
During the Inside ETFs Conference, we had the opportunity to speak with the Buzz Index creator Jamie Wise and Executive Vice President John Ciampaglia from Sprott Asset Management. We discuss their unique BUZZ index and the innovation in social media sentiment data that uses machine learning and big data artificial intelligence.
ETF Database (ETFdb): Please tell us about yourselves.
Jamie Wise (J.W.): I’ve been in the investment management business for 17 years now. I started my career at CitiBank – managed its Canadian equity derivatives trading business down in New York. I came back to Toronto and spent some time at Bank of Montreal doing a various wide range of proprietary trading strategies, capital structure arbitrage in the early part of the 2000s. I spent a year at Moore Capital in Toronto and then launched Periscope Capital, which is a current business of ours, which is a Canadian focused hedge fund focusing primarily on non-Canadian institutions investing into Canadian traditional arbitrage strategies.
And BUZZ started; before it was called BUZZ, it was a project that we started looking at a little over three years ago. I was very interested in it. At the time, it was a very early adoption of some of the big data analytics with respect to the investment climate. And certainly, consumer product companies had been doing it for a number of years by then, and I could see that it was going to be applicable to finance, and wanted to get involved in that area. So we started looking at this data and this concept back then, but it really started taking off in terms of people talking online over the last couple of years, which gave us real confidence in the predictability of the model. That led to BUZZ, which we just launched at the end of last year. Our first index, which is the BUZZ Social Media Insights Index, went live at the end of December, with historical results available for the two years prior – so from January 2013. And since then, we’ve partnered with our friends at Sprott and Alps, and will be licensing the use of that index for a coming ETF product.
John Ciampaglia (J.C.): I’ve worked on Bay Street for over 20 years. I’ve worked predominantly for different money managers, and first got exposed to ETFs when I worked for Invesco. We acquired PowerShares back in … I think it was 2006 or ’07. And that was really my first exposure in a meaningful way to ETFs. And it has been an incredible evolution of product development. And at Sprott, what we really want to do is focus on different investment solutions that are more in the alternative category and are designed to do something different in the marketplace. We have a number of products on the New York Stock Exchange that we feel are differentiated and provide value to customers.
ETFdb: Why would you say they are differentiated?
J.C: It’s a case-by-case basis. Some of the funds we have are structured in a way where they provide tax advantages to investors, for example. In the case of our ETFs, they are all based on smart beta factors that aren’t in the marketplace today. And really, what we’re doing for each fund is we’ve come up with this term called “factors that matter,” which means basically, we’re trying to come up with some kind of factor, whether it’s product structure or some kind of intelligent way to weight and pick stocks that we think are good predictors of long-term performances of stocks. Nobody needs another version of the same fund, so we’re very focused on trying to come up with concepts that are very differentiated and unique. I think the marketplace has responded to the funds that we have in the marketplace. And with this new concept of using big data and social media, we think there is a real opportunity to bring something totally unique to retail investors.
ETFdb: Will you tell us the background of how this relationship happened? Is there a little bit of a story?
J.W.: Yes. I’ve known some of the senior management people at Sprott for a number of years, since we launched Periscope. We have a subadvisory relationship on a different style fund with Periscope managers right now. So we just have an ongoing natural relationship where we talk about ideas, we talk about what we’re working on. And as we were working on the big data analytics concept, I reached out to John and Peter Grosskopf, who is the CEO at Sprott, and mentioned it to them. The discussion was sort of fluid and natural from there, and we decided to partner together.
ETFdb: Jamie, how are some of the holdings in the BUZZ index selected? What is the methodology?
J.W.: Each quarter, we calculate the average total number of online mentions for the previous four quarters for all U.S. listed equities. The top 100 securities, with a market capitalization of at least US$5 billion and average daily trading volume of at least $1 million, are selected as the eligible investment universe. Each of the 100 stocks then is given an Insight Score based on our proprietary analytics. The 25 stocks with the most bullish scores are included in the index and weighted based on their Insight Scores. The index is reconstituted monthly to ensure we extract the latest insights of the social media community.
ETFdb: What is the value-add for investors when investing in your Sprott BUZZ Social Media Insights ETF (“BUZZ”)?
J.W.: The BUZZ Social Media Insights Index harnesses the power of social media’s big data to identify the best investment opportunities in the market based on actionable investment insights derived from the social media collective.
