Value Driven Data Science: Boost your impact. Earn what you’re worth. Rewrite your career algorithm.

[00:00:00] Dr Genevieve Hayes: Hello and welcome to Value Driven Data Science, brought to you by Genevieve Hayes Consulting. I'm Dr. Genevieve Hayes, and today I'm joined by Andrew Einhorn to discuss how he is using AI to help ordinary investors like you and I find better investment opportunities than we could ever manage on our own.
[00:00:21] Andrew is the CEO and co founder of Levelfields, an AI driven fintech application that automates arduous investment research so investors can find opportunities faster and easier. Before moving into finance, Andrew started his career as an epidemiologist and helped build a pandemic monitoring system for Georgetown Hospital.
[00:00:47] He's also previously co founded another tech company, Synoptus, and has also consulted for NASA and served as an advisor to a 65 billion hedge fund. Andrew, welcome to the show.
[00:01:03] Andrew Einhorn: Thanks for having me on.
[00:01:05] Dr Genevieve Hayes: Now before we go on, because we're going to be talking about finance and investments in this episode, I have to add the disclaimer that the content of this podcast is general in nature and is intended for educational purposes only.
[00:01:22] It should not be considered financial or investment advice. So with that out of the way on with the show,
[00:01:30] succeeding in stock market investing is all about timing, buying low, selling high, and being able to read the signs to determine when things are going to change. But as anyone who's ever tried to get rich through stock trading can tell you, this is easier said than done.
[00:01:49] Given the massive amounts of financial data published each day, for people who aren't experts in the field, it can be too hard to spot the patterns and keep up with the constant change. As a result, many people are either investing in markets based on guesswork or not investing at all. And this is where AI can help because there's nothing that AI does better than finding patterns in large volumes of data.
[00:02:20] AI has the potential to democratize access to investment insights. And we're going to be discussing how it can achieve that in today's episode. Now, finance and technology uh, two areas that have always fascinated me. As our regular listeners would already be aware, I originally trained as an actuary and I'm a lifelong finance nerd.
[00:02:45] By contrast, Andrew you came to be working in the fintech space by a more unconventional path. So you began your career as an epidemiologist. How does an epidemiologist such as yourself come to be working in the fintech space?
[00:03:05] Andrew Einhorn: It's been an interesting ride. Say that there's a long story and try to get a shorter version out there. The short answer is I'd love working with data. I love more importantly, answering questions with data. For me, it's about what are you solving for? And if you're solving for a problem, if you're understanding something that you didn't understand before, then that's really exciting to me.
[00:03:30] I didn't really know the financial field when I was younger. So I was sort of drawn to the healthcare side of, of things. And, you know, went to undergraduate. At Emory, and I thought I was going to be a doctor, and I found out later through college that I really didn't like being in hospitals or being around blood.
[00:03:50] So that was going to be a problem for being a doctor, and that kind of, you know, threw me for a loop, and I had to rethink, okay, what do I do with this love of science this love of, you know, being in health care, but not necessarily so close to the patients, and somebody whispered in my ear, try public health.
[00:04:08] Okay, so I had a big statistical background was, you know, undergrad statistical focus and research and my psychology and anthropology degrees and found myself working after graduating for a very large public health consulting company. They were. You know, doing a billion dollars of business, and they would get kind of outsourced contracts from the government that said, Hey, can you go figure this problem out?
[00:04:34] Can you go solve this issue that we're having? Can you determine you know, where the tuna fish is coming from that has mercury in it? And where did the mercury come from? And how did it get in the tuna fish? Because it's affecting pregnant women. We've all seen those sorts of labels. And so As you start investigating that, and you build up your skills for data analysis, data science, you start to be able to work with big data sets and understand patterns and recognize them.
[00:05:01] And I did that for years. I was very happy doing that. And then at some point, you know, went to graduate school. Focused on more engineering type of work, came out of graduate school, started consulting for a company where I was sort of the data consultant that would be brought into large government agencies to solve problems.
[00:05:22] NASA was 1 of them. They had a desire to build a metrics and monitoring system for environmental threats. And so I was the go to consultant for that to figure out, you know, how should they evaluate those threats in a numeric way? Likewise. You know, the F. E. A. 20 years ago was dealing with trying to figure out how they allow the private sector to launch their own space vehicles.
[00:05:46] They needed to analyze the threat to the human population of just anybody launching a vehicle. They need to look at kind of the. Environmental safety and occupational health risks and so they hired the company I was working for, which is to go and figure out what those threats were over a 6 or 7 year period before the government was able to look at the data, look at the science behind it and determine.
