Wayne Ferbert has over 30 years of experience in financial services, with a focus on innovation and risk management. He is the founder of Alpha DNA Investment Management, where he applies machine learning to Hedged Equity strategies.

Previously, he co-founded ZEGA Financial, built on principles from his co-authored book Buy and Hedge. Earlier in his career, Wayne led Business Development and Product at TD Ameritrade, and held strategic roles at leading insurance and banking firms….more

Dan: Hello and welcome to QuantBeats. I’m Dan Hubscher, managing director and founder of Changing Market Strategies and your resource for all things quant. You know, my co-host, Radovan Vojtko, former 300 million euro quant portfolio manager and current CEO and head of research at Quantpedia. Rado, please say hello.

Radovan: Hello, hello everybody.

Dan: You can check our backgrounds on the QuantBeats website, or in episode one of the podcast, go click the Q logo in the corner of your screen. Today we’re joined by Wayne Ferbert. Manager of Alpha DNA Investment Management. Nice to see you Wayne. Please just say hello.

Wayne: Yeah, how are you doing? Nice to be here guys.

Dan: Thank you. Now just before we get to Wayne’s background and start to find out how decoding sentiment with quantamental strategies seeks to generate alpha, I need to show you some quick disclosures that are on the screen now (watch the full episode for the disclosure). Since I’m a registered representative. We’ll be giving you ways to contact me at the end of the episode and you can contact me with any questions.Alright. Now let’s get to it. For seasoned investors, aspiring quants and anyone wanting to dip gently into the world of quantitative finance and give them some insights into quant strategies, knowledge of market dynamics, and a glimpse into the future of algorithmic trading. So to kick us off, Wayne, can you please give us the one minute version of: who you are? What you do? And how have you been doing lately?

Wayne: Let’s see what I do. I mean, other than the last weeks talk, clients off the ledge or, you know, convince them to get out of the fetal position. Yesterday helped a little bit I guess. It’s a few days here after the Trump tariffs and the markets have gone a little, a little haywire, but all levity aside Alpha DNA Investment Management.

We’re a quant based trading shop or investment management shop. We basically, you know, live at the intersection of machine learning and alternative data. We use alternative data to get, to create more accurate forecasts on revenue and EPS. We basically see demand changes for companies products before the market does, and we use that research to, to build smarter portfolios, long portfolios, long biased portfolios, and even equity market neutral portfolios.

Dan: Okay, thank you. So that’s good for now. We’ll get into it a little bit more with each question. So Rado please, over to you.

Radovan: Okay, thank you. Thank you, Dan. So I have like first question and it’s the usual question. How you get or got into this quant world. So why quant world? Why machine learning? Why alternative data?

Wayne: I actually live on the portfolio management side of our business, right? About 10 years ago, I met two data scientists. And actually, gosh, time really flies. It’s closer to 14 years now. I met two data scientists, Rod & Arun, and they said to me: “Wayne, if you can track the digital footprint of a publicly traded company, you’ll gain insight into who’s winning and who’s losing.”

And when you juxtapose that against what Wall Street thinks about who’s winning and losing, you’ll find trading opportunities both on the long and short side. Now, I had just a few years prior to that, left a decade of running product development for TD Ameritrade. I was one of the biggest internal consumers of their information about how our clients engaged with us.

Right. And when Rod and Arun said that, that they were referring to the digital footprint, what they meant was the actual interaction between customers and the digital brands of the companies, right? The, the websites, the Twitter handles, right? The social media pages. And so I already knew that interaction and the measurement of that interaction was a great indicator of who your most profitable clients are. Or even just like , whether you’re having a good month or a bad month, more interaction generally is positive for any e-commerce type of business or even these days, any business that has to have a website just to even explain what their product does. So knowing that at being one of the biggest internal consumers of that kind of data at Ameritrade for a company with 7 million clients at scale. Right. I knew that what they were saying was going to be a very interesting investment approach, but I knew it was gonna take a while to build it. So we eventually, they left their, at the time they had a full-time job in Fortune 500 companies. I saw the potential and said, listen, I’ve got the portfolio management chops, you’ve got the data chops, right?

Let’s merge the two and see what we can build. For 10 years now, we’ve been running. Live money doing that and it’s, it’s exciting. It’s the, it’s cutting edge, right? It’s the frontier of where investment management is going, especially the use of machine learning. And also the idea that alternative data can be a proxy for, you know, real engagement and real demand is,I think, a really interesting and unique way to tackle the markets.

Radovan: I like, I like this answer. I mean, I looked at your webpage, so I looked at your digital footprint before we started this conversation. And I like, the way how you look at this from your perspective. I also consider the alternative data one of the few. Let’s call it one of a few last venues where we can find uncorrelated alpha.