Machine learning and big data artificial intelligence models have been used by some of the world’s largest quantitative hedge funds over the past few years. Now, everyone can access the approach to add value within their personal investment portfolios.
The Index is dynamic and tracks which investments are trending among social media participants. The field of artificial intelligence and machine learning has really exploded in the last couple of years, and as consumers, we see that in all our products. You see the way that Google Analytics works, the way Google Maps works, image recognition – all of that is machine learning and artificial intelligence. And that’s how I think our model differentiates itself in that it relies on those concepts more than just more simplistic keyword analysis, or more simplistic sentiment-only focused models where they’re trying to understand “Is it positive or negative?” assigning a binary one or a zero to online discussions.
Our model really captures so much more. It captures the way people are thinking about brands and stocks online from an investment perspective and from a brand perspective. It’s able to recognize when we’re talking about someone who is making a stock trade, or recommendation, versus are they focusing on the brand experience. And then even more importantly, we identify influencers, and we identify user reliability – and that’s unique. So that when we’re coming up with insight scores, if you have something to say about Exxon Mobil and John tweets something about Exxon Mobil, they can have very different weights within our model. Because you may have 10,000 followers and be deemed an influencer, and we can measure that in how fast people are, let’s say, re-tweeting your posts or sharing your blog posts. Whereas John may have 30 followers and no one really cares – so influence matters. Because ultimately, what we’re trying to score is what people are saying, how likely is that to propagate through the social network, and ultimately into that asset price. Equally important, we measure reliability.
So you may be great at having big influence around your discussion points around Exxon Mobil, and people re-tweet those often. But it may turn out that you’re actually a bad predictor of its stock price, whereas John may be the opposite. So we weight those things accordingly in our model. And all of that will lead to an individual score for you in the model relative to John. Now imagine multiplying that by tens of thousands of people, or even hundreds of thousands of people, depending on the stock, and we can come up with a more insightful measure on not only what will happen with that stock price, but how likely that is to happen based on the people who are generating the content. So it goes beyond, so much more beyond just sentiment reading, just sort of good or bad analysis. It’s really contextual. And it is smart and sort of intelligent in that a model learns over time. Because as we follow you over time, your influence ranking changes consistently, your reliability changes, and so the model reacts to that, and we believe that leads to a better outcome.
ETFdb: Are there any specific social media platforms you look at?
J.W.: We look at over 50 different platforms, websites, blogs, news services. If you look at our website, there’s a little model there; you can see a selection of the 50 on the website. But it’s all the main ones that you would think of. The explosion in the data really took hold when Twitter started cashtagging, and started the cashtag process. That allowed a real easy way for people to link themselves into stocks. And that also sort of … it was that rising tide that lifted all of the other boats. So as people talked more on Twitter, they started sharing those stories on other social forums, were able to comment on those forums, so the data just exploded, not just within Twitter but across the social media landscape.
ETFdb: John, is there anything else you would like to add?
J.C.: Well, I think what’s interesting and what got our attention was platforms like Bloomberg and Thomson Reuters, who are providing sentiment scores into their terminal today on individual stocks. Discount brokerage firms in the United States are providing sentiment scores on individual stocks to their customers. We would view that as more short-term-trading oriented, or more point in time to get a read on a stock based on what sentiment is telling them from social media.
But it really doesn’t translate into how do you put the whole package together for investors, and how do you do it in a way that’s dynamic. Because the trend is changing all the time depending on what people are following. And I think taking this engine and putting it inside of a financial product makes that technology accessible to the masses, and I think that’s really powerful. Whereas today, it’s hard for an individual investor to really put the application to work. More sophisticated investors, like hedge funds, are using daily sentiment scores and social media scoring for shorter-term price signals. But again, the typical person can’t access those kinds of investments. So I think bringing this kind of intellectual property down into a product wrapper that people can access is really the value-add as well, because we’re providing access to something they can’t get today.
J.W.: I’d add, too, that our insight scoring methodology isn’t meant to be a trade; it really is a factor, an investment theme. So you can think of growth, value, momentum and insight score as other factors that someone should employ within their portfolio that can add value. As John mentioned, there are some applications for this to try and scrape data. … A great example is Twitter. They inadvertently released their earnings online. And you could have your big data analytics engine find that and then profit from it, because you could read the news release, or it could interpret the news release before it was more broadly disseminated. That’s not an investment strategy, that’s a trading opportunity. And what we’re doing and trying to deliver to investors everywhere is the opportunity to use this type of analytics within an investment strategy, and not for that real short-term trading opportunity. It’s a different approach.
