 Las Vegas, expecting the signal from the noise. It's theCUBE, covering Interconnect 2016. Brought to you by IBM. Now your host, John Furrier and Dave Vellante. Welcome back everyone. We're live here in Las Vegas for IBM Interconnect 2016. This is Silicon Angles theCUBE. It's our flagship program where we go out to the events and extract the signal from the noise, talk to the smartest people we can find, extract the data, share that with you. I'm John Furrier with my co-host Dave Vellante. Our next guest is Prashant Buyan, who is the co-founder of Alphamotus. Welcome to theCUBE. Thank you very much for having me. Good to see you. Try to extract the data out of your head and share with the audience in real time. Do some machine learning. I would hope you would extract the knowledge. Yeah, that's right. You guys were on stage this morning. Very impressive demos, obviously showing the benefits and the value by the way of the data, using data in a new way in real time, providing real value. So talk about the story real quick for the folks tuning in. Alphamotus, why are you here on stage and what were you talking about? Sure, we created Alphamotus in 2014 as an investment technology company. Our goal ultimately is to reprice investment advice. But the way we do this is by looking for predictive patterns in market behavior that's buried in unstructured data, natural language. We discovered Bluemix organically after the Alchemy acquisition. And we saw, I mean, five years ago I saw Watson on television beat the Jeopardy champions like everyone else. And I thought immediately, how do we get one of these? So, right, but it's not like they put a price tag on a Watson, you can go get it from Amazon or whatever. So basically, when we discovered the Bluemix API and we started to look at the product size, right? Suddenly, we could start to incorporate tweet insights, Alchemy, natural language processing as a new dimension into our model making and improve the predictive accuracy of our old models. In particular, an imbalanced product that we used to predict market direction into the close of trading was about 10% predictive. After adding this NLP element to it, we were able to do that by a factor of five. So you're, you're, so. So 50, 50, 50, basically. Yeah, of 550%. Right, and so, which is basically, ever since the earth grew a nervous system in the internet, you know, the cognitive biases of investors compounded into these collective manias and ultimately our brains aren't growing fast enough to absorb, to make sense of all the complexity. So a lot of the traditional investment methodologies, which are based on theory rooted in. Diversify. Evaluation, diversification, assumptions about normality, independence, and rationality of investors, kind of out the window when you have news and information spread around the globe instantaneously, right? And, you know, suddenly human psychology plays a much larger factor than ever before. It amplifies, right? These markets. Now you can measure everything. So let's just take this nervous system, you mentioned, which call it Twitter. The CB radio for the world, if you will, all chattering away. That's data. So financial markets have always had some sort of thesis and, you know, norms around patterns. You know, the deer, the herd, deer in the headlights, investor too. You know, the herds moving against that. And so with news cycle, if I get what you're saying is happening is, you're saying, okay, we can use things like news as event traps and look at the impact to psychology at essentially real-time level and accuracy. Is that kind of? It's not quite real-time yet. Cognitive, real-time cognitive analytics is where we want to go, certainly. But it's still, you really don't have to be the guy who wins the race. You just have to be faster than the guy who's you're racing against, you know, in the world of asset management. And so the idea is that buried in natural language, whether it's the tones of our voices or, you know, the sentiment in our tweets, weather patterns, you know, I mean, what affects markets that typically, you know, traditional investment methodologies don't encapsulate? So what's- So we need to set the table here a little bit because I think for a lot of our audience, I mean, does everybody know what alpha is? Right, so alpha's beaten some index, right? Some average. Right, right. It's simplifying. It's beating the market on a risk-adjusted basis. Okay, so like the website, seeking alpha, everybody's seeking alpha. And your objective is to totally transform that business. So today, if financial advisor charges a percent of the assets in the portfolio, you're trying to- Well, our philosophy is that no one should pay a professional through a dart to the board, right? It's like, if you're paying 1%, you know, the only person who's really compounding returns is the advisor, right? If they're going to be compounding returns, you should be compounding returns at a higher rate than what you're paying, right? It's not really anything else, but many asset managers are having difficulty keeping up with the complexity of markets because of all this data, right? So, but the business models haven't changed around this kind of revolution in data that- So are you a technology provider or are you a service provider or both? We're a technology provider. So what we do is we analyze unstructured data to find predictive patterns in markets. We take those insights and then what we do is we package them up into high-performance algorithmic trading systems that we then distribute to asset managers who are our clients, who are our clients, typically, and other market participants via the cloud. Okay, and then they can choose to still charge one or 2% or they could- Exactly, because they've got the regulatory framework and everything like that, but they can still charge what they, but now they can justify their views. Right, and somebody's going to come along and say, hey, I can do Robo investing and dramatically lower your cost and my cost. One key distinction is Robo investing typically refers to like passive investment management, just kind of mimic the index for free, but we're really focused more on active management where we're saying, all right, well, let's beat the market on a risk-adjusted basis. Are you just like a fancy way of saying a recommendation engine for the clients? Are you actually doing trades on their behalf? Are they taking support? Well, the recommendation, so we are not, we're currently in the