Redefining Capital Markets through Alternative Data, with Carrie Shaw
What does it take to find alpha? Data is transforming the world of finance — this is what it looks like on the ground.
Carrie Shaw is the Chief Marketing and Product Officer at Quandl. She previously served as the Director of Marketing for Rogers Incubation, and worked across product at Microsoft in Seattle. Carrie is also the Chair of the Board of Directors at Progress Place, a mental illness focussed nonprofit.
Here’s Carrie at the August 2017 edition of FintechTO:
Who in the audience has heard of the term alternative data as it relates to capital markets ?
Not a lot of hands, which is not surprising given that it’s early days for this category. It’s been around for about a year and a half. Tonight I’m going to talk about alternative data and why it matters so much to capital markets.
All active investors seek to do is to beat the market. In the financial services industry this is known as capturing alpha. Alpha is a zero sum game — the profits earned by a winner will always be offset by the losses sustained by a loser. Not everybody can win in this game, and that means that markets are extremely competitive. Investors are constantly looking for an edge, a source of advantage that they have that no one else has. This has been the case since the dawn of investing.
In 1865, a group of traders orchestrated a scheme where they had French actors visit the Chicago Board of Trade to ask about acquiring a winter’s worth of wheat for the army of Napoleon III. Speculators rushed to buy wheat futures and of course Napoleon’s so-called emissaries disappeared within hours, leaving the speculators high and dry and creating vast profits for the organizers of the scheme. This was known as manufacturing information and it was a source of alpha.
In the 50s and 60s a man named Alfred Winslow Jones figured out that he could beat the market by pitting the stocks in his portfolio against one another. He would long one and short the other.
Today, this sounds basic but at the time this flew in the face of conventional wisdom which was all about market timing and only looking at macro factors.This notion of hedging stocks against one another gave birth to the hedge fund and Jones and his contemporaries would enjoy this edge for more than a decade before other people caught on to it. Today, the things that Jones figured out are taught in finance 101 classes. There is no edge to be had there anymore.
In the 1980s the scientific calculator still reigned supreme and investors used it to price bonds one at a time. The advent of computers meant that we could now perform hundreds of these calculations per second, which offered a speed advantage, if only for a short period of time.
So you can see that these sources of alpha come from different places. They could arise from new technology. They could arise from new investment models.
They can also arise from information. When you think about fundamental analysis and fixed income research, in the 1970s a small company called Reuters started piecing together faxes of financial statements that they had captured from annual reports and this was the source of alpha and really all it was was early fundamental analysis.
In 1982, Bloomberg launched its terminal and they had fixed income prices that nobody else had. Same deal. Today, we think of fundamental analysis and fixed income prices as table stakes. Everyone has to have them or they’re at a disadvantage, but nobody gets an advantage just by having them.
So there’s always an edge to be found, but if it is a real edge it will eventually diffuse so completely that it provides an advantage to no one. However by the time that happens everyone has to have it. Today’s edge is again from information and increasingly that information comes from what we call alternative data.
Alternative data is information that has not previously been used in the capital markets industry but is nonetheless very valuable to investors. It always arises from sources well outside of what we traditionally think of as financial market data.
Here’s an example. When you buy a book from Amazon who knows about that purchase? You know about that purchase. Amazon knows about it. Your credit card company knows about it. The GPS unit in the truck that delivers the book knows about it. The satellites watching the GPS unit know about it. Even the security camera in your condominium’s lobby probably knows about it. So there are a lot of sources out there that can tell us about this particular purchase. This data is hard to access, but if you can get it, you can paint a bottom up granular near real time view of an Amazon transaction. Now imagine doing this across millions of transactions. That is the promise of alternative data.
“Data are to this century what oil was to the last one: a driver of growth and change.”
-The Economist
So, why now? Because we have seen the dawn of the data economy. In May of this year, The Economist referred to data as the world’s most valuable resource and the fuel of the future. The technology boom of the 90s along with Moore’s Law improvements in processing power and storage, plus our outrageous need to measure everything has produced a world in which virtually everything can be measured.
The prevailing belief is that there are predictive signals buried in the exabytes of data that we’re creating and these signals have the power to move markets. It’s on us now to find them. Put it all together and we’re entering a new age of alternative data. The sheer scope of data available plus the history of data innovation have combined to produce an industry that is going to see massive returns in the coming years.
And people are ravenous for it. The early adopters of alternative data are quantitative hedge funds. But just like all other sources of alpha, we believe this trend will be pervasive across all capital markets, all forms of investing, all asset classes, globally.