ETFdb: What inspired the creation of such a unique ETF?
J.W.: The BUZZ investment process has been in development for over three years. Early on we were excited by the concept of applying the same big data analytic techniques used by leading consumer brand companies to the field of finance. As investors embraced the concept of sharing their investment experiences and ideas with the online community – an approach that took hold with the introduction of the Twitter $cashtag – the volume of data exploded over the past several years. The depth of the data from the social media collective now allows us to have confidence in our investment insight scores and predictability of the insights – an idea that simply was not possible just three years ago.
ETFdb: What is your take on the current volatility in the markets? Do you think it’s a temporary correction, or a bear market that will persist?
J.W.: Volatility is a natural occurrence in financial markets. One could argue that the environment over the past few years was abnormal in that volatility was unusually subdued. That said, strategies that have worked well in the past may not be ideally positioned going forward. Having a good understanding of how people are thinking with respect to their investments, including which sectors and themes are more likely to lead markets on a go-forward basis, is critical in this environment. As more people look to various social media platforms to collaborate, we are excited about harnessing those insights and profiting from the BUZZ.
ETFdb: We started with one of the worst weeks in January; it has been a volatile month so far. So what do you think is going to be the performances for the markets in 2016?
J.W.: We do some other analytics within BUZZ that allow for capital deployment, full capital deployment, or maybe keep the cash balance, depending on the overall bullishness of our readings, and depending on multiple factors being met. What we’ve seen going back to July of last year, it was the first time in three years that model was not fully deployed. It went back to about 75 percent invested, 25 percent cash waiting. And then it oscillated around there, got back to around maybe 90 percent in the fall, and then again started dropping – but this time at a much faster rate.
So those models, as of today, are only 50 percent invested in cash. What I think is telling is that this is for real, this era of volatility. I wouldn’t say it’s necessarily a new era – I think it’s a return to some more normalcy. And there will be some; markets can’t move 2 percent a day, every day for years and years, so you have to get through those anxieties. But what is unique about BUZZ, and what it can capture, is that even though this model was meant to give you consistent full-time exposure to the equity markets, you’re sort of always in a point where you know that you have the best momentum from a social perspective – not from a price perspective – and that’s a different concept to think about. But you have that social momentum and social demand behind these stories, and all in all, they should continue to at least perform as well or better than the overall broader indexes.
ETFdb: What is the best piece of advice you would give to investors for 2016?
J.W.: Share, collaborate and stay social!
ETFdb: What has been your biggest insight from the conference?
J.W.: My biggest insight is that there’s not more discussion about the application of big data analysis and machine learning concepts to financial products. I’m really surprised about that. Just given what I see in other industries, given what I see in some of the world’s largest quantitative focused hedge funds where they’re investing their money, I’m surprised that hasn’t yet trickled down so that the retail investor can access that – which is what we’re doing.
J.C.: I would say, generally, I’m amazed every year, just the growing adoption to ETFs. As someone who started on the active mutual-fund side of the business, it had an incredible period of growth, and I’m always amazed at how much excitement and innovation is going on in the ETF space. I see a lot more innovation happening here. I see a lot more excitement of people working on this side of this business. So I think the secular drivers that are fueling the growth are not going away. People are looking for more innovative solutions, lower pricing, more transparency, liquidity, more tax efficiency. And the growth of how ETFs are being packaged and sold, and some of the solutions with providers, I find fascinating.
Because I really do think that as millennials start to build some discretionary wealth, they are going to transact differently in terms of what advice they buy, how they buy the advice, what products they buy. And that’s what I think is very interesting about this concept; it will appeal to someone who is more in tune with living a life that’s more socially based and transparent. When you think about going on a holiday, you go to Tripadvisor because you want to hear what other people are saying. Not what the company is saying, but what is the real story? What do the real pictures look like? And we’ve applied that concept to so many different things in our lives. So that this power, this idea about the power of this crowd on social media platforms, I think is going to appeal to particularly younger people who are more in tune to social media and the power of it. Consumers are more empowered than ever with information, and I think this fits very nicely with that theme.
Find out more information on the BUZZ Indexes here.