Because it’s not very crowded and you can mix it together and there are like, lot of the alternative datasets, you have probably more information about that. What are the important datasets? Because there are like, I don’t know , 100 different types of alternative datasets that I can choose from.

Wayne: It’s a great question. So first you know, we organize our data into four categories, right? So there’s just the sort of the direct web properties. Think of it as the traffic that goes directly to the company’s websites. There’s search results. So key brands or key terms associated with the company. If it’s Amazon, it’s you know, Amazon Prime, right? Or Alexa, right? They have lots of brands. Most people might not realize though, that they also own mattresses.com, soap.com. A lot of people know they own zappos.com. They bought that shoe company. I could go on and on. They own, I think they own barbecue.com.Xthere’s so many brands that they own. And so those brands matter. We basically track every sort of billion dollar brand and who’s searching for it and why. There’s also social media, so Facebook, Twitter, TikTok, Instagram, all those important parts. And then lastly, there’s app usage.

App usage is actually the newest of the data categories that we track and think about. When we started actually using this data a little over 10 years ago to invest there was, we weren’t buying any app data from anyone. ’cause apps were not the primary way that people engaged with folks?

Right. And so the data sort of fits into those categories, but it is the consumer metrics and the consumer volume metrics of the actual interaction between the customers and the digital brands of the company or the search for the company. But we don’t care about, you know, if they are saying nice things about the company.

We also don’t get any personally identifiable information. We don’t know that Rado specifically looked for the company or search for it. We just sort of, you know, get the aggregated: how many people searched for that term? How many people clicked through on it? How many clicked through on a competitor, right?

Like that’s the kind of consumer vector metric and volumetric data that we’re gonna care about. We’re always adding more data. We’re always taking data out. Data sources go away periodically, but that’s sort of the category and clarity data that, how that data can be transformed into vectors of marketing analytics.

We really solve this problem from a marketing perspective, you know. It’s about market share. You know, the, the slope of your growth. Does it look like your engagement measures are seeing momentum or achieving higher attractions? We turn it into different vectors of analytics, but then the most important part is do changes in those vectors correlate to past changes in either the revenue or the earnings of the company.

That’s what really gives us the signal. I mean, that’s what machine learning is looking for and what makes it so capable is its ability to see patterns that the naked eye couldn’t see or that regression analytics couldn’t see. Right. I mean, that’s really what we’re looking for is data that’s informative around the actual traction that the company is achieving, with its clients.

One other thing to add about this is organizing this data is not a trivial exercise, right? I gave a little bit of example with Amazon. If you wanna go buy all this consumer metric data, and we buy it from lots of sources, right? We spend millions of dollars on it. And so if you wanna go buy that data.

You have to know the URL you’re asking for, you have to know the brand you’re asking for, right? And so you have to pass it through an API to the vendor and they pass it back to you. They don’t just give you their whole database. Imagine an entire database of every single URL in the world, the traffic associated with it. It would be way too immense for our systems to handle it. We only are gonna pass to them the URLs or the brands that we care about. But these are the URLs that are associated with the app, right? Because remember, apps are just APIs passing data back and forth, right? So it’s the URLs associated with those APIs.

It is the URLs of the Instagram page, the Facebook page. Again, from those apps it’s those key brands, but it’s also the webpage, the most important parts of the webpage. Most important parts of the frame. We have a team that just keeps that data fresh. You know, that’s always asking, what are the billion dollar brands at Amazon or Walmart or Target or, and it’s not I keep mentioning consumer brands, but it’s also John Deere, right,who is mostly gonna be a B2B type of provider, right? They have lots of websites related to supporting their products and how people buy their products. There’s pharmaceuticals, right? We cover pharmaceuticals. If doctors are prescribing a drug more, there should be more traffic to that drug’s website, right?

People frequently don’t buy drugs online. I mean, it’s in, it’s increasing the purchase of prescription drugs online is definitely increasing, but still for the most part, they get ’em filled at their local pharmacies. Let’s say you’re prescribed the drug. When you get home, you’re gonna look, you’re gonna go on the website and you’re gonna look up what are the side effects, right?

How effective is it? You might tell your husband or your spouse, they might go look at it. You might tell your children, they might go look at it, you know, ’cause, and that’s gonna create, that sort of traffic, let’s say a drug is just being prescribed more by doctors. That traffic’s gonna be seen, and that’s what we’re looking for is a change in that traffic pattern that correlates in the past to a change in the demand for that company’s products.

Radovan: I have like two questions. I’ll start with the first one. So the first one is, okay, if you have so many alternative data, do you use also like traditional data? I don’t know, price, volume, price-trend, price momentum and stuff like that? Or do you see that alternative data trumps those price action or fundamental data. So, are the alternative data so good that you do not need ordinary price data or, I mean, there is like a combination, so you need to combine those two datasets together?