process of building out different types of wrappers to distribute these insights, but what we do is we create trading strategies which are distributed as algorithms, as part of high performance algorithmic trading systems that we kind of lease infrastructure as a service kind of thing along with the strategies. So an asset manager, say your hedge fund or something, and you are having difficulty beating the market on a risk-adjusted basis, you can actually subscribe to our solutions and kind of run them almost like white-labeled strategies. So what about venture investing? I'd love to corner the market on the early stage, unicorns, one in two in 10 funds the entire venture fund as an asset class, we call it that. You guys going to anything with that or is that two gray areas? In terms of trying to predict that. Yeah, using data approaches there as it's just too gray. That's not really our primary area of expertise, but I definitely think that there's a lot of opportunity in applying these types of analytics across venture, private equity, traditional banking, I mean, across the board, but that's what we're really focused on our niche, which is active investing. All right, so you guys doing some cool stuff, so dig into the hood, share some color around some of the tech involved, and then the impact to the people watching, whether you're a customer, or someone else in another industry saying, you know, my gut's telling me I need a data-driven approach, I hear the buzzword, be data-driven, it's good for your business. Okay, kind of cliche, but the reality is it is. So talk about an example of some of the tech and how that would translate and impact a business. Well, you know, the beautiful thing here is that we really were just created on the cloud, so there's no like trying to embrace cloud or anything like that. And the reason is because cloud has kind of, we're attracted to cloud, Abe for the cognitive, but B because we're running high performance algorithmic trading systems. The technology behind the insight discovery process really, you know, in the natural language space, we're pretty much using a bunch of these blue mix APIs, Watson partnership APIs like Alchemy, Tweet Insights, Tone Analysis, these types, you know, there's almost like no tech involved. It takes us a day or two to build out really powerful insights, prototype these solutions incredibly cheap and fast, and then put them into our trading systems, you know, which we already have an infrastructure for that, right, and then deploy them right on the cloud through IBM. You basically buy in with blue mix through the APIs. Blue mix. All the IBM IP, plug right in, extract out some rule heuristic, and then from that, you use your back end. Correct, yeah, our infrastructure that we, and then we just create instances of that. And, you know, the key thing here is that, you know, the key unexpected consequence of this whole thing is that blue mix and Watson, you know, these types of technologies, they actually, you know, when I saw Watson be Jeopardy Champions five years ago, I was sure that technologies like that would replace human intelligence in short order. But actually what's really happening here is that, they actually are unleashing our creativity and extending our intelligence. Like, I mean, we're able to build in prototype ideas, like record, like if I have an idea, we can have a prototype built in a few days. Which I got a Bob Egan researcher was on earlier and he brought this out where opportunities for creativity are highlighted in an entrepreneurial way where, and also risks to companies that don't evolve is when you look at gap analysis between what you're good at, it's easy for someone to identify where you might have a gap on an entrepreneur and say, hey, there's a gap in that big whale over there, company industry leader, and my crumbs, his crumbs is a billion dollar opportunity. I can move on that. That's kind of the thing that we're seeing. Do you see that same dynamic? Yeah, I mean, if I'm understanding your question correctly, I think I am seeing the same dynamic in the sense that it's really about the use case, right? It's about applying domain or something to the technology to commercialize a product that adds value to the customer. So help us understand how natural language processing fits in. Are you presenting a reduced set of choices for an individual? Are you increasing the probability of success? So basically what it is that we, all right, we'll say we'll start with BlueMix and prototype an idea on a small data set, right? So we'll have a small corpus of news related to whatever topic, for example, which is really harder than actually building the sentiment parser. But say we build a sentiment parser, the insight makes sense on a small sample, scale out the corpus, right? And then apply to run it through the sentiment parser. And we might get back, for instance, positive and negative sentiment values that we then turn into like, then we wait those values by relevance, you know, and we'll incorporate those into our models. Like maybe those models have imbalances as another factor, for example, right? And what we'll then do is, you know, do data science and try to figure out how predictive is this thing? Is there a significant relationship here? You know, maybe we want to, you know, reduce the risk of coloniarity. So we, you know, different types of statistical methods to try to figure out how to turn this insight into something that's stable. And once we have something predictive, then we just plug it into our infrastructure where we have execution algorithms, you know, routes or smart order routing, you know, low latency co-located infrastructure. And now we can just, we have a strategy that will start running in a matter of a week. So brag a little bit. Give us some examples of some of the real successes that you guys have seen. All right, well, so there's a demo that I'm doing in my breakout on Wednesday. And it applies to sentiment analytics from Alchemy. And it's plugged right into one of our imbalanced models. And, you know, it's a stable predictive pattern. It's a $60,000 account in 85-second trade. And, you know, in the demo, it makes a real cash, $1,062 on that $60,000 capital, which is, you know, a 1.77% return. And the reason is because on that particular day, you know, there was sentiment weakness or uncertainty in Europe, central bank policy, weakness in oil. Oil was down about 8% recently. But the interesting sentiment was the strongest weighted