Jonathan Larkin of Quantopian, Christina Qi of Domeyard and Andy Weissman of Union Square Ventures sat down to discuss algorithmic trading.
Hedge funds have had a rough go in recent years with many of the industry’s top players posting below-average returns. As traditional hedge funds have struggled, quantitative and algorithmic trading firms have gained prominence. While quantitative and algorithmic trading is not new to the industry (firms like Renaissance Technologies have employed quantitative trading strategies for years), it is getting more attention lately due to its promise of delivering future alpha to the industry more broadly.
At the CB Insights Future of Fintech event, Robin Wigglesworth of the Financial Times sat down with Jonathan Larkin, Chief Investment Officer of Quantopian, Christina Qi of Domeyard, and Andy Weissman of Union Square Ventures to talk algorithmic trading.
Jonathan Larkin says the growth in quantitative trading has been due to advances in cloud-based technology as well as the dissemination of tools and information that were “once only owned by a small handful of companies on the island of Manhattan and parts of Connecticut, but which are now available to the world.” His company, Quantopian, is a sandbox platform that lets amateur data scientists test and trade securities using quantitative strategies. “It’s a platform anyone can access and apply quantitative insights to the world,” Larkin said.
Qi agreed, noting that there has been a recent democratization within the algorithmic space. Just 5 years ago starting Domeyard as a quant fund was a difficult task, but based on the data sources that are available today new funds could create strategies “we never even thought of [at inception].”
Unsurprisingly then, there’s also been an observable increase in competition of late, Qi noted. But despite what people may be seeing, the hedge fund industry is not dying out but consolidating and differentiating, with funds focusing on areas they know best and using quantitative strategies to trade within those knowledge areas.
Andy Weissman of Union Square Ventures suggested that Quantopian and Numerai (a competitor crowd-sourced quant trading fund in USV’s portfolio company) are like many other software companies in their portfolio, in that both companies are working to build communities on their platforms. Numerai, like Quantopian, allows data scientists to test and deploy trading strategies, whose pending success might warrant inclusion in each company’s meta-model trading strategy.
Not all hedge funds are turning to quantitative methods however. Larkin attributes this to “basic human anchoring bias.” When you’ve been very successful at doing something for a long time, it’s hard to change your ways. This may pose challenges to future fund raising for investment funds that rely on traditional methodologies however, as Qi said she feels many millennials in her generation bring an expectation to the investment industry that their money will be traded with the latest and most sophisticated strategies.
“These days we don’t hire MBAs. If you want to get into trading study math, not finance,” Qi said.
Math and data science are surely key skills for the industry. In addition, it is the accumulation of, and access to, the data itself that provides algorithmic models with better training and more efficiency in the markets. But whether crowd-sourced strategies or a centralized owner of data will win out in the long-run was a matter of debate among the panelists.
Larkin believes platforms like Quantopian are accumulating unique data sets that straddle both roles “with hundred of thousands of unique trading strategies” that let them train their algorithms. However when asked what would happen if major data owners like Google decided to start their own hedge fund, Weissman replied, “Do we know they’re not doing that?”
Robin: Hello, everybody. Thanks for sticking with us because this is, frankly, the best panel of the conference. And it’s gonna be wonderfully moderated by myself and we’re gonna have tons of insight from everybody here. The reason why I say it’s one the best panels of the conference is because, you know, it’s a subject I find fascinating, I think you guys will also. It’s the idea of the algorithmic hedge fund.
Now, this is a big trend in the world of investing right now. Quant hedge funds, as they’re called, have grown at a far greater clip than the broader hedge fund industry over the past few years whilst the rest of the industry seems to be slowly deflating, and closing, and struggling, and complaining all the time. Trust me, I hear the complaints.
But this is broader than, frankly, just hedge funds as well. The idea that we are going to have a more scientific approach to managing money, using more technology, more data, is spreading across the entire investment industry. I guarantee that all your pension funds have now traditional, boring mutual funds and asset managers buying alternative data hand over fist and doing most of their trades algorithmically. JP Morgan recently calculated that quant funds of some various stripes or another actually account for 60% of all trades in the US stock market. And only 10% is done by a traditional human, the stock picker of yore. So we’re gonna be diving into this.
