The use of alternative data in the trading world appears to be on an upward trajectory and according to some stats, there is showing no sign of slowing down. What’s behind this move from traditional forms of research and data to that of the not so traditional? In this episode, we’re chatting with two experts who have plenty to say about this space. Tim Harrington, is the co-founder and CEO of BattleFin and Stewart Stimson is the head of data strategy at the systematic trading firm, Jump Trading. Listen in to learn about what’s behind the alt data explosion, which data sets are on the rise, what the regulatory picture is and how they see things moving forward from here.
Jamie: [00:00:05] Hello everyone and welcome to another episode of Hedge Fund Huddle. I'm your host, Jamie MacDonald, and today we are talking about the fascinating world of alternative data. Now, back when I was at UBS Warburg, that dates me a bit saying UBS Warburg, but when I was on the research side then, it was all traditional forms of research that we used, SEC filings, management presentations and that sort of thing. But as the years have gone on, the world of alternative data has grown and gone from strength to strength. And from the numbers I'm looking at, it looks like it's going to continue to do so. So luckily, I have two experts to help me out, to talk about this topic today. They are Tim and Stewart. And before I go into what topics we're going to cover in this podcast, I'm going to ask each of them to introduce themselves. And if you don't mind, going to a bit of detail about what exactly it is you do on a day-to-day basis. So, Tim, perhaps we'll start with you.
Tim: [00:00:57] Great. Tim Harrington, CEO of BattleFin. So, BattleFin, we're about ten years old. We have two businesses, a technology business that created a platform called Ensemble, which is a SaaS platform where people can log in, search for data, get insights into data, pull up different data sets and do that. And then we run a global event series where we travel around the world. So, in September, we're in Singapore, last week, we were in London, we'll be in Miami in January. And what we do is we go around the world looking for different data sets and then also connecting the providers and the buyers that are on our platform because we really find that these in-person meetings are where you start to really dig into some of the topics. My background has always been mostly by side, so I had worked at Sigma, which is part of SAC Capital running a TMT book, then JP Morgan Ventures, and then like I said, started BattleFin about ten years ago.
Jamie: [00:01:55] So Tim, maybe I can ask you a quick question straight away. At these events that you host, who are the buyers and sellers that are there?
Tim: [00:02:02] On the buyer side, it's mostly the hedge funds, PE firms, VC. And then what we've seen recently is a lot of corporate buyers coming in. And what they do is they come in, fill out a profile, and then we try to match them up with the right data providers. So, on the other side of the one-on-one meetings, it's geolocation providers, employment data providers, and as well as the larger market data firms the Refinitiv and SNPs, they're also moving very much into the alt data space. So, we started out with probably ten data tables and 100 people. And at our larger events now we're at 150 data providers and 1000 people. So, it's grown pretty rapidly.
Jamie: [00:02:45] And it looks like it's going to keep doing so.
Tim: [00:02:46] We hope so.
Jamie: [00:02:47] Tim, Thanks very much. Stewart. Why don't you tell us a little bit about yourself and jump trading?
Stewart: [00:04:00] Absolutely. So, I'm the head of strategy at Jump Trading and Jump Trading, I should sort of explain that, but we are a systematic trading firm. We aren't technically a hedge fund in that we do not trade investors’ money. We don't manage assets, but we do a lot of the same style of trading that you would see at large quant hedge funds. And I've been here five and a half years and my mission on the DSG team, the day strategy group team is to ultimately discover the data that's going to be predictive in nature or additive to our existing predictive models. So very much I would be a data buyer and have been at one of Battle Fins events. And we use a lot of the things that Tim mentioned and in a production sense by way of background for me and arriving in this space, I've been at Jump for five and half years. Prior to that I was at PDT Partners, which is also a systematic trading firm, though that is a hedge fund. Prior to that I was also at SAC, which is now Cubist and worked on the quant side there. And prior to that I spent 13 years at HBK, which is a Dallas based hedge fund multi strategy fund, which included statistical arbitrage, systematic trading as well as discretionary research.
Jamie: [00:04:15] So, I guess we all work for Steve Cohen at one point.
Stewart: [00:04:18] I think we find that out way too often.
