPEAKER_00 28:15

Certainly. So I think you know, I've been thinking more and more about um sort of communicating to potential clients and saying that the value of alternative data is either faster, which I think is you know how a lot of people think about it. So, you know, more coincident data that hopefully gets you a read on the next earnings print earlier than than than other people. But also, you know, richer and and and it's able to enrich your view of the company. And I think that is where so it's not necessarily something that has a long um sort of term predictive signal. It's rather data that gives you an enriched view and an ability to ask a question that perhaps you couldn't if you didn't have that data set around how a company operates, how the dynamics of an industry are operating, um, etc. And so I think that that there really is the question that um a lot of these investors are trying to answer. They are looking at, you know, how do we use a data set that perhaps helps us test a claim that management has made? So perhaps management makes a claim, we you know, we we have to either take that at face value, or perhaps we can test that against some data that we've got, or perhaps management is executing on a certain thesis. Can we go and find quite granular detail? So, you know, some of the examples that we've looked at, um there's a um I was in a discussion with a client just around a uh hospitality uh companies in Africa, a um uh hotel operator that has been refurbishing a number of their hotels. And can you use um some proxies of volume, but as well as ratings? So you know, using sort of some of the travel sites, getting ratings of those hotels. And can you then go to granular level and break that out on a per location basis and identify those locations that have been refirmed? And are you seeing a move in either some sort of proxy of volume there and occupancy, or even just in rating? So if you're spending money on refurbishing these locations, are they moving from four to four point two stars over time, or is there no quantitative response to that? Um, you know, looking at, for example, clothing retailers and understanding relative um range, relative pricing, um, et cetera, between between the different um clothing retailers and the competitive dynamics um between them. There's been some interest in that um type of data. Um a question that that we use some of our data to to look at is like a lot of um the rest of the world, particularly EM, South Africa has seen a very, very strong growth in um in Chinese um vehicle sales in South Africa, going you know, from almost sort of the standing start to now more than one in five vehicles in SA is now Chinese. And that has had a big impact on um a lot of the motor dealers uh uh dealerships, as well as also the secondhand motor dealerships. So um we've seen new car sales accelerate quite a lot faster than secondhand vehicle sales because you you know the new vehicles are just so much more competitive. Some of the feedback from some of the um the used vehicle dealerships is to say, well, you know, at some point this becomes a tailwind as those vehicles then filter into our inventory. Um we've been able to look at that question in quite a lot of detail and um look at the representation of various different brands across used dealership inventory, but also look at the age of that inventory. Um and so they're certainly used used um vehicles dealers that focus on a very old inventory or very old old market, so sort of a low-cost average eight to ten-year-old vehicle. And they aren't seeing any penetration or meaningful penetration of these vehicles into that. And you know, we would argue that perhaps given the the average age of the vehicle across their um their inventory, it's actually an eight to ten year lag. Um, not necessarily, you know, the argument that's been made is that it's it's going to become a tailwind very soon. And so those questions, you know, there's no, you know, there's no requirement on sort of very up-to-date data to ask those sort of more thematic questions, but they can often give an analyst um the ability to ask a question or perhaps answer a question that they weren't able to do just looking at financials.

SPEAKER_02 32:47

And do you find the I can imagine the differences between emerging markets and and more developed markets is the types of data that you're trying. Different types of data be valuable in different markets, like the US and the UK and Europe are very digitalized in terms of payments. Do you find that more physical receipts versus e-receipts are would be more valuable in somewhere like South Africa?

SPEAKER_00 33:10

Yes, I think certainly you know the penetration of um of online retail in in Essay is a lot lower across certain um you know sort of clothing retailers, anywhere between two and seven percent. And even there, you do get feedback from analysts saying, you know, if you were to provide me with an e-receift data set, it's 7% um of sales, and it's possibly not representative of of the overall market. Um, I'm not sure how much how much value there is. Um, I think even further than that, you know, there are um so so the large um listed food retail groups in South Africa have got very, very good coverage and penetration, but um, even outside of that, there is a very strong informal market. And there are certain food producers, for example, that are highly exposed to that informal market. And so, you know, that becomes not only is that not digital, I mean, that's all sort of cash and paper receipts, um, it's a market that falls outside of your sort of traditional listed retailer. And so that becomes an even more difficult um market to penetrate. So I think you know, there certainly are um these sorts of differences. South Africa is a particularly interesting case because it is such a bipolar, um, unequal market. And so, you know, it's not perhaps just that it is less digitized. It's um that other markets, it's highly digitized and um in a certain segment of the market, and then very different in a different segment of the market because of this very large um inequality in South Africa.