The first guy I want to ask is Jonathan about this because you worked on the hedge fund side as well before coming over to Quantopian. What’s been driving this push towards more algorithmic and scientific approach to investing?
Jonathan: Thanks, Robin. So there are a number of factors that can explain this growth, some are structural just in the world. So the growth of cloud-based technology, for example, makes all of this opportunity available. But what’s also happening is that tools and not just tools but really opportunity which was once the sole province of a very narrow set of organizations, typically on the island of Manhattan or in Connecticut, are now available to the world.
So at Quantopian, for example, we’ve created a free online platform tool chain for quantitative research and quantitative strategy production which five, six years ago would only be available to some of the largest traditional asset managers in the world. And we’ve taken that and we placed that for free on the internet. So we’ve taken the tool set, which was once available to a very few, and we’ve made that available to the world. And along with that, we’ve made the opportunity to apply quantitative insights to finance available to the world.
Robin: Andy, you also…I mean, sticking on this more democratized approach to investing, you’ve invested in Numerai as a similar internet based, open platform for sort DIY quants around the world. I mean, why did you do this? This is a bit of a departure for you guys, right?
Andy: Well, it is. It’s a departure, it’s a little incongruous to think of a venture fund investing in a hedge fund, however, Numerai is…it’s a community business. It’s a community of 17,000 data scientists who are building models, and we’re good at understanding community businesses. The business model of Numerai happens to be a hedge fund trading an ensemble model that the data scientists are creating. So it’s a combination that is fueled by some of the trends you said, cloud technology is really powerful computing that allows more people to build models.
And Numerai is open…well it’s open to anyone. I’m sure some people here are members of it. I’m a member of it, anyone can create a model you don’t have to know anything about quantitative finance, there are lots of biologists that are on the platform as well.
Robin: Bit of basic data science helps though, right?
Andy: All you need to know is data science.
Robin: Yes, exactly. So I think it’s almost like a Kaggle for a hedge fund cum Kaggle.
Andy: Yeah, in an anonymous manner though also.
Robin: Christina, one thing, I mean, you work…it’s quite an unusual model, low latency, sorry there’s a high-frequency trading hedge fund that operates at great speed and that’s an interesting model. But, I mean, I’d be interested in hearing how far do you think we’re going to go towards the path of having truly algorithmic, systematic hedge funds and what are the limits? Like, where do we still need humans?
Christina: That’s a really good question. So I think in order to look at how far computers can go, maybe we can take a look at like…an example would be, like, an area where computers are really good. So, like, for instance, there’s AlphaGo which a lot of people know about, beating the best Go players in the world, the best human Go players in the world, right? Why is AlphaGo so good is because Go is one of the oldest games in the world, it’s thousands of years old, there’s millions of games to train upon for this algorithm. And there’s millions and billions of data points really that it can look at and that’s why it’s better than a human.
But if you look at areas like…trying to think of one area. For us even in high-frequency trading sure, the reason why we do that is because we also have billions upon billions of data points to look at. And that’s why we know that relying on a machine actually is better than doing whatever, relying on our human emotions and intuitions. But that is not to say that there is a limit to that, of course. So one limit I can think of actually is in the venture capital space, surprisingly. So if you look at what a VC does, every VC has a different set of rules, right? Go is a game where the rules are pretty straightforward, for thousands of years it’s been the same set of rules. Look at the VC world, every VC has a different set of rules.
When we’re talking to VCs, some people stay far away from us and there’s other VCs who love us, so you never know. So the rules are different and also you probably made…maybe you have a portfolio of some VCs on average maybe make a couple dozen investments, there’s not a lot of data points to train upon. And so if there’s a lack of data and if the rules are different between different firms, then that’s when you’re gonna need a human to be there. One other example we went through, we actually went through an audit recently and that was a giant nightmare. I mean we had like, what? Hundreds of email threads…
Robin: Well it hasn’t been like one of those 20 year audits that Trump goes through, right?