Wayne: Yeah, good question. The single vendor that we pay the most money to is FactSet. But we don’t really buy any alternative data from FactSet, so that’s just sort of answer your question. Traditional data is still a really important input to what we do. It’s not the part that’s giving us the insight to who’s winning and who’s loosing necessarily. So the data, we buy a lot of fundamental data. Like I said a minute ago, you wanna see a correlation between a change in that digital footprint and the change in the revenue or the earnings of the company. So obviously I need to get the revenue and the earnings of the company from FactSet or from, you know, FactSet is the best providers they’re still considered sort of the, the leading company in the earnings tracking, right?

Sort of aggregating analyst views of companies and, and their models for revenue and EPS. And so we collect a lot of fundamental data from them, right? Sales revenue by month, by quarter, to the extent companies break it out by month instead of quarter. Obviously every quarter a company has to announce its earnings.

That’s a key feed into the comparison. ’cause what we’re eventually doing is we’re determining, hey, we think Reddit, you know, just as a stock, let’s use ’em as an example. We think Reddit the stock, we think their engagement projects to a 9% growth in revenue this quarter. But all the analysts are projecting 4%.

So, so we’re at 9% based on our AI models. And the analysts are at 4%. So what we’ll do is grab all the data from FactSet, look at all the analyst data, see that they’re at 4%. That comparison is what’s going to create a buy signal for us.

Wayne: 5% would be a very material gap for most companies. And we are right a lot on our forecast for those companies.
We’re traditionally accurate, you know, to a close to 90% for calling surprise in the very tail of what we project, right? The sort of top 5% are most attractive companies. And so, and on either info. And so if we’re gonna be 90% right and we’re projecting it’s 9% and the analysts are 5%, that 4% gaps material.

We have a lot of confidence we’re gonna right. That’s what’s gonna create the buy signal. Or if it was inverted, it would create the short signal. So yeah, we buy a lot of fundamental data. Price and volume is not important to us.

Wayne: We don’t engage in a lot of discussion around, is the company trading at a discount?

Does it have price momentum down or up? Yeah. There are no technical analytics, but you should know this is a really interesting insight. Right. You should know. There is a high correlation between digital momentum and price momentum, right? So I wouldn’t wanna be doing extra price momentum work. I’d be doubling down on price momentum.

There is already a high correlation between digital momentum and price momentum. It’s an interesting insight, right? It shouldn’t surprise you that much. If a company is really executing online, it’s selling a lot more of its product. It’s the slope of its growth, it is really improving in terms of engaging people online and therefore selling more products.

More people are gonna see that, and those are stocks that a lot of investors are gonna start buying and leaning into. We are just good at getting that insight a little bit earlier than everybody.

Radovan: We are basically looking for a gap between alternative data and traditional data.

And you are faster than the others. So there are other people who are doing something similar to you. Maybe they’re not using the same data and it’s causing the price momentum afterward when all of the people start to realize: “yeah, this is the company and it’ll have a great earnings announcement in the future.”

Wayne: Yeah. And in fact, it’s even, we don’t talk about this a lot, and Dan, you and I even talked about this, but I’ll tell you guys just now. We don’t just analyze the historical patterns of how the companies’ changes in the digital footprint project to changes in their revenue and EPS. We also look at how changes in the analyst expectations for revenue and EPS correlate to the changes in the digital footprint.

Right. Meaning, remember the analyst community is aggressively trying to get at a lot of the same data that we get at, right? And so they’re sort of doing a digital channel check also, right? And so one of the things our AI has been trained to do is sort of say: “Hey, let’s find the companies that tend to have a pattern of having information that as it gets into the market, tends to move the analysts”. Right?

The actual objective function that we’ve designed around is likelihood of analyst change. It’s not likelihood of surprise. Tthat they’ll be surprised at earnings. It’s likelihood that the analyst will upgrade their revenue and EPS view. So it’s actually revenue or EPS view.

But we like to see when they upgrade both. And that is because. Now if a company’s going to surprise, there should be upgrades in the buildup to the surprise, and then if there’s a surprise, there should still be a gap. So there’s a high correlation between change and analyst upgrades and the actual surprise event from the company.

But remember, every time an analyst upgrades a stock, traditionally it’s gonna probably be rewarded a little bit. The stock’s gonna go up a little bit, but we’re actually looking for these chances where the analysts are sort of, they sort of see the changes too, in the digital footprint, but they see them slowly.

Right. And we see them ahead. So the AI also looks for that pattern and then tries to, like I said, find the matches. Right?