[00:06:12] Whether or not they're going to allow a space X or other companies blue origins to send their own rockets into space. And so the section that I was assigned was orbital debris. So I had to determine all the space junk and whether it was going to fall on our heads at our cars. If all of a sudden, you know, fast forward 30 years, everybody's launching rockets that are not reusable into space and parts are falling back to the earth.
[00:06:38] Was it safe? So it was a lot of math, a lot of data science. And my last assignment that I had was to build an event monitoring system for the defense department. They wanted to globally look at all the different threats facing. The bases that the soldiers were living on those could be environmental contamination issues.
[00:06:59] That could be safety risks like rockets, not being strapped down properly during tests and going skipping through the desert towards towns and they needed a system to do it. And we needed to build a software system in order to capture those threats. And that was when I was 1st kind of tasked with.
[00:07:16] Customizing a software system that was very kind of vanilla, not designed specifically for the military use. And I just absolutely fell in love with the process of building technology. I thought, wow, this is even cooler than analyzing data. I can actually shape. And form something without physically, you know, welding it or, or coming up with a different way to build it, like you would build a house.
[00:07:40] Cause I didn't have those skillsets, but this I could do, cause it was just based on needs, outputs, requirements, and where the data would be housed and stored. And so that kind of got me very interested in realizing that I can not only analyze data, but solve. Certain problems with software, so that led me down the path of creating a software company that looked at publicly available data from social media from news media and analyzing that data in a way that allowed for organizations to understand their reputation or how certain policy decisions were being evaluated by the general public.
[00:08:17] What the perception of the company was, for private and public companies, and we grew that company from a napkin sketch at a bar, literally a napkin sketch to sell it 10 years later to a private equity firm. But along the way, we had these big clients that were publicly traded stocks.
[00:08:34] ExxonMobil was a client. Discover, credit card was a client of ours and they were always having different types of issues. And these issues were events and those events were affecting the share prices. And so over time, over 10 years of watching hundreds of different clients have, you know, dozens of different events throughout the year, you start to see these patterns and especially kind of having a data mindset, you see the patterns, you're analyzing the patterns, you're giving software to your customers to analyze those patterns.
[00:09:04] And a lot of them were, event a happened, and this is how much the share price moved, moved 2%. It moves 5%. There was a cybersecurity breach at discover card and the stock drops. There was a new product launch at a company and the stock goes up. There was a train that derailed and it was carrying some kind of chemical and there was a spill and the stock price goes down 5 percent and our job was to just get the information as fast as possible to the chief comms officers so that they could kind of spin up their talking points and try to figure out how do they.
[00:09:36] Position the companies say, oh, wow. Well, it wasn't our fault that the train derailed. This company overloaded it with oil. We told them that could only have 60 tons and they put 80 tons and then it spilled all over the place. So it's their fault. And you'd see, you know, as they were able to do that, the share price would change.
[00:09:53] So over the years, you sort of got used to being able to get that kind of information really rapidly, we developed novel techniques and methodologies to source information from first parties to validate that they were true to identify kind of how information moved in a distribution pattern. If you ever read, a Malcolm Gladwell book, he talks about I think it was a tipping point, you have kind of your mavens and then you have your connectors.
[00:10:20] Right? And so we built a system where you can monitor that visually. You could see the mavens pushing information to the connectors, the connectors pushing it out and the journalists connecting with that information. And, we were just doing kind of very novel things at the time. And because of that, I got, Called up or, , ask for an introduction from someone who was working for a large hedge fund.
[00:10:42] And he was very interested in the types of research that we did looked at the software that we had and said, this is mind blowing what you guys are able to do. But you're in the wrong industry, you're working for peanuts, you're working for an industry that really has no money because most of the money is in marketing or it's in financial services.
[00:11:02] And at the time, it kind of took a hard look at it and said, yeah, you might be right. I don't know enough about financial services to evaluate that. He said, well, let's get you introduced to some people. I'll bring you around the hedge fund. You can talk to the traders. Just help me do a better job at my job as a portfolio manager.
[00:11:19] Why don't you outperform? It's like, if you teach me your methodologies for information aggregation, then, exchange, I could help kind of advise you to work in financial services or introduce you to people who can at least explain what they do and and then you can figure out how you might be able to help them.