Christina: Yeah, exactly, right? Hundreds of phone calls and email threads and this is all manual. And I just saw the CB Insights thing they had a couple audit firms that are coming up that are trying to disrupt that space which is awesome, I hope they do. But it’s gonna be maybe a couple years before that actually takes place, so those are just some examples I could think of.
Andy: I’m happy to think you believe my job is safe, I’m less optimistic.
Robin: Well Anand made that point, didn’t he? That all VCs think their jobs are undisruptable but the reality…
Andy: Yeah, but we have to because we’re human beings. At the same time, my firm at least is trying to invest, profit off of our demise through investments like Numerai. We have investments in another company called CircleUp which uses quantitative techniques to invest in consumer products companies. The idea there being that consumer products companies all have the same business model, they sell something to a consumer and the consumer buys it, therefore, you can normalize the data.
So I’m less optimistic that the trend you’re talking about will not be applied to venture. I think they will be applied to everything. I just think the technology and the science is getting so good and it’s getting better and it’s accelerating, that it will spread everywhere.
Robin: Yeah, I mean, John you’re Quantopian’s CIO, you’re the, sort of, the human interface. I mean, are you basically working to make yourself unemployed, unemployable even?
Jonathan: Well, first I would say we have a community of 140,000 people in 180 countries. In our investment process, we’re sourcing fully automated strategies. But those automated strategies come from somewhere, they come from the idea generation and the work of all of these individuals. So we’re tapping into this global mind share to make something valuable for our investors. And internally, everything that we do we seek to be based on a process and based on scientific reasoning. So as we move forward, internal data that we continue to grow about the different strategies that we have, strategies we select, strategies that we deploy, we follow in some sense a hybrid approach to our investment process. But we’re always looking to bring machine-based approaches to everything.
Robin: I mean, just to follow up on Christina’s point as well, what are the limits here? I mean, AlphaGo is a great example because it’s so unimaginably complex, but it’s a game with clearly defined rules. And one thing I love about markets is because they’re a messy, messy, constantly evolving, chaotic system with no rules. And there are so many different agents that act differently and irrationally and with human biases. Can a quant hedge fund really, over time, always do well in an environment that’s still so human and still so messy?
Jonathan: Well I think we’ve seen that historically. If you look at the hedge fund performance over the last 20 years, the history of the industry, you know, quant funds, by and large, have been the superior performers consistently. It might not be as transparent as some other industries, and a lot of these funds are privately held or they’re closed to outside investing. But, you know, as you get to know the industry it’s not something new this year that quantitative investing or trading is beating out the human counterparts.
Christina: Also one of the biggest misconceptions in this space is that everyone thinks that we just have one strategy that we run constantly over the years and it just is always successful.
Robin: Which would be great.
Christina: Which would be awesome but that’s definitely not true, even at the biggest quant funds out there. What we’ve seen is they’re constantly evolving and shifting their strategies based on whatever’s happening out in the markets. And some strategies have a lifespan of weeks and it can just be dead in a week, and then you’ve gotta move on and continue generating ideas and generating insights.
Robin: Yeah, I’ve heard people call it “alpha decay” that something that makes money in a short period of time. Some of them last for a few days, a few weeks, a few months, a few years sometimes, but it’s pretty rare. How often do you guys iterate your models or change or add new signals or new trading algorithms?
Christina: Oh, all the time. All the time, yeah.
Robin: All the time.
Christina: It’s a constant process of iteration and how can we do better? Even if we’re doing good enough, okay can we do even better than that? You know, and just seeing what the limit is of what we can do.
Jonathan: Yeah, I absolutely concur with that. I mean, that’s a foundational principle of our organization which is that all strategies, whether they’re discretionary or systematic, always face competitive pressures and they don’t last forever. So how do you bring to bear the greatest urgency and the greatest amount of idea generation? You bring to bear the internet to the problem, and that’s really what our business is about.