Radovan: Yeah. Sorry. You’re a front running analyst basically? Not directly, but you are trying to out guess them.

Wayne: We’re arbitraging Wall Street analysts.

Radovan: Yeah, exactly.

Wayne: There’s no doubt about that. Yep. And listen, they move markets and as long as they’re split markets they’re a great group to arbitrage.

Radovan: Definitely. And here is another question. So you mentioned the deer and industrial companies et cetera, et cetera. But I still think from what I listened to you, that there will be huge sector differences. So difference between the sector, like consumer staples, consumer discretionary, where I can imagine that those data are the most useful and other sectors that are more B2B, like, I don’t know, energy. So do you see these differences or those differences between sectors?

Wayne: Great question. And we get this question all the time. You should just know that across every single sector we create consistent stratification between the winners and losers, right? In terms of beating Wall Street analysts and not. But we also engage in an exercise in which this just is another way that makes us very different than most quants or even most Wall Street investment jobs, is we actually score the company similar to how a credit bureau scores a pool of people.

Right. When you have a credit score, if you’ve got a really great credit score, the 800+ credit score, right? Your score is actually relative to everybody else’s, right? The stratification from like the 200 level or 300 level up to the 800 level is actually sort of like, if they were given a pool of people, that’s sort of how they would stratify ’em.

And so across millions of people you have the, that sort of stratification. We are actually creating like a credit bureau approach to how we score stocks. So we score them within market cap, which means, you know, B2B companies are scored against B2C companies. We’re trying to find the companies most attractive, right?

And so as a result, we are going to find growth companies most attractive. Why? Because I’m trying to find the biggest gap between Wall Street’s expectations and what the company’s growing at. An insurance company is rarely ever gonna. Llike think about a company like Allstate. If they have a 5% growth in a quarter, that’s huge for them.

But if the Wall Street expectation was two 2, the gap cap’s only three points. We regularly find companies in which Wall Street expectations like 7 and real growth is 15, right? And, and so. That can only happen in a real growth story or a real growth company. So as a result, we do get more measures in technology.

Wayne: Consumer discretionary and healthcare. We tend to find the most bullish calls there. If I then took each of those sectors though, when stack ranked the companies, like I said, we still consistently stratify. So if I went into the, say another sector like. Consumer staples or finance. Right. And banking. I’m still gonna find a very consistent stratification between the best and the worst.

What really matters is your ability to stratify the two to call the winner and the loser. Not necessarily call that the winner’s gonna be an absolute winner, but he’s gonna be a winner relative to the loser. Right. And so it’s, ’cause you know, like I said earlier, we’ve got long strategies, we’ve got long bias strategies, but we also have market neutral strategies.

And our market neutral strategies are much more about building pair trades in which there’s a winner and a loser. And so even though we sort of score the entire small cap universe, the entire mid-cap universe, the entire large cap universe, and therefore, a staples company as compared to a finance company as compared to a technology company in terms of the magnitude of their growth outperformance to Wall Street’s expectations are the ones we’re gonna find most attractive, are gonna be the ones with those wider gaps. The reality is, within each sector, when you’re building a pair trade, you can still find meaningful differences. That’s what the sort of statistical arbitrage we’re delivering is able to do.

Radovan: Yeah, I like where we are going because my next question is related to portfolio construction. So once we have the alpha signal and we have a strong alpha signal how we build the portfolio? I mean, what is the period for which I should hold the stock? I mean, is it the months or weeks?Or days?

Wayne: It depends on which implementation. Right. In our long strategies, typically the whole time is probably close to 90 days. Remember, every 90 days a company has to announce its earnings, and so if we are right. That there’s a gap between the Wall Street analysts and our view.

And we were right. The analysts are gonna have to change their view and move their numbers up, right? It’s that difference is what’s gonna shake us out of the stock if we find a company that has a, still has a really big gap, and the Wall Street analysts haven’t moved up yet.

That’s something we’re gonna wanna still be long, right?

Wayne: And so, but what happens every 90 days, the company announces, once the company announces and says the word how we’re doing and what our guidance is, analysts have to come up. Now, sometimes our best trades are the ones in which we were right.

Company beats by a lot. The analysts were down here, but the analysts don’t believe it. And they go: “eh, I think this was a one off. Right? I don’t think they’re, this company’s quite executing.” We then look at the numbers for the next quarter, and there’s still a big gap, and so we stay bullish. Then they beat again to the analysts.

All the analysts capitulate when the analysts are wrong, two quarters in a row, they all capitulate. You know why they capitulate? ’cause they’re not allowed to be wrong three quarters in row, right? That’s how you lose your job on Wall Street is to be wrong three quarters in a row. And so they capitulate, that capitulation really propels the stock.