[00:11:35] So I spent. Weeks, if not months interviewing traders, bankers, investment bankers, traders of all types, options, FX, bonds, stocks, and I was more working like an investigative reporter, just trying to understand the problems of the industry. And at the same time, I was helping this portfolio manager outperform his peers by, four or 500 percent just by showing how to get information quicker and how to validate information faster through some of the methods that we had developed.
[00:12:06] And when you saw it, the first thing he said was, you just figured out in 15 minutes, what has taken me 15 years in a career. To know who to listen to, and who's credible and how did you do that? And I said, well, it's just a scientific methodology for determining influence online and credibility. And, from that, we started just kind of spinning up ideas that, hey, maybe you could use your tool into financial services.
[00:12:32] And we were just really busy, we had a big client base. It was impossible to shift the business that much to a completely different vertical at the time. And we said, we're going to do this, we would need a lot of money, but we talked to our board and the board said, if you're going to bother raising a lot of money and going into that sector.
[00:12:50] You should try to figure out what your valuation is now. It might be worth, just liquidate or selling the company now instead of waiting and taking out a lot more capital and diluting the shareholders and then waiting for an exit 5, 10 years later. So we ran that test and, the board ultimately said, I think, , the better option is sort of bird in hand.
[00:13:08] And we had an offer to buy the company sold the company. And then I took a bit of a sabbatical as my wife called it, just to kind of regroup, figure out what was next. We were looking back in the technical team, and I were thinking about all the things that went right, the things that went wrong in the previous company, and we realized that, one of the limiting factors was really Boolean search, that it was just always became a problem to not be able to do context very well, or understand humor, satire, sarcasm, all the nuances that were proliferating social media speak and he said, what's going to solve this was AI.
[00:13:51] So we started kind of fiddling around thinking about, well, let's just try to see if we can solve that search problem. We didn't really know where we were going. But I always had this idea of like, maybe we should be in financial services because, we had some opportunities to do that before that we didn't get into.
[00:14:07] And so sort of experimenting on my own with, , could I take these methodologies that we had previously and apply them, just as an individual investor and then COVID hit, right. As that happened and we were searching for really like an industry and a problem to solve and COVID hit.
[00:14:24] The markets went nuts, everything tanked. People were panicked. We got a lot of calls, or I got a lot of calls that were, people worried about losing their savings, losing the college savings for the kids, losing their retirement and having actually worked the CDC through the last pandemic.
[00:14:41] Having been an epidemiologist and having been unemployed at the time because I had sold my company. I had time on my hands. So I went in and sort of spent 80 hours researching everything about pandemics. I wanted to know how does the government respond? Has there anything like this happened before? What kind of parallels can we draw from COVID from previous pandemics, even Zika or swine flu, or Ebola virus, or even going back to.
[00:15:10] The original in the early 1900s pandemic and trying to figure out what the response and the response time. And then after a lot of coffee and not a lot of sleep and like 80 hours of work, it came out and said, had these big spreadsheet models and said, Mark, it's going to be fine in 6 months.
[00:15:24] It's going to be right back to where it was. I got a lot of flack for that from family members and friends but that's exactly what happened and it wasn't that surprising. It was just based on the data. I just look back, I looked at how did airlines respond? How did cruise line stocks respond to Zika when they banned cruises out of the port of Miami?
[00:15:45] Which they did. They were closing schools. So there are small examples of what had happened in the past. And so all I have to say is that The conclusion was that events were changing everything. Everybody has a plan, and then an event comes and punches you in the face, and your plan is gone, and you have to really react.
[00:16:04] But, we had no data set that told us what was about to happen. And I thought that was a huge opportunity in the market. I thought that was something that was a calling for me, but I had kind of this weird set of skills that we could put together, you know, software design meets event monitoring meets.
[00:16:26] Epidemiology and all of a sudden it was like, aha. Okay. I know what we need to do. We need to get analytics around events so that people can make better decisions when an event happens and understand what's about to happen and not completely overreact or completely under react, which the market is. Has a tendency to do, and that's because people don't have data to make decisions.
[00:16:50] They're just going on good instincts often when something happens, and it makes everything nuts, including, the sell off that happened during covid. So that's sort of how I arrived at that conclusion that, okay, we have. Event driven system. We have event analytics. We have a software system that helps people understand how to make decisions when events occur.
[00:17:14] We have a system that allows people to find investments really rapidly and then not just find them, but see the forecast for what's about to happen with the share price. And the closer we looked at that, we realized. We were onto something pretty big because most of the market was driven by events and the rest of it was just riddled with opinion and kind of BS, uh, , people who said that they can figure out what the next Amazon is going to be if you just pay me 200, I'll tell you.