sentiment scores that we had were in gold, flight to safety to gold. And these happen to coincide. It's strong buy imbalances in gold, weak imbalances in energy inequities on that particular day. And we're able to book a really good return. Yeah, it's interesting how much, you know, Dave and I talk about this all the time when we geek out about some of the tools we built and run data science and look at the history of the web and the interactive world since the web started. It's really the, we're back to contextual and behavioral data. You're essentially writing code and using blue mixes and saying, hey, here's the context, trade, banking policy in central Europe. The behavioral is going to be the pattern you can predict. And the prescription is the trade. You're saying, okay, you're doing that just faster and uniquely. And the key is how much of a particular phenomenon can you explain? Right? Like how, you know, that's really what we're after, right? If I could explain, and by the way, human behavior is almost impossible to explain. You can ask any psychologist, right? So it's like, but you don't need to explain all of human behavior in order to make a lot, you know, to make decent returns. You- No matter if the NFC team wins the playoffs, Superbowl, the stock market goes on. How do you explain that? Right. The giants are in New York and that's an NFC team. People are happy. I mean, I don't know. I'm guessing, but it is a guess. I mean, you know, it's like, you know, sentiment could be affected by weather. I mean, who knows, right? But it's all about looking for these deeper connections in the data. And you know, it's like the famous quotations, right? It's like two and a half quintillion bytes of data created every day, 90% in the last couple of years. Most of this data is unstructured, right? So like, we're not even tapping into, we're just at the surface of really what all of this stuff means, right? And applying it in real time is really the future, I think. This is not your first rodeo entrepreneurially. You've done a few ventures in the past. Well, it is actually kind of all the same trajectory, but it's an evolution is what it is. That same niche as they say, entrepreneurial. So what are you doing differently? Let's talk about the entrepreneurial journey. It always is the case where you scratch that itch and just kind of stay on that same track. But now, as a startup, entrepreneurially, there's new tools available. So you're cloud native, you mentioned that. What are the things that you're doing now that weren't available from a funding company? What we're really doing is we're leveraging infrastructure that was built over the last decade, right? But that infrastructure itself is really just a delivery mechanism for our trading insights or algorithms, you know? What's different now is that we have this focus on this, below the tip of the iceberg data. And it's like, wait a second, instead of just looking at how the prices of this asset relates to the prices of that asset, and trying to find a relationship, let's also add in the dimension of something in unstructured data. Natural land, like one thing we're doing right now is we're prototyping a tone analysis so we can kind of like kind of parse sentiment from public company executive earnings calls and see like how it's not just like the news they deliver but how they deliver it kind of thing and see how that can affect our market prices. We're very pleased with the quarter. Yeah, right. Crash, right? It's like the other thing. Yeah, right, so things like that affect investor behavior, right? But they're not necessarily captured in the price. At least not at all. I mean, you're basically picking up nuanced signaling. I mean, essentially it's taking signal theory. The price will factor it in but there's like a lag time, right? It takes time. And that trade value is just a blocking and attacking chip away in volume. If you're knocking down on 60K, $100,000. Well, it's more about, it's not so much about the absolute dollar amount or the return. It's just the idea that, you know, this is a predictive pattern that works. It's about monetizing insight that we're able to extract using Watson alchemy, right? Okay, so talk about, so this is always hotly contested since we're geeking out on it, we'll go there. Predictive versus prescriptive analytics. Obviously a lot of people get confused by the two. Predictive and prescriptive and you're seeing it in ad tech and now you're seeing it in all verticals. And any mobile app pretty much has some data layer in it with some aspect of analytics. Sure, so. What's the difference between the two prescriptive? Well, I mean, if I understand you correctly in terms of prescriptive analytics, your suggestions, right? I look at data and I suggest, you know, I was at Partner World last week and I heard, I think, Steve Gold talk on stage about Fitbit and the idea is that, you know, well, with Under Armour, I think they're trying to now, like it's not just, all right, well, you know that you have 10,000 steps and now what do you do about it, right? It tells you kind of what to do. So to that end, we're actually building out a recommendation engine that will actually allow portfolio managers to upload their returns and take, you know, from our universe of different strategies, some kind of recommendation back saying, all right, well, if you incorporate X amount of this strategy and X amount of that strategy, this is how your portfolio could evolve over time. You know, you know, so we're kind of getting into that. From a predictive point of view, it's really just about finding those insights. For Sean, I really appreciate you coming on theCUBE. Share with the folks out there that are watching. What's the vibe of the show here? What's your take on the show for the folks that aren't here watching live? It's a great show. I mean, it's a lot of fun to talk to you guys and you take a real data-driven perspective on things. Awesome. Well, we have a recommendation for you here and the recommendation engine theCUBE. Check out our CUBEcast on SiliconANGLE.tv, a lot of insight in there and potential prescriptions. We'll figure out how to create that in the future. For Sean, great to have entrepreneur data science doing some cutting edge, creating value, not just arbitraging with value, congratulations, alpha modus here on theCUBE. We'll be right back with more extraction of the data in real time right here on theCUBE Live. We'll be right back after this short break.