Jamie: [00:04:20] Wello, we’ll maybe come back to that a bit later. Tim and Stewart, what I would love to do is talk a little bit about how this world has changed over the past ten, 15 years. I know it's growing at a very rapid rate. And when I was at the London office of SAC, we're talking ten years ago now, alternative data 15 years ago now, was still a relatively small thing. I knew it was around, but it wasn't something that we'd spent a lot of time or money on. So perhaps we'll start with you, Tim. So how has that world developed? I've read one statistic which I can't remember where it was now, but it was certainly took me back was that of all the data that's been collected in the world ever, 90% is in the last two years, which I guess if I thought about it longer, maybe it wouldn't surprise me as much. But it really did catch my eye that there's just so much out there that tells us about consumer behaviour. So, if you could talk a little bit about how it's all changed over the past 15 years or so to where we are today.
Tim: [00:05:17] Yeah, I always relate it back to if it was 15 years ago you were doing channel checks. I covered TMT and you'd do everything from going to the Sprint store to see which phones were selling to just trying to do as much feed on the street research as possible. And that was highly inefficient to what really interested me in the whole space was this idea that I could come across data providers, that I had no idea that there were ways of collecting data in this way and once again, fully legal, fully transparent, anonymized, why would I need to go try to put all that feet on the street research together? When I could find a data company that was able to track every time a cell phone got registered with, whether it was one regulatory body or where it was a sales, and just looking at the IMEI number and getting pretty much up to date numbers on a daily basis, that to me was an edge that that you wanted to make sure obviously was legal and anonymized and all such things. But the amount of data being created was one thing. Then it was finding the data providers that could actually make sense of it and organize it. And then the last thing that's always the most important is, ok, are you focusing on the key debate or what's going to move that stock? I always thought about things was coming up with ideas, it was like okay, first you have to establish what consensus is. So, you did need to have the sell side and do that. And then it was really going out and finding the information to build a model and then coming back and seeing where that that discrepancy would be and saying, okay, there's an opportunity here. It's not within consensus. Maybe I want to put a bet on. And that's what we did. We went out and developed a business that said, okay it's hard enough to run money on its own and then to have risk management and analysts, how are you going out and sourcing this data and testing it and evaluating it and organizing it in a simple way. And what we saw was initially it was. It was the two Sigma’s, World Quants teams that were the first to engage. And it was very systematic at the beginning 80% or 90% were our clients and now it's probably 50/50, fundamental discretionary and 50% systematic. So, there's been a shift in that direction as well.
Jamie: [00:07:45] What about in terms of the different resources you can source for information, whether it's credit card receipts, whether it's satellite imagery? Just the pace of these things as well has obviously just become almost instantaneous. Can you talk a little bit to how the different areas of alternative data have changed?
Tim: [00:08:01] Yeah, Stewart and I were talking about this. At the beginning I think satellite imagery was the one that everyone talked about because they were like 'Oh, I can track everything, I can see every space on the earth, every 24 hours'. And believe it or not, that was one of the slower ones that we saw take off. I think the credit card data probably was one that grew fastest. And then you even saw that develop into another layer, which was email receipt, because credit card was very categorical. We can look at how some of the consumer stores are doing or the Home Depots. But when you went to email receipt, you want a whole other layer where you could actually see what people were buying and what types of products. And then there was another layer after that which was, okay this is interesting, but you're still looking at last quarter, so what do I want to see for next quarter? And that's where we started working with these survey companies, like Momentive, which is the old SurveyMonkey, and they're actually going out and asking questions like 'okay that's great during COVID, you were buying home improvement stuff, but what are you buying post COVID?' What are you doing post COVID? Are you going to the gym again? Are you doing this? And so, it was kind of the reason we called our platform Ensemble was you really want to take all of these different things credit card, email receipt, hiring, reviews, survey data, and really start to put it together to get more of a holistic picture.
Jamie: [00:09:26] Stewart, why don't we turn it over to you and maybe you can go back to how you've been using alternative data and how that's changed. And perhaps you can touch upon what Tim said about satellite imagery perhaps being a sort of false start.