[00:17:45] And a lot of that gimmicky stuff that was just garbage can go away if you have access to direct information, if you have access to this type of analytics, which we knew only the largest, most Hedge funds in the world had access to because they built it in house. So we had a very clear mission and the mission was make this very high tech system available to everybody and level the playing field is where the brand name level fields comes from and sort of take all this institutional knowledge, all this data analysis, knowledge, and these methodologies that have been built up and just democratize access to it.
[00:18:24] And sell it for a couple of hundred dollars a year subscription service so that everybody can do it.
[00:18:29] Dr Genevieve Hayes: Wow. So it sounds like even though level fields can't possibly have existed for more than what, three or four years, is that about right?
[00:18:39] Andrew Einhorn: Right? Yeah,
[00:18:40] Dr Genevieve Hayes: The idea has been basically in the making for your entire career.
[00:18:48] Andrew Einhorn: it seems that way. Yes. We don't always know it at the time. You know, when these things are going to all get pulled together, I had always followed my interests. Sometimes it took me down interesting paths, like, epidemiology and other times it drew me into financial services. And, as you develop each skill set, it starts to come together later.
[00:19:13] And you start to realize, wow, I can do this and have a unique angle that I'm looking at it. And when you see that in use, and you have to get it validated a few 100 times, from beta testers and advisors, friends and family members. But, when the answer keeps coming back, like, yeah,
[00:19:30] you're actually really good at this. You should do more of it. , you're in a good spot at that point.
[00:19:34] Dr Genevieve Hayes: This whole concept of event driven investment insight is something that's actually quite new to me. I'm familiar with a lot of quantitative investment strategies, but a lot of those are things like trend analysis and stuff like that. So how exactly does event driven investment insight work?
[00:19:58] Andrew Einhorn: it's pretty simple. An event happens and the market is driven by people and people react to that piece of information. If you know how they're going to react positively or negatively, then you can profit from knowing the directionality of the event. The reaction is a very much like human psychology.
[00:20:19] So, for example, Jeff Bezos leaves Amazon quits as a CEO. Is this the end of Amazon? As we know it, you know, question mark, question mark, big, bold headline predictably. As with almost all the cases of large technology or even just large companies where they lose a CEO, particularly a founder CEO, the share price drops on the first day, about 2%.
[00:20:43] Then over the next month, it's going to continue to slowly drop as the doubts are sowed by newspapers who are taking advantage of the situation to. , sell clicks by saying, Oh, is Amazon dead? Is this the end of growth? Can the new CEO do it? Tons of great opportunities to sell articles that mean absolutely nothing.
[00:21:04] And that drives and sows doubt into investors. And so the share price will usually go down for a month or two before starting to when the buyers come in. So selling ends because everyone who's going to sell sells and the buyers come in and within two months. The share price is usually right back to where it was at the time of the announcement of the CEO departure.
[00:21:27] So venture investing will tell you that on that event, you could do one or two things. You could short the stock or you could wait for the dip. And if it dips down 10, 14%, you buy that dip. And you wait for it to rebound and you make 10 to 14 percent within two months, which is a return that most people would hope to make within two years in the market.
[00:21:48] And so what you do with that is you obviously make the money, but you also limit your exposure to just general market turmoil that can happen. Because the longer you're sitting in a position, the more likely there is a war that's going to break out and affect the share price and politician comes in and change the rules.
[00:22:08] Or there's going to be a pandemic or there's going to be a natural disaster. All those things that can change your share price and the value of your investment without having anything to do with the fundamentals of the company. So I'll just market sentiment ships. And so the idea is instead of. So falling prey to those events, you use them you take advantage of them to find pricing discrepancies in the market, looking at when stocks are undervalued, oversold or overvalued because they're overhyped you look for things that are changing the way we do business, right?
[00:22:45] COVID came, COVID was an event. What did COVID change? Well, not a lot of people are back to the office anymore. So short term, what would that mean? It means more Lululemon at home and less, fancy shirts from Nordstrom's. It means more zoom and less gym memberships, so Peloton was great for a while and now it's terrible.
[00:23:06] And, those are the sorts of trends That events change and they change the trajectory. And so if, understanding the correlations or inter linkages between things of how an event is going to impact an industry or sector or particular company, it's actually quite easy to determine what to do with your money.