I want to talk about what Quandl’s role is in all of this. Quandl is a marketplace for alternative and financial data. We’re an online marketplace just like many other marketplaces you know about. The difference is we don’t sell a product or service, we sell data. But we didn’t start out by selling data.
We started the way most startups do back in 2012 when our founders Tammer Kamel and Thomas Abraham needed to solve a problem in their previous lives as hedge fund professionals. There was no good way to consume the myriad amounts of financial data they needed for their trading strategies. They set out to solve that problem and the first iteration of Quandl did just that.
Fast forward a few years. In April of last year, Tammer and I were at a conference in New York called the Battle of the Quants and he said to me, “you know, Carrie, I really think we need to get some alternative data onto the platform.” And I said, “What is alternative data?” And so he explained it to me like I explained it to you and I said,
“Gee that kind of sounds like a pivot. Can I think about it? Can we talk about this next week?” And he said, “No we can talk about this right now.”
So Tammer, once he saw the trend, realized that we needed to move on it very quickly. We rejigged the Quandl marketplace to support the incorporation of alternative data on the platform. And it turns out it wasn’t so much of a pivot as it was an evolution to what we were already doing, and it was a lucrative one. In the first quarter that we had alternative data for sale on the platform, our revenues increased by 900%, and they’re still going.
But it’s not always easy to find it. One of the big challenges with finding data supply, beyond the fact that there just so much of it and you have to know where to start, is that there are many steps to determine whether it is going to be saleable or not.
The data needs to be a leading indicator, not a coincidence or a lagging one.
There’s a lot of data out there that might say something about Amazon or Alibaba or Exxon or any other security that you can think of, but if it doesn’t say it faster or better than what’s already available it’s going to be of no value. That’s why the data needs to be a leading indicator not a coincidence or a lagging one. In another example you could have data that’s really highly correlated with some type of security, but the correlation makes no logical sense. I’ve always enjoyed this example. The divorce rate in Maine happens to be very highly correlated with the amount of margarine consumed. But there’s absolutely no logic to that whatsoever. It’s a useless thing to know.
There’s about one hundred other ways a data set can go wrong. We estimate that for every ten data sets we assess to put on the Quandl platform, only one makes it through. There’s a real challenge in sourcing supply. At Quandl, we employ a team of data scientists whose sole purpose is to be out there hunting for and evaluating the best on offer from the data economy. The size of our data science team rivals the size of our development team which shows how much of an investment we put into the data itself.
Even if you have a saleable asset, you still need to make this data consumable by a Wall Street audience. You can’t just flip a CSV file over to a hedge fund and be done with it because that’s not the way they want to work. You have to harmonize your data. You have to make it available via API. We support other modules that financial services people like to use, like Python and R. Investors, especially quantitative investors, typically don’t want to consume pure raw data.
You have to package it up as an index for the quantitatively minded and level it up with some analytics for the qualitatively minded. To that end, Quandl offers data in a manner designed for quantitative investors at the moment, most often consumed by API and we’re now working on an offering that will service the broader discretionary side along with the sell side that will be more analytics driven.
We divide alternative data into three generations.
The first generation contains data you’ve probably heard of — sentiment data like data from Twitter, data from the news. Sentiment data is the oldest form of alternative data and it predates the term. Most people believe it’s been priced into the market already so there’s no advantage to be had there anymore. In the second generation of data which is where we are right now, there is a lot of what we call business exhaust data, which is a byproduct of what is already being done for a business’s operation. I’ll give you an example from the insurance industry in a minute and it sounds excruciatingly boring but it probably has the most potential for alpha out of any of the data sets out there. Logistics and supply chain is very exciting too.
The third generation is where you hear all the fun stories about drones taking pictures of agricultural crops. But that’s really just fodder for the media right now and people aren’t using that in the institutional investing world yet, but we’re keeping an eye on it.
Quick case study, Tesla. Tesla’s been in the news a lot lately. Why? Because they have too vastly ramp up their production in order to support the demand for the Model 3. Quandl has a dataset that is created from several U.S. insurance providers who give us a daily count of the number of policies that they have written for new cars. This is an extremely powerful way to approximate new car sales because when you buy a new car the first thing you do is you buy insurance for it. For somebody holding a position in Tesla, we can tell them on a daily basis whether Tesla is making good on their promise to deliver these Model 3's from the policies that they’re issuing. That’s a powerful dataset and one that does well.
We find ourselves in the precarious very startup-y position of being early in the hype cycle. While it’s early days for the adoption of some of these more esoteric data sets, the competition is heating up and this is what keeps me up at night. How do we sustain our advantage here? I think the answer really boils down to two things. We hire the best people and we find the best data.
Ready to meet and learn from Toronto’s best?