So if it’s a long strategy, typically that’s gonna be the construct in our more absolute return strategies. The ones that are pair trade, right. So it’s more relative value. You typically it’s, we’re gonna hold them until the relative value shrinks to our favor, right? Or expands to our favor.

Right? So those whole times are shorter. There’s a lot more risk management deployed in that strategy in the sense of negating away factors. We look for pair trades that are relatively factor neutral, but then we weight our pair trades. Such that the overall portfolio factor could be as close to neutral as possible.

There’s sort of a little more effort to neutralize and by neutralizing that factor exposure, what we’re able to do is really just isolate our signal, right? It’s our ability to be right versus wrong. On those calls because we’ve, you know, at a portfolio level, we’ve neutralized the way much of the factor exposure.

Dan: Yeah. And from the sounds of it, it also sounds like, at least in that market neutral idea, that’s where the Rado’s earlier question, the traditional price and volume data tends to creep back in as a source, because you need that for the comparison.

Wayne: Yeah that, that’s right. So first off, when we build the pair trade, we’re using a traditional factor model that we license to find pairs that are remarkably factor similar, but they’re never gonna be factor identical.

I mean, think about it, like even Walmart and Target, right? They’re very similar, a lot of the traditional factor measurements, like maybe their growth measurements or their balance sheet related measures, right? Quality measures. But they’re gonna be very different on measures like momentum and valuation.

Why? Because price is a key input to momentum and valuation, right? It’s amazing how often or how frequently Walmart and Target divergent price performance, right? So their momentum measures on a factor basis are always gonna be pretty different. They’re still remarkably factor similar in like many ways that matter.

So Walmart and Target, whenever they’re really far apart on our signal, would be a really good pair trade, right? But when you enter the trade, you’re going to have some net exposure on like a momentum factor. Right.

Because again, they, they trade so differently. You’ll probably have some differentiation on the valuation factor.

So wherever that beta exposure is positive or negative, right. We are gonna calculate it and say, okay, I need to find some other pair trade somewhere. That has the inverse.

Radovan: Yes.

Wayne: Right?

Radovan: Yes.

Wayne: And what we’re doing is we’re doing it on the incoming 84 days. It’s a blind factor study. It’s just we isolate on the seven factors that are most important.

They’re actually blind factors to us, but you can always bet that. You know, market momentum, growth, value, size, the most traditional ones are gonna almost always leak into those top seven factors. But, you know, there’s that extra level of risk management, which is why we’ve been able to deliver, what we had planned in inside the risk management design.

Radovan: And do you see the difference between the longs and shorts? In regarding the analysts, et cetera? So when you build a per trade with the long length of the per trade and the short length of the per trade, and I can imagine that there is like a different price action between the longs and the shorts because the analysts, they tends to differently look on the companies that improve.

Then the companies that are not improving. But around the other side of the trade.

Wayne: Yeah, I mean, one of the things about analysts, right, is there is an analyst bias towards bullishness, right? And they don’t tend to pile on the stocks that are underperforming. I mean, they will downgrade them if they’ve legitimate downgrade reasons.

It’s interesting. Remember the whole earnings game is skewed in the company’s favor. It’s not a 50-50 beat game, right? Companies beat on expectations more like 60% of the time because they try to low ball expectations. Yes. They try to press them down. And so the gap on the top side, right?

You would think that, that would say: “Oh, okay, well they must be harder for you to find beats”. The reality is the downside is only gonna ever be so far down, right? So we actually tend to find a wider gap on the bull side than the bear side when it comes to beat expectations. So in other words, our metrics on beats are slightly better to the bullish side than the bearer side.

But price response there’s no comparison. We will have more separation to the market on a good short trade than we will on a good bull trade. Right. Meaning when a company misses, it really gets punished, right? When a company beats, it gets rewarded, but not quite as bad as a company gets punished when it misses.

Radovan: Yeah. So there is a skew.

Wayne: Yeah. The only difference being right, the march to zero still finishes at zero, right? So, you know, but the march up can be more than a hundred percent. But in the long run, in the long run, maybe the way to think of it is on average there’s more price separation on the bear side to the average stock then there is on the bull side. But in the end, if it’s a pair-trade, I just want separation and performance. Right. Even if it’s modest, that’s still a winning trade.

Dan: Yeah. So I’m thinking of a AAA analogy here. Not the American Automobile Association, but it’s sounding like Asymmetrical Analyst Arbitrage.

It’s kind of what we’re talking about.

Wayne: Yeah, that’s right.

Dan: And I’m also thinking of the data collection and cleaning you were talking about Wayne at the very beginning when you’re explaining how you got to be quant and the jobs that, that some of these quants do. I’m looking for the mop in the bucket in your background to see if you have one.