[00:23:27] I can make a quick amount right now, or I've been watching Amazon for years, and it's always been, too expensive for me, but all of a sudden it's on sale. But wait a minute, the CEO left. So maybe it's dangerous. Well, I can look at this database in level fields, and it shows very clearly that after 2 months, most all the companies are right back to where they were that share price.
[00:23:47] And so you have the confidence that comes with it. You don't have to do this just by sheer gut instinct anymore. And it's been largely missing from the market for a long time. There's a lot of people who invest in the basis of momentum or candlesticks. And you look at different patterns and all those patterns are either driven by events or they're screwed up by events.
[00:24:10] So, when there's a bullish engulfing, candlestick, it's not just because all of a sudden people decided one day to buy a stock, there's an event that drove that change, changes it, you see it every day. And so the system is looking for. Events that are material to the share price, and it looks at 6, 000 stocks at once, and it's reading through 30, 000 documents a minute, determining what kind of event just occurred, and then pulling out the analytics to say how this will probably impact that particular share price, or at least creating, an analytics array where you can look at past events have affected it.
[00:24:49] And it prevents bad decision making just as much as it helps making money, for instance, we have an event type on there, which is about Tesla product launches whenever they launch a new car or product the prevailing viewpoint would be their stock's going to go up on the excitement.
[00:25:06] Actually, it's about 50, 50, 50 percent of the time it goes down because people are disappointed and the hype was too much and 50 percent of the time it goes up. If it's, a success. And so we get a lot of kind of flack for putting that on there. Cause some of the users like, it's 50, 50, this is terrible odds.
[00:25:23] Why are you putting this on there? And our response is we put it on there so you don't make bad decisions because knowing what not to do is important too. We don't want you chasing after the next test, the product launch, just because Has the name Tesla in it, , look at the data. And also for sophisticated options traders, there's still many ways to make money from that volatility, if you know what to do, so.
[00:25:44] I'll stop there because I'm sure you have questions, but hopefully that answers,
[00:25:48] Dr Genevieve Hayes: For something like this to work, you must have a massive data set underpinning this. How big is your data set?
[00:25:54] Andrew Einhorn: At this point, I don't even know we're terabytes of data we're processing it's, it's huge.
[00:25:59] Dr Genevieve Hayes: How many years does it go back?
[00:26:01] Andrew Einhorn: It depends on the data that we're looking at. We have stock data that goes back to like 1910 , and, different profit and loss statements and things like that. Share price movements for our data set with the AI, the linguistics layer.
[00:26:15] Okay. Goes back 5 to 10 years, depending upon, the event type that we're looking at.
[00:26:22] Dr Genevieve Hayes: The AI and data science that underpins this fascinates me because I can imagine how you could do this sort of analysis as a manual investor, like you described that analysis you did after COVID started. Buzz. I don't even have the first idea how you would build a artificial intelligence model that would be capable of doing all this analysis on its own. What sorts of AI do you use under the hood in this product?
[00:26:50] Andrew Einhorn: Well, so if you think about it, in kind of lay terms, our AI is more like a speed reader. It's really for analysis of unstructured text. And so we had to build our own language model. To understand financial services in that space, particularly, because there's a lot of nuances.
[00:27:12] We couldn't use something that was broad based because in finance, there's just oddities that happen. For instance there's a company called blue bird bio, and this kind of thing happens a lot where you've got company names are the same as creatures on the planet. So, like vanilla off the shelf kind of language model would not be able to discern the difference between a blue bird.
[00:27:33] That's referred to as a company and a bluebird that's flying in the sky. So we had to develop those types of linguistic markers and algorithms to determine those things with kind of proximity to other types of terms. And using different types of numeric algorithms to determine the likelihood that this was, in fact, a company versus, an animal that's flying around.
[00:27:57] So if you take that, and then kind of multiply that use case by, like, 10, 000, that's kind of what the system is doing. So that's part 1. Part 2 is we had to figure out not just, what was an event, but what was an event that actually materially moves the share price. That's a combination of knowledge, testing and, allowing for a platform where we could discover it ourselves if we didn't know about it.
[00:28:25] We could throw in a couple ideas and then get a data set and see, does this actually move the share price? Is there a correlation here? And, many times it comes back, no, there wasn't. And so we said, okay, well it's noise then, like it's an event. It might be interesting, but it doesn't predictably move the share price.
[00:28:43] And so we omit those types of things. And so we have this large filter that starts with, somewhere between 25 and 30 million events. And narrows down to about 000 events per year for the stocks that we're monitoring that we know are actually material. They're going to move the share price north or south.
