Digital Assets Report


Like this article?

Sign up to our free newsletter

Discover the future of data at the Amsterdam Investor Forum

Related Topics

With the 8th annual edition of ABN AMRO Clearing’s popular Amsterdam Investor Forum looming large on 20 March, Global Fund Media Editor-in-Chief James Williams previews what promises to be one of the event’s essential panel sessions, the Future of Data, which will explore how alternative fund managers are utilising data to inform their investment decision-making…

Cloud platform technology and an ever-growing interconnected global world have led to an unprecedented explosion in the volume and diversity of data. Data dominates both our professional and private lives; an endless stream of information that informs our health and fitness, recommends music and TV choices, gives insights on what the latest consumer trends are, not to mention corporate and financial market data. 

Everywhere we look, we have information at our fingertips. According to Statista, global IP traffic from data centres in 2018 was a staggering 10.6 zettabytes (one zettabyte is 1 billion terabytes). Big data has become big business, with revenues reaching USD42 billion last year. As the volume of data continues its exponential rise, these revenues are forecast to reach USD103 billion by 2027. 

How then, within a financial markets context, are alternative fund managers thinking about this brave new world? How are they utilising data in novel ways to inform their decision-making? Against this backdrop, a panel session – The Future of Data – will explore the implications of this data explosion and what techniques might come to the fore to make sense of, and generate actionable intelligence on, myriad data sets. 

The session, moderated by Stewart Jardine, Director, Market Technology & Data Services, CME, will take place on 20 March, 2019, on the second day of the 8th annual edition of ABN AMRO Clearing’s hugely popular Amsterdam Investor Forum; a leading forum for institutional investors and alternative investment managers. 

Panelists will include Rani Piputri, Head of Automated Intelligence Investing, NNIP; Michael Weinberg, Chief Investment Officer and Partner, MOV37, and Guillaume Vidal, CEO, Walnut Algorithm. 

Speaking with Hedgeweek ahead of the panel, MOV37’s Weinberg shares some views on how machine learning, combined with massive computing power and sophisticated data analysis has given rise to the “third wave” of investment management. 

Deep learning algorithms are being used by a new breed of hedge fund managers, which MOV37 has named autonomous learning investment strategy (ALIS) managers, to find patterns in large data sets that in turn drive their investment strategies. 

Whereas the second wave was characterised by quant fund and was largely hypothesis-driven, this third wave is data-driven.

“There are two primary attributes we look for in these managers: either the strong use of alternative data sets and/or the strong use of machine learning,” explains Weinberg, who is also a Co-Founder of the Artificial Intelligence Finance Institute. “We want to see evidence of how they are using machine learning techniques on the data and that they understand where the alpha is coming from and why.”

The future of data will likely evolve as the world becomes even more deeply intertwined, as evidenced by the Internet of Things (IoT). According to Ericsson, around 29 billion connected devices are forecast by 2022, of which approximately 18 billion will be related to IoT. 

As more devices speak to one another, it is giving rise to a vast array of unstructured data sets. 

“They include data sets that people have obtained creatively, particularly those that have been scraped off the internet (on government data, municipal data, corporate data etc.): anything that you would not categorise as traditional market data sets,” says Weinberg.

And the more data contained in the data sets, the better the machine learning techniques become. For context, the global algorithmic trading market was valued at USD9.30 billion in 2017, and is projected to exhibit CAGR of 10.1 per cent over the forecast period (2018-2026), according to data released by Coherent Market Insights. This is a new arms race and those who process and analyse data in original ways will steal an edge on the competition. Hedge funds are subsequently evolving to hire data scientists and engineers, not just financial whizz kids, and putting sophisticated data management programmes in place.

According to MOV37, structured and unstructured data is growing exponentially. By the year 2020, new data four times the size of the entire US Library of Congress will be created every second. 

As such, investment managers must have a robust process in place. Care needs to be taken that data sets are not over-fitted, which machine learning lends itself to. To guard against this, MOV37 has a PhD on its team to help validate the robustness of the investment process, when scouting for the next best ALIS manager to invest with. 

One of the potential risks to the future of data is how they are sourced and how well protected they are from cyber attacks. If systematic managers increasingly ingest third party data sets, they could contain information that nefarious actors attempt to misappropriate. Also, there are risks that some data could be deemed Material Non-Public Information (MNPI), exposing managers to regulatory investigation. One only has to see what happened to Cambridge Analytica to appreciate this. 

Weinberg believes there is much more risk of discretionary managers trading with MNPI, however, which they might have obtained from a corporate management team.

“The nice thing about ALIS strategies,” he says “is that all of the data sets that are scraped from the internet are anonymous and already exist in the public domain. 

“If data is acquired there has to be no personally identifiable information and the manager has to ensure that the person giving it to them has no fiduciary duty and isn’t violating any regulatory standards.

“We would argue, therefore, that in a world where you have SEC regulation FD (Fair Disclosure), machine learning investment strategies actually avoid potential insider trading issues that discretionary managers might fall afoul of.”

Hedgeweek spoke to Deloitte on this topic last year. Doug Dannemiller, Research Leader, Investment Management, Deloitte Center for Financial Services, believed that there were “some ticking time bombs” with respect to MNPI. 

“Innovators (such as quant hedge funds) saw the opportunity for alpha and went headlong into it, although so far no investment firm that I’m aware of has blown up as a result of using alternative data. 

“But it is a problem. Investment managers have to avoid MNPI at all costs. In a firm like Estimize, it has algorithms and user input profiles that can scrub and eliminate any signal generated by MNPI. The rules for MNPI as they translate into the cyber world don’t necessarily apply. Just because somebody with a server and some bot-programming capability can get data does not mean it is public. The rules on what’s public and non-public are difficult to navigate. They need to evolve,” said Dannemiller. 

One other intriguing aspect to consider in how data might be used in the future relates to the recent phenomenon of ‘fake news’. With so many different social media platforms and chatrooms and messenger groups, stories on world events – real or not – are spreading like wildfire, fuelled by the likes of Twitter. The problem here is that chat bots are being used to generate entirely baseless stories, which trading algorithms need to be able to spot and differentiate fact from fiction. 

When asked on this point, Weinberg concludes: “The algorithms need to understand what is important and what is not important. When a piece of news is put out into the public domain, you have to be aware of it and/or aware that if it is in fact fake news, what the impact is. The reality today is that fake news moves the markets in much the same way as real news.”


Like this article? Sign up to our free newsletter

Most Popular

Further Reading