Andy: So Numerai takes a very different approach, and its approach is that when you have…humans involve bias and strategies will inherently involve bias, so how can you reduce or remove bias from investment decisions? And the proposition for that fund is you make it a math problem. And so the data scientists for Numerai actually don’t know any…you download a series of integers and you create a model. You actually don’t know what those integers represent because it’s encrypted. In fact, no one knows what it represents.
And so the theory there, if it’s true, is can you achieve superior returns or consistent returns over a long period of time by removing human bias? And the way you do that, the theory is, is by making it a math problem. And the way to build the best model if you make it a math problem is to have as many people building models as possible, and then build an ensemble model on top of that. And the way you do that is to create the right incentive structure to do that. And one way is to make sure the people are anonymous so that they can submit their models from anywhere. I’m not saying that’s right or not, we don’t know, but it’s a different approach.
Robin: Yeah, one thing that fascinates me because it’s anything quant is really hot these days, and, like I said, I talk to so many big, old boring, mutual fund managers also moving into this space and hiring as many data scientists as they’re able to get their hands on, really. What happens when everybody’s doing this? I mean, for example, with this alpha decay are you noticing this happening quicker? Because back in the day when markets were far less efficient than they are today they were far slower, Renaissance could run some fairly simple trading strategies that would consistently make money. All they hear now, it’s getting more and more brutal that more and more money flowing into quant will eventually sow the seeds of either just people doing badly or some epic disaster at some point.
Christina: Yeah, definitely it’s gotten much more competitive, even over the last, like, two, three years, compared to when I first started off. And what’s happened recently is…it’s not that the industry is dying, which there’s a lot of news articles out there saying HFT is dead, it’s permanently gone and that’s not true. What’s happening is we’re seeing a lot of larger firms that are consolidating, they’re starting to differentiate their businesses, and they’re realizing that you can’t just go out into every single market and trade every single thing, you know, that’s not a practical business model here. Why don’t we focus on the areas that we do best? And so we’ve seen a lot of that too and for us also we focus on the niche areas that we’re good at, and that’s the only reason we’re alive today actually.
Robin: So what kind of speeds are we talking about? What kind of holding periods…how long [inaudible 00:13:16] hold a position?
Christina: So, like, we process data at the nanosecond resolution and in terms of speeds we can trade a couple microseconds in terms of holding period and up to a couple minutes sometimes. But this is…I’m saying this, like, it’s not like that’s our edge, everybody does this in our space, in high-frequency space. So maybe a 100, 200 firms out there that do that as well and that’s the area that, you know, the microscopic level that we’re playing in.
Robin: In the slightly slower trading side, do you worry at all that there’s so much money going into this space now that there’s just gonna be crowding and a lot of these signals are going to harder and fainter to find?
Jonathan: Oh, I certainly expect they’ll be harder and fainter to find but we’re set up to capitalize on that environment. I can imagine a world where if you’re a hedge fund and you do not have access to a massive global research capability that would be very, very difficult for you to continue to innovate, continue to find new signals, continue to process, you know, the avalanche of data that’s coming your way.
Andy: Maybe the endgame is there’s just one fund that’s left.
Robin: Yeah, well this is why a friend of mine, Matt Levin, wrote was, “One perfect capital allocating robot that just sits there and perfectly allocates capital to companies around the world.” It’s a great idea.
Jonathan: I mean, that’s unlikely because that fund has to trade with somebody else, there is a marketplace here.
Robin: But we’re starting to see signs that, like I said, I mean, I feel that everybody uses algorithms now there’s no kind of stock, there’s no paper being traded, it’s all electronic now anyway. So I feel like even talking about algorithmic investing sometimes gets redundant. But at what point do you things start feeding on themselves and eating themselves? Are there enough trading algorithms operating out of the wild interact in ways that we can’t foresee and that causes glitches. And we saw things like the flash crash, sometimes the big sell order by big, boring mutual fund interacts with people in your world and all hell breaks loose. I mean, that should logically be happening more as more money goes into more quant strategies, right?