[00:29:04] And the rest of them, we still have in the database we store, we can analyze them for different things. But this is sort of what we came out with, at the onset, and we're thinking really hard. Okay. During our beta test period of like how to make this a really easy to use interface, so that you didn't have to be a data scientist to do this.
[00:29:22] You could just be a regular person with no data background. And we wanted it to be as simple as looking at the weather. And that's sort of the best analogy I can give. It's like, when you look at the weather forecast, you know, it's going to rain, where it's going to rain, how long it's going to rain, roughly.
[00:29:39] The percent chance of rain, we're doing the same with events. The event occurs, you know, the share price is going to move. How long is it going to move? What's the probability it's going to move that way. If you can read a weather forecast, then you can read level fields. It's quite the same.
[00:29:54] Dr Genevieve Hayes: So from what you're saying, the way I understand this is that you've got a tool that takes advantage of the fact that AI is really, really good at pattern recognition. It's got massive amounts of data and use that to find events that are highly correlated with either upward or downward movements in the stock price.
[00:30:22] Obviously, you can't guarantee. What will happen after a particular event, but your tool is able to alert people to the fact that after an event has happened, it's one of five or 6, 000 events that it's seen before, where there is a high probability that this will lead to it. an upward or downward movement in the stock price based on correlation?
[00:30:53] Is that about right?
[00:30:55] Andrew Einhorn: Yes, that's right. And every event down there has a win rate, which is basically the correlation. We'll call it a win rate, just to say that the expected direction is bullish, stock price going up and the win rate is 90%. That's roughly the correlation. You're looking at a 90% match. Now, it doesn't take into account some of the confounding events that might happen.
[00:31:18] Meaning you could have a bullish event happening on a day where, some massive flood hits Manhattan in New York and, , All bets are off that that event is still going to be positive. So we have to accommodate for those types of things in the correlation, but often without those intervening events, many of them are 100 percent correlated.
[00:31:41] Dr Genevieve Hayes: And even though it's correlation analysis rather than causation analysis, and every data scientist has it hammered into them that correlation does not equal causation, if you're making enough bets on the stock market. It doesn't matter that you're using correlation rather than causation. You are making bets where the expected value is positive.
[00:32:06] So you should end up winning in the longterm.
[00:32:08] Andrew Einhorn: Yes, and I would say it also passes kind of the common sense test. If bad news happens, you would expect people to sell some of the shares. Good news happens. You expect people to buy some of the shares. So there's sort of that. Yeah. Like it's clearly the cause so long as the price movement is significant enough above the normal price movement of the share price, right?
[00:32:33] So if by chance Tesla goes up and down 2 percent today and you capture enough for those events Then you have to say, well, if it drove the share price 7%. Then you're looking at standard deviations from the mean of 2%. It's like, you can do, normal statistical analysis. And so we see those patterns as well.
[00:32:54] We have to kind of guard against some of the noise that might be out there. Like, for instance, we have this scenario, it's dividend increase. On the platform. So company increases the size of its dividend, the increase in size of the dividend, 2%, nobody really cares, you know, business as usual, doesn't move the share price much, and likewise, , the average volatility of the shares about the same as that movement for that 2 percent dividend increase.
[00:33:21] But when a company comes out with 105 percent dividend increase and the share price jumps 14 percent that day. You're well above your standard daily move in the price, and it happens over and over and over again, and we see, some absolutely amazing moves in the short term. And we've seen as much as 600 percent price increase in a day on some of the events.
[00:33:45] These are companies that, let's say, was a small company. Nobody paid attention to the share price, and all of a sudden it's a billion dollar contract. Now it's worth. , 6 to 10 times the amount or a company that comes up with a new drug discovery and they get it to market and it goes from being a, clinical company to now a commercialized product able to source a 10 billion dollar market.
[00:34:08] So those are absolutely events that are causal that you can put your finger on and say, yeah, of course you would buy this. Like they just found the cure for Alzheimer's. You're going to buy that stock and you can say that's the cause. So those are things that we look for. And I think putting it simply, everybody's been in that seat where they're watching TV, they hear from a friend that some stock is up huge.
[00:34:32] And they're like, well, why can't I find this earlier? And it's largely just the way the market works. It's like, as an individual, you can't monitor so many stocks and so many events by yourself. So then you rely on the news. To filter it for you and new staffs are short staffed. Usually they focus on what drives the most clicks for their online ads and subscriptions.