Stewart: [00:09:39] I think 15 years is exactly in line with when I thought about what alternative data is and how I wanted to incorporate it into my own career and for the benefit of the firms that I work for. A little bit of a story there. I was at Continental Airlines. I forgot what it was called, but I was their rewards member. And one day I woke up to an email that said, Continental and United have merged. But don't worry, your reward points are all safe. Go to UnitedContinental.com and check it out. And certainly, at the time there was a discretionary component research that I supported and just natural curiosity took me down the path of well, there's this URL that seems to have been known by the company before they sent me this email. There are DNS records that are updated across the globe 24 hours a day. When did this get registered? So, I found out when it got registered. And then at that time, obviously, now these things are private, but at that time it was fully exposed who registered it. And it happened to be a law firm that represented Continental Airlines. And it was registered eight days before I received that email before they announced the merger. So, all of a sudden, this became a very interesting path. Now there's lots of things that need to be explored, such as is that public data, can that be used? And these started to develop into processes by which we would evaluate. And it really is defined how I view a large chunk of my career, which is oftentimes ideas are generated by post-mortem scenarios. So, an event happened. I want to figure out, could I have known that event before it happened? And going down that path, tends to yield answers. 15 years ago, 12 years ago, it was very difficult. You were talking about companies that often times would have exhausted in the process of doing their business and they'd often delete that data. Shortly after, the concept of business analytics as we know it today was not nearly as robust in terms of reliance on data. So, a lot of it was convincing them to not do that and then also convincing them to get to a place where they were comfortable exposing what their state of their business has been. And so, what I think we see now as your statistic about the last two years of data, you're seeing companies truly understanding that when they produce data, it's an asset and is sellable likely to someone. I think that's really where we've gotten to be this acceleration of data. Let me talk about satellite data. I think that's always a fun one for people to talk about because it's cool. It involves outer space and stuff. So, it's fun. I think that where I go with something like satellite data, where I did in present time as it sort of was hitting the market, was the frequency with which it updated. It's not real time. And from a systematic background, I really want real time as close as I can get it. So, it's every 24 hours. And if there were clouds in the way then you had all this noise, maybe a delay. And so, the concept of counting cars was never actually interesting to me. We're two seconds away from being able to see where people are actually standing at a given time because we all carry smartphones. And so, I felt from a retailer perspective, that was a much more interesting thing. That being said, satellites are really cool for monitoring things crops, which are very important to lots of firms. We know lots of futures traders out there, lots of commodities traders, and so there's definitely a place for it. It just wasn't quite the use case that they initially hit the market with.
Jamie: [00:13:27] You just made me think because I used to trade insurance companies and obviously hurricanes hitting Florida was always a big deal. And there was a joke that what you would do is in about May or June before hurricane season really started, you’d go down to Miami and just stick your finger in the water. And the idea was if it was a certain degree centigrade, then it would be more likely to be hurricane season. But satellite imagery would have been far more useful to me then. But what about the discretionary side of things? I know, Stewart, you're not so much on the discretionary side of things, but is it being used more and more by the discretionary side? So, this is really just a growth area for more quant traders.
Stewart: [00:13:59] Tim, do you mind if I grab that? I think discretionary tends to have more opportunity to use a broader set of alternative data. Often times data sets that we'll see are a little too narrow for us to use in standalone mode where I think a discretionary researcher is going to have the ability to give that human component and that understanding of the context of what that data is revealing to what their industry expertise is. Systematically, we have difficulty in saying this one narrow data set, we can make a decision off of a large number of securities that we're trading based on that it's not impossible, but it's more difficult. So, I think that what you are seeing are firms that are doing, first of all, on the larger hedge funds. They do have the technical ability to do this for the benefit of their discretionary traders. When you talk about smaller firms, you're starting to see a lot of vendors provide platforms that will allow for them to extract these insights out of that data. That speaks a little bit to ensemble. I think that the world is trying to centre around allowing firms to use this. And as those platforms develop it's going to be smaller hedge funds that are able to pick this up without having to have vast technical staff and technical stack.
