Finding Alphas, 2nd Edition

10Controlling Biases

By Anand Iyer and and Aditya Prakash

INTRODUCTION

Investment biases have been well studied, as is clear from both the academic literature and the field of behavioral finance, which is dedicated to understanding them. Although a quantitative investment process offers investors a means of managing and arbitraging such biases, many quantitative portfolio managers wind up introducing bias, systematically or otherwise. This chapter is a practitioner's reflection on how to control bias and is aimed at quantitative portfolio managers and researchers.

We start by identifying two categories of bias, then explore various types of bias that exist within these categories. We conclude with some practical suggestions for quantitative practitioners and firms.

CATEGORIES OF BIAS

We broadly categorize bias as systematic or behavioral. Investors introduce systematic bias by inadvertently coding it into their quantitative processes. By contrast, investors introduce behavioral bias by making ad hoc decisions rooted in their own human behavior. Over a period of time, both systematic and behavioral bias yield suboptimal investment outcomes.

SYSTEMATIC BIASES

There are two important sources of systematic bias: look-ahead bias and data mining.

Look-Ahead Bias

In a simulation or backtest, when a signal or investment strategy at a given point in time uses data from a future point that would not have been known or available, it introduces look-ahead bias. This often makes ...