[00:34:56] And that's talking about gossipy things, like, how many kids does Elon Musk have and, whether or not apple is going to come out with a car, things that really. They're interesting, but they don't necessarily make any money. And so you have this fixation on the magnificent seven or the Fang, it's always a small group of stocks that gets 80 percent of the coverage.
[00:35:19] And meanwhile, you don't hear about all the rest of these great companies that are out there grinding it out, coming up with new things that they're doing, changing the trajectory until. They become a 50 billion company. Then all of a sudden you start paying attention. Then maybe the ride was they're up a thousand percent.
[00:35:37] And so being up a thousand percent makes it interesting to suddenly talk about. And that's when you hear about a super micro computer for the first time. But you're late. And we want to get the information directly from the source. We're trying to get information to people. So they don't have to be reliant on the news to filter this information for them.
[00:35:55] And we want to help them. Come up with faster, easier ways to get that information. So you don't have to spend all day staring at a screen, I should say most of what's on the platform is in the form of an alert, a couple of clicks, and it will monitor, even literally an AI search agent monitoring what you want to look for all the time, and then sends you an alert when it finds it.
[00:36:15] And you say, all right, do I want to invest? Yes or no. So you don't have to sit there.
[00:36:20] Dr Genevieve Hayes: So you mentioned it'll monitor what you want to look for. Suppose you don't know what you want to look for beyond the fact that you want to look for things that are going to make you money. How would you? Program the tool so that it doesn't constantly ping you with alerts.
[00:36:36] Andrew Einhorn: Well, so on the platform, we have ready made event driven strategies. You just browse them and say, , what kind of strategy you're looking for. Some are long term, some are really short term, some are midterm. And that's noted on each strategy card, if you will. She gives you an overview of, you know, how fast does it move?
[00:36:54] What's the average move? What's the win rate? What's the average hold time? And so with that, you can just kind of browse and say, this one looks easy enough. I just buy it and hold it for 12 months because I'm not going to be day trading it. And then you literally click one button that says turn alerts on and the alert is going to you.
[00:37:12] Now, if you know a little bit more and you want to customize these scenarios, you can. For instance, you might say, I like this event type, but I never going to invest in a company like a raw materials company. I'm never going to invest in utility companies because I think they're boring. So you can just kind of filter out those say, I'm really interested in tech, healthcare, financial services.
[00:37:36] And so you customize that strategy just with those 3 clicks, click a button, and then you can determine. , what time of day you want the alert and it will send it to you. You can obviously do more more things with it. That's like, the very basic. But if you're more sophisticated, you could set things like.
[00:37:54] I only want to get events that happen in the morning because I don't have time to react at night. You could look at the financials of the company and say, I want to invest in companies that have. Earnings growth of 10 percent revenue growth above 20, you know, valuation of less than 40 price to earnings ratio in the following industries, or you could just let the system, , send them all to you and kind of pick and choose as they come.
[00:38:19] So it's really flexible. We try to make it like booking flight.
[00:38:23] Dr Genevieve Hayes: Is it just the U. S. stock markets that it deals with or do you have other country stock markets on it?
[00:38:29] Andrew Einhorn: It's the U S exchanges right now. But because it's the world's largest exchange, most large companies around the planet. Have a listing in the New York stock exchange or NASDAQ. So some big Australian companies will be listed as ADR as big British companies, like AstraZeneca will be listed there.
[00:38:48] And at some point we'll expand different exchanges, but, if you're looking at the largest market cap companies by country , top 10 or 20 are definitely going to be in there.
[00:38:56] Dr Genevieve Hayes: So for someone like me who's in Australia, I could set it to just focus on the Australian companies that are listed on the US stock exchanges.
[00:39:06] Andrew Einhorn: You could, yeah, you create a list of those companies that are interested in and then set it to only look for those companies. This is 1 of the features that we have. You can have a watch list of companies that you're interested in.
[00:39:16] It might be ones you own. It might be ones you're thinking about buying. And then you just say, at least 20 companies, if this bullish event happens, let me know if this bearish event happens to take on the share price. Let me know.
[00:39:26] Dr Genevieve Hayes: So yeah, that sounds really useful. The other thing that you mentioned a few minutes ago was you're saying that you want these Events that you're alerting people to, to pass the sniff test. Does that mean you've got a human in the loop as well?