Tim: [00:151:19] Just following up on that. We just had an event in London, and the part of the reasons we even do the events is to get insights from data providers. A lot of times we have to go to them and say, our last event and it was with Salt that one of the co CIOs of Bridgewater was talking and he was, well look, I think we have one more shoe to drop. I think that Europe could be an area where there's significant downside. And as he's talking about it, some of the things that came up, like, look what's happening with energy prices. And he said, look, one of the big fears we also have is employment. If unemployment, especially in the US, hasn't really done anything. So still really low, that's another significant area where if you think there's a recession, you want to keep an eye on. But when we're in London, they're talking about energy prices going up something like 300% this winter and so much of it is going to be a factor of how cold it is. So, one of the things we were doing is we work with a company called Weather Source. So, we've been trying to figure out how do we use Ensemble, our platform, to showcase some of these use cases and let people log in and say, Hey wait, you're right. I'm connecting the dots here. If it's cold weather, it's going to affect these regions. What the price of oil? Then let's actually look at some of the maritime data sets you have, like, where are people getting this oil? They're not getting it from Russia anymore. Russia's sending their oil to India and maybe China. So, is it the US that's making up that difference? And it's such a cool matrix to start seeing where you're tracking weather to predict oil, which then is going to relate to consumer spending because, I don't know if you've seen what's going on with rates too, you try to buy a car, you're not buying at 1.9%. It's 5.9% or higher. Your home mortgages, if you're buying anything now, are going to be way higher. So, there's less discretionary spending that can happen. And how is that going to trickle through to a lot of these names? And Stewart said, let's watch if they're walking into Lululemon or not, it doesn't sound it. $100 stretch pants aren't as attractive as they used to be. When you go next door and get some for 25 or 30. So I think that that mosaic is interesting to watch.
Jamie: [00:17:36] Tim, a couple of questions just popped into my head as you were speaking. Firstly, with regards oil, how much detail can you actually provide? Obviously, you can provide where ships are, but do you know how much oil is on that ship when it's going to dock? And then secondly, I'm so interested by the employment thing, because how would you go about trying to source that kind of data? Because, as you say, so many people are actually still surprised at how strong the labour market is. But where do you go to try and see if there are cracks starting to appear?
Tim: [00:18:07] I guess first on the oil side. You can see the AAS signals of where they're going. Then you can look at the bills of lading in terms of what's coming in and what's not. A lot of them I think, have to register. There's even some public web websites you can go to that would get more into the web scraping side of things. What we do now, which hopefully makes it easier for the discretionary side, we have something called data hunter. So, people just come to our site at BattleFin.com, click on data hunter and they tell us what they're looking for, similar to what you are saying. I'm an oil trader. I need to know what forward crude looks and which data sets you recommend. So that's the energy side is pretty advanced. I mean, on that side too. You're also trading against sophisticated oil traders. So as a consumer, I wouldn't say don't go in and try to become the next one unless you have a differentiated view. But on the employment side, there's everything I think on Ensemble, we probably have seven or eight different data providers that just look at employment data and they do it differently. Some go to the websites that the companies have and they're scraping what jobs are available. Some will look at what they're paying, some will look at various other factors. Some will go to the job sites and look at those and aggregate how many are currently available, what's happening. We have another group and I guess Elon Musk found this out the hard way, if you're about to downsize and it relates to anyone that has health care, you actually have to be filing that before you actually do those cuts. So looking at that data and saying, okay, some of these major corporations are starting to file, we should be on look out. It's interesting because I think in the past employment data and maybe you have a view on this, Stewart, it was a little bit of a lagging indicator. It finally got to the bottom, but the market had rallied 15 or 20%. Now we're almost flipping it and trying to figure out could this be a leading indicator and what are we looking at for next year in terms of will we be in a full recession or not? I have a different view because I think we've already been in one and I read somewhere that they actually just change the definition of what a recession was. It used to be one or two quarters in a row of slowing growth. Now it's let's wait until the elections are over and kind of change this. But yeah, there's just a ton of data around that. And there's systematic ways of using it and there's fundamental discretionary. I think what Stewart said, which we see it very much as systematic needs, a lot more data. They're give me everything, show me as long of a history as possible. If I can get it right, 53% or 55% of the time, that's a win. Whereas the stuff that we see from fundamental discretionary is more Hey, I own Apple. I'm worried if there's a recession that the iPhone 14 or 15 is a disaster. I don't even care about the historical stuff. Just give me daily inputs on how many they're selling and then I can consensus and that helps me. I'll buy that data set versus the systematic funds. Like 'what do you mean you don't have five years of history?' I can't back test that. I can't make sense of it. Then the third area, which I think we've always been hoping for, and it's probably the next stage of growth for alt data is the corporate market. And what Stewart said earlier, which was they're generating so much data now and they're actually getting to a point where they understand it and have organized it, that's been impactful because now they're saying, 'Hey, oh, and what can help me look at my data in a better way or supplement it to make it more powerful' versus if you talk to them three years ago, they're like ‘oh my God, I don't even know what the hell is going on here. Just delete it so I don't have more AWS charges'. So, I think corporate could be a pretty exciting time, hopefully next year. But once again, it's been one of those satellite imagery where it's like 'okay, next year it's corporate.'