Christina: Yeah, absolutely that’s definitely a possibility but at the same time, I guess, just from our perspective what we’re seeing is we’re seeing a lot more democratization of the space which actually makes it easier for…when we were starting up it was really hard, this is, like, five years ago by the way. It was extremely difficult back then to start up a quant or HFT hedge fund. But today if you started up…if you want data, there’s so many sources of data out there that just weren’t available back then. There are so many new sources of different things that you could use for your strategies that we had never even thought about back then because it wasn’t available to us.
And so definitely I think the opportunities are…it’s actually becoming more opportunities even though, sure, the marketplace has been relatively the same, but in terms of the ideas you can have and the types of transactions you can have even, that’s all constantly changing over time, for sure.
Jonathan: I wouldn’t agree though that the ascendancy of quantitative investing, you know, implies there’s greater systematic risk in the markets. I mean, I would take the opposite view. It’s really kind of the human conflict between greed and fear and poor incentive structures historically that have led to manias and crashes. So quantitative strategies are very attractive because, you know, they’re testable, risk management is encoded in the process. You remove the emotional aspect that can cause human bias. In our organization, we marry that with full trader oversight. So although our infrastructure runs in a fully automated way, we have a very, very experienced professionalized trading staff that oversees the systems in real time, and is able to act should anything ever seem questionable.
Robin: Yeah, and obviously humans were doing really dumb stuff with markets a long time before computers came around. But Andy, one thing that Christina mentioned that it was harder five years ago. One thing I’ve also noticed, a bit of an inflection point, in the willingness of people like you and the institutional investors to actually give money to, sort of, young computer scientists. Before the way to hedge fund to heaven was to be a star prop-trade at Morgan Stanley, Goldman Sachs and you went out and you did that and you raised a fund. These days, well they technically don’t have prop-traders anymore, but that’s a lot harder. But it does seem to be far more willing to say, “Hey, you’re a cool person who’s got a bit background in machine learning, you’ve got an interesting idea. I’ll try and support that somehow.” There’s seems to be more willingness now.
Andy: I think that’s a trend that has been in the macro or the innovation market for 20 years, it’s just starting to be applied to different areas, right? And that’s why we as a venture fund would participate in alternative financing that we have no business in doing it. We’re seeing characteristics of it that are…these are software companies, right? Or even Quantopian is, like, there’s software and community as part of what they do, you know? And so VCs should be investing in that, or you may wanna take money from VCs. So I think that’s what it is, it’s kind of that collapsing that’s happening in many different markets, this is no different than any other.
I think at the same time that we from the venture side participate in increased computing power and capabilities of technology and that’s the other trend that’s affecting this as well. And that we’re trying to think about and process which is that the machines are getting smarter, they just are, they’re not…
And I don’t see a scenario where that stops, I only see a scenario where that accelerates. And so there are lots of new businesses and challenges that will come from it, but that just is a reality.
Robin: What do you think are the challenges though? I mean cost of data, storage, [crosstalk].
Andy: No, I view the challenge as a little more of kind of like a cosmic level, right, which is, you know, the Turing test, right? So any computer that can pass the Turing Test is smart enough to know to fail it because it would. And so, and that’s like up a joke but it’s not really, right? When you have computers doing lots of different things, creative things, and intelligent things, how is that gonna affect different industries, how it’s gonna…finance in particular for today, but other industries and the implications of that are pretty profound. I don’t think we understand them by any means whatsoever but it is happening.
Christina: So, like, 10 years ago for instance, if you try to start a hedge fund 10, 20 years ago what do you do? You just have a strategy, you plug it into a computer and you’re all set, you know? And you can have a decent hedge fund going. But today if you try to start a hedge fund, there’s no way that’s gonna happen. You’re gonna need millions of operating capital, you need to develop…we developed our own order management system, our own data feed handlers. We developed everything in-house because five years ago about wasn’t available and that wasn’t… I’m sure a lot of you guys work…I’ve talked to many people here who work at companies that provide data, that provide feed handlers or provide different parts of the pipeline that we would need. But anyway the point I’m trying to get at is that we need venture capital, we need the financing from firms like that in order to succeed. And so that’s why, like, Quantopian raised a Series C earlier this year or last year…
Jonathan: In November.
