Clearing up some misconceptions about alternative data
FinTech Influencers, a technology leadership forum founded by The Realization Group (TRG) together with Harrington Starr, recently launched a new series of Advanced Technologies Special Interest Group meetings to explore use of new technologies in the Financial Markets sector. This inaugural meeting, which looked at the changing role of quants in an age of alternative data, featured Steve Wilcockson of Geospatial Insight as a guest speaker and was moderated by Mike O’Hara of TRG. Steve and Mike report here on the findings from the discussion.
It’s time to clear up some misconceptions floating around in the markets about alternative data.
You don’t need to search long before you’ll hear stories about quantitative analysts (“quants”) who have used mobile phone- or credit card-based information to build valuable snapshots of the specific shopping habits of millions of people. And there is no shortage of fund managers who have begun to use satellite data to gain advance notice about crop yields or consumer retail traffic trends.
But as word spreads about just how powerful alternative data can be and the many ways it can be applied, so too do a lot of misplaced ideas, leading to myth, obfuscating hype and a biased skewing of an otherwise exciting industry.
One of the more common, and potentially dangerous, ideas is that alternative data can simply be plugged into a buyside system to produce miraculous alpha. It’s certainly true that new forms of data can reap rewards – but not in the magic wand-waving way that some people may believe.
Part of the reason this plug-and-play notion has gained traction is because of conflicting ideas about the role of the fabled quant and more recently, the data scientist.
During our discussion, we reminisced about the “seven ages of quant”, recalling their contributions in recent history to derivative pricing, portfolio theory, arbitrage, high frequency and systematic trading across most asset classes, and now machine learning and artificial intelligence. Whether on the way they helped contribute to the credit crisis was an accusation the audience generally refuted!
However, audience participants noted that “quants can be odd things” and that the job title itself is self-prescribed. True, but we noted there were different sorts of quants, for example risk-neutral “Q” sell-side quants and “P” probabilistic portfolio construction buy-side quants (as Attilio Meucci explains so well in “P versus Q: Differences and Commonalities between the Two Areas of Quantitative Finance”). There are others too: risk quants, central bank quants, even quants working in insurance, with all types slicing and dicing alternative data in different ways.
This brings us on to the subject of alternative data, where the traditional view is that the quant is at the forefront of the new buy-side revolution, generating alpha. The audience was largely united in asserting that this perspective not only oversimplifies what quants on the buy-side actually do (and what they can do), but also forgets or ignores those working in other sectors.
We considered the buy-side first. As one audience practitioner noted, rather than throw “data mud” to see what sticks, a better approach is to start with a hypothesis and test it first. Another frustrated discussant noted that data sets have quite limited asset class focus – “it’s funny how every alternative data discussion descends to consumer retail” (cue some embarrassed speaker foot-shuffling from our speaker, who is currently working on a retail trends data-set) “although there’s certainly value for physical assets such as oil and crops” (happily the speaker’s firm works on those too!).
Another participant noted that “data lends itself best around the edge”, but that the edge is often limited and subject to alpha decay risk as everyone jumps aboard the gravy train, even those crowd followers suffering from fear of missing out. Most agreed that data sets were unusable in isolation, having to be ingested into more comprehensive aggregate databases designed to serve the business appropriately, their rules often designed, enforced and “engineered” by quants, with (data) engineer being an appropriate descriptor for the new quant.
Refuting the buy-side myth is only part of the story. Remember those sell-side, risk, central bank and insurance quants? They’re using alternative data too, but not necessarily for alpha. In risk and on the sell-side, alternative data sets are informing fraud detection systems, chasing alpha generator misdemeanour rather than creating alpha. Bank research teams, in reorganisation mode after MiFID II, are increasingly taking on some of the buy-side functions, also ESG research too. Bank risk teams are starting to explore satellite, drone and ground-based imagery in order to assess loss-given defaults (for example retail mortgages and commercial lending) in areas of environmental or social stress, against which they increasingly take out insurance for regulatory capital protection as well as loss mitigation. Indeed, in insurance, imagery is almost mainstream in claims detection, for example in environmental catastrophes.
In short, the changing role of quants in an age of alternative data is not all about the buy-side. And where it is about the buy-side – as the audience discussion vigorously asserted – the value in alternative data comes from the idea of uniqueness, not ubiquity.
Find out more about FinTech Influencers – join the discussion