SPEAKER_01 34:50

I I really like what you said about start almost starting with the question rather than the data. Start with the question, figure out what data might be able to answer, and then go source that data. Is that essentially how you you work with with buy-side clients? You you speak to them, find out what problem they're trying to solve, and then go, right, now I'm gonna go and see if I can find an alternative data source that um allows us to better answer this question.

SPEAKER_00 35:12

I would I would love it if if that was always the case. Um sometimes, and I mean, I'm sure you know anyone who services hedge funds knows that um you know the information generally flows in one direction often. And so it's a little more, I think, sometimes trying to understand the questions that I would be asking in the market, trying to go out and look for that data, um, looking for data that is then accessible, accessible cost effectively, etc. And then going to go and present that. Um that, you know, discussions then happen. Um but yeah, you know, you do end up, I think sometimes just the nature of a very competitive market and very concentrated market, you know, naturally some of the hedge funds, I think, are you know sort of keep their thinking close to their chest. And so they are, you know, the feedback is provide me with ideas, provide me with insight, and you know, I'll ask you five questions.

SPEAKER_02 36:10

That's fair. Moving on then to maybe the closing question, given you've been on both sides of the aisle in an industry which is changing quite quickly, especially on the emerging market side, where do you think the industry will be sort of going over the next couple of years?

SPEAKER_00 36:25

I think it's such an interesting question because so much is is is is changing. Um and to some extent, I'm probably biased here in talking my book because I hope that this is the uh the trajectory. But really, the sort of bet that I've taken is that um a few things are going to happen. So really the one is just a an EM catch up to develop markets. Um I think you know this data is it adds value, it's consumed, it's useful globally, and across emerging markets, um we'll see more penetration. And you know, the spend and the penetration of the alternative data will catch up to to develop markets. But I think the you know a big theme that is going to influence that is um the effect of of AR tools on both the cost and the ability to fund this data and to deliver it, and then the ability to consume it as well. So I think you know, as I've as I've spoken about this um this sort of equation of getting the economics to work in in emerging markets or in smaller markets, or even just you know, in sort of less less liquid stocks, I think has been a constraint to some of this with the industry uh developing in certain sectors. As the cost to acquire data drops, but also um you know, given analytical tools, perhaps the the b the ability to consume more data on the other side, both um sort of as an analyst who's perhaps supported by um by you know agents and analytical tools, but also as more and more of your data is consumed into um AI processes. And you've you you've got a sort of drop in the cost of supply and then an increase in demand should then hopefully tilt that equation to make um you know the economics work in a lot more stocks or more sectors in these types of markets. So that's the bet that I'm I'm taking. Um I think you know, then on top of that, which I think is a big question, is then what is the delivery mechanism? Um and you know, how much is there still a a market for for platforms, terminals, portals, etc.? And how much of this turns into APIs and data feeds straight into to agents?

SPEAKER_02 38:48

Yeah, and that was gonna be my follow-up question, because uh I guess we're equally biased, we're in the same market, and and therefore we're all making that bet that more buyers come to market uh and data flows into workflows a lot more easier. But if we're in a market right now where there are thousands of providers and less buyers, let's say, how does one stand out as a data source versus you know six or seven peers even in a data set if you're just listed on a you know an MCP or on a Claude or whatever it may be? What how how how are you thinking about standing out in that market?