Christina: Yeah, in November, and we were quietly also raising a Series C at the same time, you know? And back then you would never expect a hedge fund to be raising venture capital, you know, that made no sense. But today that’s becoming more and more of a reality because of how technologically intensive these businesses are.
Andy: So an interesting parallel or even contrast to that is in the crypto assets world where you can actually bootstrap a hedge fund. You don’t need capital to start. You literally need a generic Apple computer, Dell Computer nothing, no particular software and you could run a hedge fund in the crypto asset world.
Robin: Jonathan, you’ve had a long career in sort of the traditional hedge fund industry before you went to Quantopian. So why has the investing industry been slower? I mean, I think all industries… I’m a journalist, we’re incredibly slow, and conservative, and boring in a lot of areas. It’s still hard to get some journalists to use Excel. But why has the finance industry where it has a lot of money, a lot of brainpower, a lot of prestige been slower to adopt a different model? And what has been…I’m interested in hearing what the biggest change or the biggest “aha” moment you had from going from the old world to the new?
Jonathan: Well, I think it’s basic human anchoring bias. So there’s large organizations that exist for 20 years they’ve done very well, they’ve been doing something for a long time, and they’ve been successful. It’s become less successful so now but still historically it’s been successful so it’s very, very difficult for somebody to, you know, completely change their mode of business that’s really elevated them to a stratospheric level of income or wealth. So you really need new entrants to the space to do that.
So, like you mentioned, for the last 10 years I’ve worked in the traditional hedge fund business and I’ve observed a number of trends. And observed that alpha, the opportunity to find alpha has become diminished. Not only that but the ability to find talented people has become much, much, much harder. There are some regulatory changes in the industry that’s kind of cut off the feeder system from bank prop desks to hedge funds. And when I learned of Quantopian it was…kind of took me 10 microseconds to come to the conclusion that the business model that, that firm was growing into was really the evolution of the hedge fund space.
Christina: I actually disagree with you. So I actually think that the change has come extremely quickly in the financial industry. You look at like five years ago when “Flash Boys” came out everyone’s like, “Woah, this high-frequency trading is so cool,” and now it’s like the lamest thing ever. It’s like all old news and firms are consolidating etc. But I think the biggest…one of the reasons that I can come up with in terms of why is actually because of the Millennials. If you look at…these days, like, my nieces and nephews are taking computer science class in, like, elementary school, you know? And so what’s gonna happen is, you know, we have more control, at least for us I feel like the Millennial Generation, we have more…what do you call it? We want more control over our investments and where our money goes.
That’s why we have brokers, we have kind of people democratizing trading, we have all these different companies coming up. And, yeah, maybe that could be one of the reasons why. I think I feel more safe putting my money into, you know, a company that uses the techniques that I learned when I was in school, analysis techniques out there, that have been democratized that I learned back when I was in school. Rather than…I wouldn’t feel as safe putting my money with an analyst who uses, like, PE ratios to guess whether or not Google will go up or down tomorrow, right? I think that’s, kind of, the old school thing now actually. So quant is already becoming mainstream and I think in the next couple of years it will be.
Robin: So anybody thinking of doing a CFA, what you would say to them?
Christina: I mean, we don’t even hire people with that.
Robin: Sorry, guys.
Christina: Or like MBAs, you know, the value is no longer…maybe in consulting sure, you know, if you’re doing things like that but for the trading world definitely, it’s all about…these days if you want to enter finance you study math, you don’t study finance.
Jonathan: I mean, one thing that we’ve done is we’ve created a full educational curriculum free to the world on our site. So you can come and know very little about finance or algorithmic trading or even computer programming. And we have over 50 lectures now which you can take self-paced in our environment which will take you from very, very basic knowledge all the way up to the point where you can actually develop your own strategy on our platform.
