There are an estimated 445 alternative data providers in the funds industry, serving the needs of both traditional and alternative fund managers. This is, according to alternativedata.org, an industry that is projected to be worth USD350 billion in 2020. But while there is no doubting the quantity and diversity of data sets one can acquire, how valuable really are they?
There are an estimated 445 alternative data providers in the funds industry, serving the needs of both traditional and alternative fund managers. This is, according to alternativedata.org, an industry that is projected to be worth USD350 billion in 2020. But while there is no doubting the quantity and diversity of data sets one can acquire, how valuable really are they?
It is a valid question when one considers that the average dollar spend on alternative data, for 2020, will be USD158K for those running less than USD1 billion in AUM, rising to an estimated USD764K for those running north of USD1 billion.
Investors will no doubt want to know what kind of return on investment such an annual outlay is yielding, and whether these data sets are truly able to improve the investment process. To some extent, the sheer pace of technology advancement makes it nigh-on impossible for any serious fund manager to completely overlook this option. After all, we live in a time where “AI”, “machine learning” and “alternative data” are thrown into conversation with great gusto, as managers seek to position themselves as being on point with market trends. Heaven forbid one may look out of vogue.
How one utilises these data sets, though, should be observed with the same focus as any other aspect of the trading process. Picking which alternative data to use is far from straightforward. It is also, arguably, more additive to discretionary managers than the higher frequency traders and quants who need to act on signals in short timeframes.
Most of the alt data sets are built on equity markets, given the enormity of publicly traded information. Applying them to futures markets, where trading on OTC market still dominates, is more challenging.
I spoke recently to the head of quant trading at a London-based fund platform, which supports a mix of both discretionary and quant fund managers. On condition of anonymity, he shares the following anecdote: ”I have heard about stellar data sets on UK oil reserves but the reality is, if someone really had found such a data set to premise the production or consumption of a commodity, they wouldn’t tell the marketplace about it.
“I think the real value of these data sets is still marginal.”
Speaking at last year’s ABN AMRO Amsterdam Investor Forum, Rani Piputri, Head of Automated Intelligence Investing, NNIP (a Netherlands-base investment manager), said that how one defines a good alternative data set depends on one’s investment process. “We’ve found some alternative data providers who can provide us with very good equity analyst estimates,” she said.
Alternative data is typically speaking, relatively slow. It tends to be related to company fundamentals and usually it is based on monthly corporate updates. There isn’t much alternative data in real time that can have an impact on price. Most alternative data sets are used as leading indicators ahead of big fundamental announcement such as corporate earnings.
For a short-term trader, monthly data does not necessarily offer much value in the search for alpha. But for discretionary managers with an active short book, alternative data sets have the potential to help a manager build an investment thesis around shorting a particular stock, ie using satellite imagery to determine car production in China, for example, or the number of cars in supermarkets or retail parks.
In some respects, picking the right data sets is a skill in and of itself, just like stock picking. The biggest quant shops on the street have the budgets to buy the best data available; and by default, the highest quality data. That’s not the case for the vast majority of the hedge fund firmament.
“The managers on our platform trade futures, they are short-term traders and they are modest in AUM. Combine those factors and that is why, for us there is no real use for alternative data sets,” remarks the quant manager.
He is quick to point out though, that he does see value in using them in the discretionary trading space, noting they can be additive when doing a deep dive into assessing a company’s true value: “A discretionary trader looking at aviation stocks might use an alt data set that provides information on employee satisfaction, or supply chain costs, to build long and short positions.
“We are at an age where technology is allowing us to aggregate data from the bottom up and apply statistical tools to improve our reasoning. It is a big shift.”
One manager who is actively utilising alternative data sets is New York-based Wavelength Capital Management, which focuses on factor-based fixed income analysis. Its co-founder, Mark Landis, tells me that one basic premise behind Wavelength’s investment process is that technology has commoditised many traditional data sets, thus making markets more efficient. The traditional, one-dimensional edge of older traditional money managers is harder and harder to identify and take advantage of.
“We absolutely utilise non-traditional data sets along with traditional,” says Landis.
“But remember, in fixed income which has historically been an OTC market, attaining proper historical data sets in such markets as credit, MBS etc., has been very difficult. We utilise both machine learning and deep learning (AI) to process historical, fundamental and non-traditional data. Advances in processing power have made it possible to identify patterns through filters with more depth than a simple chart of prices.”
I ask Landis about the quality versus quantity challenge faced by managers, and the issue of cost versus opportunity gained.
He responds: “One needs to know what to do with it. The volume of financial market data is enormous and growing daily. Deciphering signal from noise, as with all data sets, represents the key to unlocking opportunity and a true edge versus wasting money.”
He thinks that applying deep learning to the universe of big data offers the greatest investment potential.
Over the next few years, the 5G revolution is going to change our lives even more considerably than we’ve seen over the previous decade. It will unleash enormous processing power, allowing computers to crunch through slews of alternative data in near real-time.
The quant trader pauses when I suggest this and offers the following concluding remark:
“Before going to the millisecond, or something closer to real time, the next step should be to get right the quality and accuracy of data. There will be a big premium for those who can gather high quality, clean data to gain an edge in the market.
“I think managers will become more qualitative in how they use alt data sets in my opinion. Quantity is not the challenge … quality is.”