Because if I’ve stolen this quote from somebody, once I’ve stolen it a thousand times, 80% of the data scientist’s job is janitorial. And from the sounds of it, there’s a lot of janitorial data cleaning going on to figure all this out, isn’t there?

Wayne: Yeah. I mean, luckily we’ve been doing it for 10 years, so, we’ve automated much of it.

Dan: Yeah.

Wayne: When we first started just for perspective, when we first started doing it, we had a team of six in India that would just, you know, keep our data footprints scrubbed and clean. We don’t even have an equivalent to FTEs doing that anymore. We’ve automated much of it. When you’re doing it for 10 years, you get a lot of experience.

We mostly just bring contractors on for a little bit of that work here and there. But that’s only ’cause of all the early experience with it. But you should know we’re always, we’re onboarding new data, and so there’s always something that needs to be scrubbed. And we also always source every data point from at least two places to do a quality check on it.

It’s not uncommon actually for us to find a mistake in the data, send it back to the vendor. The vendor corrects it for us. It doesn’t correct it for anyone else. That’s just, it’s not uncommon. And so that’s just the nature of this business. You know, we’re committed to quality, we’re committed to making sure the data’s useful and in how it’s gonna be deployed because obviously here’s a lot of data involved and the data’s gonna be what points us to our wins and losses.

Dan: So you’ve got an informational asymmetry there as well. Sounds like, because if you’ve got better data or you’re creating better data, that gives you not an unfair advantage, but it gives you a competitive advantage.

Wayne: Yeah I don’t wanna overstate that part. A good vendor is gonna make sure he fixes, at least he’s gonna fix that data for all future purchasers of it who buy it historically, right? Yeah. But anyone who bought it contemporaneously is not guaranteed to get, you know, that sort of clean data. So we do it a long time and it’s a lot of data and we’re always adding data. And some data even goes away on its own. There’s been data that’s been taken private by companies.

You know, that’s a always an interesting outcome as well.

Radovan: What is the rockstar of the data? So what are the popular at the moment? I know that there was a time when everybody was looking for, I dunno, satellite data. So everybody was looking. I know how many cars are on the park lot next to the Target or Walmart or whatever, whatever.

Wayne: It’s so many vendors, right. That we buy from, and it’s a, you know, it’s a fairly contained set of data that we buy from every vendor, so. Although, you know, we ask about lots of URLs. If I showed you the digital footprint for Amazon, how many URLs, for instance, Amazon, we have to ask about because of how many billion dollar brands they have.

I dunno if you guys remember, Amazon built its business early by acquiring every website that sold a lot of product like I mentioned earlier, barbecues.com, soap.com, casa.com, right diapers.com. If you were like the main deliverer of a product. Amazon, probably in the early days, probably just came in and bought you.

Then all of a sudden your website would redirect to amazon.com/diapers. But by buying you, they also, they also bought your, sort of your delivery capability. And people just don’t realize that that’s sort of, that’s where like where Amazon came from. So when you look at their map, it’s huge, right?

Obviously the bigger the company, the bigger their map. Apple’s got a huge map, right? Walmart’s got a huge map. Also the more they engage with customers online, the bigger the map, right? Meanwhile, does a B2B company have a smaller map? It does. But remember if you’re a B2B company. People still spend a lot of time on your site for service reasons or to research buying the product.

Imagine you’re gonna buy a $150,000 combine for your farm, right? You’re gonna buy that from the local dealer, or you’re gonna buy it from a salesman for John Deere or Caterpillar, right? But you’re gonna go onto their website and you’re gonna read the manual. You’re gonna look, you’re gonna look, you’re gonna go on message boards.

There are Reddit boards where farmers get together and talk about Caterpillars, right? You’re gonna read the message boards, you’re gonna read about, and you’re gonna do a lot of research. If Caterpillar’s selling more combines for some reason, ’cause there’s more demand for combines, that’ll be something we pick up because more people will be going to that site. Even though they don’t buy it on site.

And so, sort of keeping all that data becomes really relevant. In terms of data, you know who moved into the business smartly over 20 years ago was Nielsen, right? Nielsen, the guys who do your TV ratings, they sort of realized more than 20 years ago that advertising is moving online, and what they really measure is that when they were measuring how much commercials you watched on tv, they were really measuring advertising breach, right?

So they realized over 20 years ago, they had to move into that space. Plus, remember the companies that sell us advertising space online give a lot of this data away. Like, Amazon will give it to you. Google will give it to you. Facebook will give you measurements about key terms, key brands ’cause they want you to buy those terms, right?

So some of the data you can buy, you can just get from some of these vendors. That’s not the lion’s share of the data, the lion’s share of the data you gotta buy. Right? And it’s not cheap. Like I said, we’ve spent millions of dollars on the data.