[00:39:41] Andrew Einhorn: We have some validation that has to happen for the training of the AI and, , certain scenarios require a little bit more, particularly if they're from new sources. So, we find that news is not quite the best source of information because it could be rumor and the rumor could be complete garbage, but they put it out anyway as fact.
[00:40:06] So for certain scenarios, yeah, we have more of a human validation to say, did this really happen? Are they someone just kind of doing a pump and dump here?
[00:40:14] Dr Genevieve Hayes: So a piece of information that was released as a press release by the company itself would have more validity to it than something that was tweeted on X, for example.
[00:40:26] Andrew Einhorn: Correct. Yeah. And, similar, you see these sort of nebulous articles that will say, according to sources, familiar with the matter, this happened, sources are not named we can't verify any of that information. So we're trying to get away from news altogether, a hundred percent. And so I would say 90 percent of what's on the platform is driven straight from the company itself.
[00:40:51] Now, not to say the companies don't lie either. I have to say, there are obviously cases where companies will commit fraud. That's harder to guard against. There are, some ways to look for that.
[00:41:03] Dr Genevieve Hayes: So what's next on the level fields development roadmap?
[00:41:07] Andrew Einhorn: So we are rolling out a new version of the platform. Hopefully in the next few months, it's going to give, More control over the system. So users can kind of come up with their own hypotheses and test those. We'll have some more kind of ready made strategies for individuals who might want to have a longer term view.
[00:41:26] Sort of think about them like re filters for looking at different stocks. So, if you're just trying to browse casually. Not necessarily, , looking for events to trade, but looking for longer term. We have some solutions for that as well. So those are the main pushes right now. We also have a more premium version of the system where you get more access, more flexibility, and you get actually some human kind of editorial.
[00:41:56] Favorites, if you will. So you have analysts that are looking at the platform and then we see all these events. They'll flag the ones that they think are going to be the most successful for level 2 users who don't really want to do this themselves, but love the idea. And the idea is, you have somebody who's an expert on the platform and expert in interpreting these events the flag it and put kind of a.
[00:42:19] a synopsis of their thinking behind it and send it out. And so those have been wildly successful over the last year and a half. The returns are actually 3500 percent for the cumulative level two alerts, which is just baffling to even say didn't think we'd get even close to that, but we've been very successful , in looking at those opportunities.
[00:42:42] And if you're playing them enough, we send about a hundred a year. They add up quite a bit.
[00:42:48] Dr Genevieve Hayes: What final advice would you give to data scientists looking to create business value from data?
[00:42:55] Andrew Einhorn: I would say, think about interoperability. I see a lot of. Products in the market, not a lot of them talk to each other very well. So you have, , 1 or 2 companies like a Zapier that's there to kind of integrate, but they're clunky. So, if you're thinking about it, and you have a solution or a data set also think.
[00:43:16] A little bit bigger in terms of the ecosystem of other data sets that could augment your data and how you can make it more valuable. That's sort of what we did at level fields where we thought about, solving 1 problem, which was unstructured text. Then realize if we mesh market data with that, then you get concoction with the price moves, making it a lot more interesting and forecasts.
[00:43:38] And so we're still looking for ways that we can adopt or integrate with other products, services to create additional value adds, and they might come in surprising ways. So, , I would say look for that. A lot of people want to solve one problem, but a buyer wants you to solve all of their problems.
[00:43:58] You have to think about it from the buyer perspective. They don't want to have 20 different tools open and have to have 20 different tabs open on their system. So, if you can find something that integrates well, that's going to make it easier.
[00:44:12] Dr Genevieve Hayes: For listeners who want to learn more about you or get in contact, what can they do?
[00:44:17] Andrew Einhorn: Go to levelfields. ai. You can read tons of case studies, some of which I cited today. You can look at the platform, get a demo, maybe a YouTube account. We're putting out, free tips and information on stocks all the time, so you can get some of there. And if you want to sign up. If you're listening to the podcasts, and you're exercising or you're doing laundry, just, send yourself an email levelfields.
[00:44:40] ai. You can use a promo code podcast the word podcast with 23, the number at the end for a discount.
[00:44:48] Dr Genevieve Hayes: Excellent. Well, thank you for joining me today, Andrew.
[00:44:52] Andrew Einhorn: Oh, thanks for having me. This was fun. I appreciate the questions.
[00:44:55] Dr Genevieve Hayes: And for those in the audience, thank you for listening. I'm Dr. Genevieve Hayes, and this has been Value Driven Data Science, brought to you by Genevieve Hayes Consulting.