Big Data and Machine Learning in Quantitative Investment

CHAPTER 4Implementing Alternative Data in an Investment Process

Vinesh Jha

4.1 INTRODUCTION

In August 2007 there was a wakeup call in systematic investing when many quants across the Street suffered their worst losses – before or since – over a three‐day period that has been called the ‘Quant Quake’. The event wasn't widely reported outside of the quant world, but it was a worldview‐changing week for portfolio managers who traded through it. In a sense, the search for alternative data sources started during those days.

In this chapter, we look at this foundational event and how it motivated the search for alternative datasets, the degree to which alternative data has in fact been adopted and explanations for why adoption has been gradual, and some prescriptions for fund managers to adopt alternative data more widely. We then examine some important issues with alternative data, including data quality and quantity; we examine how alternative data can realistically help a traditional quantitative or fundamental process; and we look at techniques for finding alpha in alternative datasets. Finally, we provide four examples of alternative data along with backtest results.

4.2 THE QUAKE: MOTIVATING THE SEARCH FOR ALTERNATIVE DATA

After poor but not hugely unusual performance in July 2007, many quantitative strategies experienced dramatic losses – 12 standard deviation events or more by some accounts – over the three consecutive days of 7, 8 and 9 August. In the normally highly risk‐controlled ...