Radovan: Here is another question. So sometimes there is a difference in academic research.

So when the academic tries to explain that some factors work better in large caps and some factors work better in small caps. They’re like two different explanations. So in small caps, some factors work better because I mean the, it’s harder to obtain data for the small caps, and they’re harder to arbitrage.
And for the large caps, the problem is that the large cap companies are so big that, as you mentioned with Amazon, there are so many different parts of the company that you have a problem to come up with the one number that I mean will explain the whole company. And here is the question. So, from your point of the view as we work with alternative data, so, which companies are like more inefficient?

I mean, in a market form. So where are more opportunities in the large caps or in the small caps in large cap because they are harder to calculate or in a small cap is because the data are harder to obtain.

Wayne: Yeah. So understand we create alpha in large cap. We create alpha in mid cap, and we create alpha in small cap.
But the best risk adjusted alpha that we create comes from mid cap. And that is because it is the best mix of just enough analyst coverage and just enough digital data. ’cause they’re good sized companies. They’re $10 billion, you know, companies, right?

And so just the right mix. And also these are companies sort of emerging. They’re not big enough being the S&P 500. These days, you can go from being a, you know, a mid growth company. I mean, look at like the sSnowflakes of the world, how quickly they go from being, you know, small.

There are companies now, there are examples nowadays, right, of companies that just jump right past. I think AppLovin is one that we’ve been in for the last year. I think it jumped right, and what jumped from the small cap index to the large cap index, right? It never, it never landed in the S&P 500 mid cap index, right?

There are so many interesting stories about companies that are sort of executing in that mid-range that can become very big, very quickly. But we do create alpha in the small cap. But yes, small cap has the problem of less digital data and less analyst coverage. Will you get the biggest moves in the small cap space?

Of course you will, right? They are the more volatile stocks, but you’re gonna pay for that and volatility, right? On the large cap side. Right. Do you still get some pretty good moves? You do, but you know, analysts have picked over Apple. I mean, I’m I, you know, aApple is the most watched company, right? It’s the most covered company.

Nvidia might be challenging them for that, but it’s probably still Apple. And so there’s a lot of analyst coverage and there’s a lot of digital footprint data. And so, you know, it offers a lot of insight to the digital footprint. But a lot of people are scouring that digital data.

And there’s a lot of analysts covering it. So we still create an alpha absolutely in each of those categories. It’s just mid cap tends to be the best. For our absolute value strategies, the relative return, relative value strategies we actually only focus on the large and mid cap stocks.

We don’t pull any small cap stocks in, into those portfolios, because of the volatility and because there’s shorts in there. Right. And you know, small cap companies can definitely on the short side get punished disproportionately. And so for a little bit of risk management reason, it’s all large and mid-cap also, that just makes it more scalable, right.

Shorting doesn’t scale as effectively across a small cap universe. It scales more effectively across a mid and large cap universe.

Radovan: This is the long-short story. And when you build just the long portfolios and you want to hedge those portfolios, so do you hedge with shorts only or do you use also the options or futures?

Wayne: In our long bias strategies, which are hedged, we actually just use index puts.

Radovan: Just index puts.

Wayne: Puts. Just index puts, yeah. But luckily we historically create enough alpha from the long book. That it pays. It pays for that drag. So you operate with a floor in your portfolio at all times.

But it’s effective, so which produces lower vol and yet we still mimic index like returns with lower vol and an absolute floor in your portfolio. ’cause there’s a hedge.

Radovan: So yeah, understand. I understand.

Wayne: As opposed to most hedge equity strategies, they have lower vol, but they also only capture about 70% of the upside return.

We capture closer to a 100% of the upside return. 90 to a 100% .

Radovan: Maybe the last question. When you have the long-short book and you, I mean, combine the longs and shorts and you try to be neutral for all of the factors. Do you still see some of the factors that appear for a short time in that portfolio, like I’m, for example…

Wayne: factors will bleed.. factors will bleed through. Yeah.

Radovan: Exactly. And also what I’m talking about is like the quantum meltdown in 2007 and stuff like that. So I mean you can be neutral as you want, but I mean, something happens, something geopolitical. And it’ll bleed to your portfolio even that you, I mean, you are neutral.

Wayne: Yeah. Yeah. I would say the best way to think about when we deliver the best outcomes versus the worst outcomes is when markets are moving from micro reasons, right? Meaning all the individual stock stories.

Combined to cause the market to move. That’s always good for us ’cause we understand the individual stock stories really well. Who’s winning and who’s losing. When markets are moving for macro reasons, right?