Robin: But, I mean, just quickly, and we’re running a little bit short of time, but interesting hearing broad level where we’re heading over the next 10 years. Because, for me, one of the reasons that I find this really exciting is because I think there’s really interesting, important stuff going on today but what really gives me goosebumps is thinking what does the world of investing look like in 10 years time and how does that actually affect the markets? Because as markets become less human, as it were, that’s gonna change their reaction to funds. Maybe they’ll be steadier or maybe they’ll be more volatile, I don’t know. But Christina, maybe start with you, how do you see 10 years time?
Christina: Like I was saying, quant is gonna be mainstream. It’s going to be kind of the boring mainstream thing and, I guess, in other words, the face of the hedge fund industry is gonna be people who look like me, which is kind of different. Today people still aren’t used to that but I think you’ll see 10 years down the road you’re gonna find Millennials who run these billion dollar hedge funds and it won’t be as shocking as it is today, perhaps.
Jonathan: Yeah, I’d say much greater data, greater access, more efficiency in the market so steadier markets as markets react faster to data arrival. And organizations that will be successful, they’ll be concentrated, they’ll be large and they’ll have access to tremendous, tremendous resources that we can only imagine today.
Andy: I think it’s not inconceivable to think about in 10 years that Quantopian or Numerai could have millions of people building models for their vehicles, and that feels like an incredibly profound change if that happens. I know that’s part of your aspiration, it’s part of my aspiration, we don’t fully understand the implications of it but that’s a pretty profound change. Because many of those people will be the Millennials you mentioned, math students, kids, building models. They can learn data science from your courses and start building models. That’s a pretty different world, I don’t know what that means but…
Robin: What about VC? Is a robot gonna be doing your job in 10 years time?
Andy: We’ll see, I think so.
Robin: I went on a demonstration on natural language processing program a half year ago, it was scary, it was like…as a journalist. It still can’t make bad jokes like I can though, so that’s… We actually have just under two minutes left, I thought I’d end there, just in case there are any questions. Okay, we have one up here.
Man: Can I shout?
Robin: Shout it, yeah.
Man: [inaudible 00:28:13] model where millions of people are building models [inaudible 00:28:12]. Doesn’t that seem unlikely because [inaudible 00:28:16] isn’t it more likely that we’ll have big companies like Google that use supercomputers to make their models and all the little people trying to make machine learning algorithms in their basements are just completely outpaced by that. [inaudible 00:28:31] Numerai and this cryptocurrency system where you can wager your bets and bet [inaudible 00:28:35]. Isn’t it likely that we’re going to look in the future and it’s gonna be like Google owns 99% of the Numerai cryptocurrency?
Andy: I hope not because I think the incentive structure that’s built into the cryptocurrency, right, incents people to build models and then stake their Numerai on the models, right? The interplay between computers and people there are the creativity and coming up with the idea of the model or math problem. What you explain could happen, absolutely. But, I believe the incentive structure makes it really unlikely.
Jonathan: I would add to that, that in order to train AI to be a kind of super investor, you need to have data and we as an organization have this ever growing very, very unique and privileged data set of hundreds of thousands of independent trading strategies which really no one else has.
Robin: What if Google set up an internal hedge fund? Given the people that work there, given the data that they have which must be an absolute goldmine to any quant investor.
Andy: Do we know they don’t have that?
Robin: That’s a good point.
Jonathan: Probably not because they’re all spending their time on Quantopian.
Robin: Yes, exactly, well exactly, maybe that’ll be easier. Any last question, one last one? Somebody else with a quick one? Okay, well we’ll end it there, thanks so much for listening to us, and thanks to the panelists.
Christina: Thank you.
This report was created with data from CB Insights’ emerging technology insights platform, which offers clarity into emerging tech and new business strategies through tools like:
- Earnings Transcripts Search Engine & Analytics to get an information edge on competitors’ and incumbents’ strategies
- Patent Analytics to see where innovation is happening next
- Company Mosaic Scores to evaluate startup health, based on our National Science Foundation-backed algorithm
- Business Relationships to quickly see a company’s competitors, partners, and more
- Market Sizing Tools to visualize market growth and spot the next big opportunity