Stewart: [00:22:09] I think to add to that, there's obviously a tremendous amount of ability to sell into corporates. From my perspective, I'm nervous because those tend to be easier sells and the data vendors tend to want to deal with us less in that scenario. But on the same token, that's the way Tim just mentioned, you've got to file for layoffs. If you're on layoff in a specific location and that's the WARN Act. That's Congress. I think this is how we hunt some of these things down. We look at laws, we look at what should be happening. We look at transparency that's beyond just the SEC required filings. And we try to piece that together. I wouldn't from the quantitative perspective, from systematic perspective. Tim did mention the dream data set right, which is wide and deep. And we love it when there's lots of history because yes, numbers of observations give us levels of conviction on what we're doing and allow us to trade in the automated way which we do. But I think the community's next venture and where we're already sort of at is piecing together where these data sets are valuable but narrow, piecing them together with other narrow but valuable data sets that can inform in the same way across sectors. I think that's where sophisticated firms are heading. I think that's ultimately where data sellers are going to head to, I think partnerships, consolidations, things like that will continue to happen, which in the traditional space, we see massive consolidation in the last few years.
Jamie: [00:23:47] Interesting. I just want to ask a few more questions we're running out of time here, but I feel I could talk to you guys for hours. I should have asked this perhaps earlier. Maybe Tim, you can speak. What are the current rules and regs in terms of what corporates can sell to data collectors? I'm over here in the UK at the moment and GDPR was a big thing about data protection. I mean, for example, American Express, what exactly can they sell? It all has to be anonymized. How do they do it? And then do you expect any law changes in this space?
Tim: [00:24:17] Yes, it'll keep evolving. I think it's actually better that we get more regulation rather than less just because people generally want to follow the rules. They just need to have the rules be clear so that they can follow them. I mean, the basics are you can't give out non-public information. Obviously, one you touched on it. It does have to be anonymized from the aspect of being able to go down and track someone that deeply. A little different in finance, because we always talk about this, I don't think hedge funds want to know who you are. They want to know the number and the trend, and it makes no difference if it's Stewart or Tim or whoever. It's more of the trend that they're trying to figure out. So, I think that related a lot more to especially the health care industry and not being able to go too far down the rabbit hole to see who might be in testing or doing something like this or that. So, it's actually come a long way. There haven't been too many legal and regulatory issues. We had the app one a couple of years ago that wasn't even it seemed like it was more related internally to what they were doing than the general industry, I will say the SEC knows it exists and it's part of a lot of the different inquiries that can happen. Usually at one of our events, even the next one, we're doing a compliance seminar and the data sourcing and the people that come get a certificate of participation. So, what the SEC wants to know that you have, you're following the rules, that you have your policies and procedures and that you're doing them every year. So, you're doing that annual check. Are you doing these webinars or these seminars to teach the people, 'hey this is a red area, don't go there'. But in general, GDPR I think took it to the x-extreme, the US at some point will do it but we'll see it's not unfortunate it's not on the top but they're aware of it.
Stewart: [00:26:15] Jamie, one thing I'd like to add using American Express in the example of regulations, American Express would need to be very sensitive about what it was that they were selling of their data because anything that might predict their performance as a publicly traded stock would ultimately be non-public information or could be construed as that. So, they'd want to be very sensitive, and we would engage companies making sure that whatever it was that they were offering us would not allow us to actually understand the performance of them themselves.
Jamie: [00:26:49] I wanted to change tack slightly and ask about AI and the role that AI is continuing to play in finance. You mentioned earlier that data is coming so thick and fast now that we haven't got as many years to back test against. But the more years we have, the more maybe machines might be able to use that information and play the part of a fund manager. I mean, right now the edge that discretionary fund manager has is one of psychology. But I'm wondering in the future whether even the machines are going to be able to make that psychological jump too.