That tends to be worse for us because, you know, we know which guys are executing, but when, the baby’s being thrown out with the bath water.

Yeah, that can be. One of the interesting ones was, but you know, I say baby thrown out with the bath water. Sometimes it can be, it’s not just markets going down. ’cause because that strategy is not correlated historically to equity markets. One of the worst reversals for us was like November, 2020 when the news came out that the vaccine was really close, right?

Everything that had been a loser that year was like, you know, anything that was low mobility or reduced your ability to get around. Right. Those stocks were all the ones that rebounded in that November. Right. That sort of whipsaw was indiscriminate. Right. Yeah. And so it hurt a lot of quant shops including RV at the time.

Right. And again, when markets move for macro reasons and news reasons, that sort of thing. That tends to be worse for us than when they move for micro reasons.

Radovan: Yeah, I understand. So this is like an idiosyncratic factor that’s, I mean, there and was not there in the data for 20 years or whatever so.

Wayne: When we look at our long short returns. Over time so I told you we license one factor model to find companies that are pair trades that are remarkably factor similar, and then we do a whole blind factor study to neutralize the pair-trade sort of exposures, right? When all that’s done, we actually license a third factor model from a different company, and we analyze the returns and say what amount of our returns came from factor exposure based on this third model, right?

What we find is three quarters of our return comes from idiosyncratic it’s stock selection. The last quarter does come from factor exposure. But there’s no consistent factor exposure. Tends to, it changes over, right? It changes, yep.

Dan: Almost time to land the plane. But let’s give a little bit of thought to people who are still new to the topic and approaching quant strategies and artificial intelligence usage.

We were talking to Wayne the other day about some of the basic differences when it comes to AI investing like what’s the difference between investing with. AI and investing in AI and how’s that changing?

Wayne: One thing I would say is, you know, if you’re an allocator, if you’re a person who makes decisions about investing clients’ money, right?

You need to be asking whether your managers are using AI as part of their process of stock selection, you know, portfolio selection, portfolio analytics. It’s the future, right? If they’re not, you should be asking them also, well, what is your path to starting to do it right? I was at a conference recently.

I was at a table of 10 people and I asked them how many of you are sure that in five years when you do a diligence on a manager, you’ll ask them a whole series of questions about how they use AI in their selection process? Every hand at the table went up all 10, and these were 10 allocators. I said, great.

How many of you asked those questions today? Not a single hand, right. And I said, well, gosh, AI is already starting to get into this business, right? It’s already starting to be used. Don’t you think? You should probably be asking these questions now? You know, there it is. The future, whenever you’ve got a lot of data AI can be a great solution for turning the data into better insights. Right. And so, and we have a lot of data in the investment management space, and so I think people should be asking how folks are using it. What I found so far is probably only about 10% of investment management companies are actually doing anything really creative and interesting on the real selection side or investment management side.

About 80% are using AI to organize analytics or to make their CFAs more productive. Be able to cover more stocks and ask more questions, compare different documents to each other, right. You know, 80% of the companies are at least doing that, giving those analysts those tools to make them more productive.

But it’s not really affecting selection in the sense that the math is doing something predictive to tell you what, you know, which of these companies is gonna be a better performer. Right? And the last 10% who are doing nothing. Those companies are really in trouble. They’re not at least leveraging it for productivity today.

There is some old school thinking inside that firm and they’re, they’re probably in trouble. So, you know, AI is the future in so many industries. It’s definitely the future in this industry. And so I think people need to get trained in sort of how to ask the question about, you know, what are you doing with AI?

What’s your objective function? How do you organize the data? You know, how thoughtful are you about this? How often does it change? How often do you change it? Right? There’s a whole series of questions they should be asking in that space. And actually, you know, come meet with us, we’ll tell you what we do and you can use us as the standard to compare to others.

Dan: Very good. Okay, well it’s about time to wrap, so I’m gonna say thanks Wayne. Thanks Rado for showing us a bit of the quants crystal ball today. So now to meet Wayne from Alpha DNA or to meet Rado from Quantpedia, to join us as a guest. Or if you’re interested to otherwise support the channel, contact me, Dan Hubscher at these details that are here on the screen.

Just note that any questions regarding investment offerings will be deferred to an email follow up for compliance purposes, as these videos are not intended to be about investment offerings. But if you like this video, please do come back. For our next interview with a Quant manager guest, you might also wanna watch the other videos on the QuantBeats YouTube channel.

So don’t forget to like, comment and subscribe to quant-beats on YouTube. And again, you can get there via that Q logo in the corner of your screen. Or finally, you can visit our websites and YouTube channels for more shown right here. So thanks for joining, and as always, have a quant day.

Radovan: Thank you. Thank you, Wayne.

Wayne: Thanks guys.