SPEAKER_00 39:26

Certainly, I think um at this stage it feels you know just purely being you know your data is consumed by, as you say, in an MCP or being consumed by Claude, et cetera. Certainly doesn't feel like the reality of far quite far from the reality in terms of you know the the the way that uh that that we're operating um at this stage. But things are moving quickly. I think you know we've thought of uh I've thought about building this business to ensure that it is sort of API driven, you know. So if you're gonna be building any sort of product, the API is is um the first sort of call. And and then if you are layering tools on top of this, if you're layering a web portal or an Excel ad, etc., that's all a method of of distribution. Um and then I do think that you know it's there is then still room to add um insights, to have um an insight into the market, into what data may be valuable um for that particular stock for that particular client, um, adding the research on top of it, I think will still add um significant value there as well.

SPEAKER_01 40:45

Are you using uh LLMs in your own workflows at the moment? Or you know, there's tools like ChatGPT's got deep research. Have you have you used any of those tools for trying to source data sets? Or do you generally find that like your knowledge of say the South African market is so so deep that like uh an LLM is going to kind of just provide you generic answers that you already knew, or even you know, hallucinate and make up some data sets?

SPEAKER_00 41:07

So we're certainly using using these tools a lot in in the workflows, but I think it's been quite a lot of learning as to how to use them correctly. You talk about making up data sets. So do a lot of, you know, I have found the following data in this market or in the sector. What is find me something similar to the following sector, et cetera? And a huge amount of it is is hallucinated and sometimes it's valuable. Um finding very limited um ability to actually use the data for me the these tools, so chat GPT for data collection. And if but a huge amount of value in using the tools within the workflows, so um uh more structured data extraction, classifying of documents, you know, turning unstructured data into structured data, huge amount of value there. Um as a solo founder, a significant amount of value on the software development side. So I mean, and that's you know to me, I found um using the tool as a data analyst or to the data sourcing tool, um unless it's you know very much part of a structured workflow, I find it often the experience is quite disappointing. Um using it as a software development tool um has been the biggest productivity gain. Um so so so my view is that uh because I think you know there's this question, the sort of risk of do data providers get completely um disintermediated by these by by LLMs, by the sort of model providers, and anyone who wants data is just gonna spin up a clawed co-work instance that's gonna go and write a bespoke scraper, etc. I think that you know the the the sort of need to have someone running those processes, ensuring that they're running efficiently, ensuring compliance, data integrity, et cetera, will remain because it makes a lot more sense to use those tools in one place in the most efficient manner possible, as opposed to everyone else out there having these individual, bespoke um sort of data sourcing um pipelines that are then sitting on their machine, not maintained, um, et cetera.

SPEAKER_01 43:26

Yeah, I think the data integrity point is is a critical one because the the fact is that they're they're trained almost to they're trained in a way to generate answers that you that you like or answers that are plausible, and it's very easy to make data look plausible, but for it to be complete garbage. And so I think um, particularly if you're consuming multiple data sources, having someone in that that loop, making sure that the integrity really is there, um is so important.

SPEAKER_02 43:52

One thing I was always thinking about uh the way the the world is going and talking about how you know are data providers gonna get separated from the end user. One market that's quite similar to that right now is advertisement, where you have data providers that plug into a platform and brands that plug into the other side and they use the platform to pick which audiences they're gonna use to target against in activation. Um and one thing these businesses that are providing the data have to do is spend a lot of resources in trying to influence that brand or that platform and how they're positioned on that on that platform and whether they're the ones that get selected in that um in that process. So I assume if we do go that route, the sales becomes more influential rather than uh direct licensing, probably. Um especially if we go down the consumption route. But yeah, that seems to how that's gone in the sort of the advertisement route. And I could see if we end up going down that route generally with how LLMs are being used in the finance vertical.

SPEAKER_00 44:49

Yeah, I think um we all have so many open questions that um yeah that uh we'll the next few years are going to be fascinating to see how they how they play out. A brave new world.

SPEAKER_02 45:00

Absolutely. Maybe in a couple of years we can reflect on uh on this conversation.

SPEAKER_00 45:04

I'd love to come back, yeah. Yeah.

SPEAKER_02 45:06

Awesome. Well, I appreciate the time, Brad. It's been lovely having you on.

SPEAKER_00 45:09

Thank you so much, Brad. Great. Thanks, thanks so much. I really appreciate the opportunity to share and chat with you guys. Thank you.