Stewart: [00:27:19] I think we'd have to go to what your definition of AI is. If we back up to machine learning, there's a huge role for machine learning. If we back into fully automated AI, I'm not sure, especially when it comes to asset management that investors are ready for that. There are there are certainly capabilities that exist maybe not in actually execution but in learning prediction models. I think quite a few firms are trying to capture the component you mentioned. Which is the psychology of it. So, I don't think you'd have to look very far to find giant hedge funds trying to codify management decisions, investment decisions, things like that. And that is all input for a process I think you're describing with A.I.
Jamie: [00:28:08] So we're going to wrap it up pretty soon. But Tim, I did want to ask, if you're not at somewhere like Jump Trading and perhaps you can't afford to get access to some of this data, how do you recommend people who are at home just running their own portfolios? How do they get access to some of this data, which if they've got Apple shares, I'm sure they'd love to know in real time how many phones are being sold.
Tim: [00:28:28] There's a number of data sets too that aren't just exclusive to the hedge fund world. I was reading an article this morning about TipRanks and they've come to some of our events. Even the jobs data stuff. There's a lot of publicly available stuff. It all comes back to where do you start and what question are you asking and what are you trying to solve? I think you can get a reasonable percentage of the way there and just become smarter and better with A, public tools and B, finding a couple of these ones that have user interfaces. Part of the of the problem, but also the advantage of if you have a ten person data team and you're able to purchase these data sets that are $10,000 a month or $250,000 a year, and it's your job you're going to have an edge because if I'm not focused on building a house, you probably don't want me building your house. Find someone that's really good at it and manage their money or find an ETF if you don't want to pay the fees. I think going that far down to the consumer is going to take a little bit longer. There are some firms trying to do it we tend to sit in the middle. Our service is $2,500 a month, but you get access to this large catalogue, you have sandboxes, you have three or four data hunters that are just going out and finding your specific need and trying to help you answer that question. I think that's probably the opportunity because the big funds do this, they have to do it to be competitive. I wouldn't want to go to an investor meeting and be like what's your data strategy? How many people are on the team and how do you plan on building it? And they are like, we’ll get to that later. I wouldn't give anyone money that didn't have a data team. as data is advantage. Your point earlier, there's more data coming every day. Do you want to make a better decision with more information or wing it? And I think that over time, data's the answer.
Jamie: [00:30:31] Well listen guys, this has been so fantastic. I've really loved chatting. I should also say for anyone listening, there's some fantastic videos on BattleFin's website. Tim has a show called The Data Drop, which I thought was particularly entertaining. And if anyone wants to get in touch with you guys that you have Twitter handles or anything.
Tim: [00:30:47] I'm just Tim@BattleFin.com. There are so many ways to get in touch with us. Our website BattleFin.com. Come to one of our events. I always just walk around being amazed at how many new vendors and stuff is out there. You can hang out with Stewart and I, we were just in London, and I have to say that was another great event because so much happens in different continents and people are looking at things in different ways. So, events are another great way.
Stewart: [00:31:11] For me, LinkedIn is the best way to reach out to me. I'm not very good at social media outside of that, but I do respond well to LinkedIn and always happy to talk about data. I hope people listening to your podcast thinking, I wonder if I should sell something. Call me!
Jamie: [00:31:29] There you go. Stewart, Tim, this has been such a pleasure. Thank you both so much for joining.
Steward and Tim: [00:31:34] Thank you. Thanks for having us.
Jamie: [00:31:35] Thanks very much for listening. And if you like what you heard, please give us a follow wherever you get your podcasts. The information contained in this podcast does not constitute a recommendation from any Refinitiv entity to the listener. The views expressed in the podcast are not necessarily those of Refinitiv and Refinitiv is not providing any investment, financial, economic, legal, accounting or tax advice or recommendations in this podcast. Neither Refinitiv nor any of its affiliates makes any representation or warranty as to the accuracy or completeness of the statements or any information contained in this podcast and any and all liability, therefore, whether direct or indirect is expressly disclaimed. For further information, visit the show notes of this podcast or refinitiv.com