Quantamental Investing & Alternative Data: A Look Under the Hood

By PJ Kaplan, Director, Americas Sales, AQMetrics

Alternative data is fast becoming one of the hottest topics in finance today, due in large part to the role it can play in another key emerging sector: quantamental investing, a hybrid fund strategy which combines fundamental analysis with quantitative investment models. But the use of alternative data - data about firms obtained from a variety of less traditional sources - to gain these investment insights may introduce some challenges that come along with its opportunities. 

Bloomberg and PRMIA hosted an event dedicated to exploring this topic in greater detail, addressing many of the key questions around alternative data, its potential use in the investment space and what wider implications this may have on the financial market as a whole. 

Defining alt data
Marc Groz, Regional Director, PRMIA Stamford chapter, began by introducing these two new concepts - and also the expert panel who would be looking at their potential consequences in greater depth. To tackle the definition of alternative data and its characteristics, Damien Weldon, Managing Director, Business Development, Vichara Technologies, delivered an in-depth presentation:

Three Key Areas 
  1.  The data must be at a very low level of granularity;
  2. That it must be obtained from non-traditional sources, not typically used in investing; and
  3. That is has some impact on the valuation of a financial instrument 

Aim of using large, non-traditional data sets
  -  Gain greater insights into the behavior of a firm's’ customers 
  -  Better understand the firm, and the value of its assets

Key Risks
  -  Data matching
  -  Data governance

Damien concluded his presentation suggesting there may even be a need for a regulatory framework in firms, allowing them to not just talk to investors about how good such models are but also about how robust they are from an accuracy and compliance standpoint. This was an area where he expected to see an increased focus in the years to come. 

New world, new issues
Michael Ho, Founder and CEO, Quantavista Research, agreed, adding that the important difference of alternative data is that is can provide real-time insights and enable funds to make increasingly accurate predictions. He also warned that the world of fundamental data is moving very slowly, so firms such as his were helping to ‘reimagine’ investing and use alternative data in a different way. His view was that such data is effectively a digital business, providing the delta between what your system will tell you and what it is being priced as. 

Rob Reider, Senior Advisor, Quantopian and Adjunct Professor, Courant Institute, NYURR, argued that because everyone thinks big data and machine learning are doing great things elsewhere, that they must inevitably also be able to work in finance. But there are a few differences to be aware of, he claimed. For example, in finance the signals are extremely noisy and the data sets are extremely small. This can lead to issues of overfitting – modeling errors created by using a limited set of data points.  

Here to stay?
Reider added, there is an advantage of quantamental over quant: 
  -  The quantamental model makes a bit of sense and can grow
  -  Whereas using alternative data for pure quant is much more difficult 
  -  In a multi-strategy fund there simply aren’t enough resources to analyze alternative data sets

Ho warned that although people say that this is the new generation of hedge funds, in reality it isn’t. Instead, he believes that massive computing power has instead reduced volatility and the role of sentiment in general. He also claimed to be a sceptic when it comes to the use of machine learning in finance, explaining there is a danger that machine learning leads to huge overfitting. In his view, a lot of machine learning talk is more of a marketing message rather than its practical use in strategies. 

In some sense, the data wasn’t designed to be used for investment purposes, explained Weldon, adding that this creates a major issue. And because there are so many layers, this may give rise to product issues which could lead to regulators seeing a need to get involved. In five years’ time, will the Fed need to have a committee on alternative data? That is a real possibility, he predicts. However, the availability of the data is not going to go away. 

Conclusion:
  -  The data pools will become oceans – they’ll get larger and larger
  -  How it’s deployed in investment strategies will become the key question 

In the current climate of diminishing returns, this debate around the practical uses of alternative data and quantamental investing will continue. It is vital to hold events such as this to ensure that as an industry we are working together for a greater understanding of such new developments in our sector.