Finding Alphas, 2nd Edition

5How to Develop an Alpha: A Case Study

By Pankaj Bakliwal and and Hongzhi Chen

In this chapter, we explain how to design an alpha, the logic behind an alpha, how to convert an alpha idea into a mathematical predictive formula by using appropriate information, and how to improve the idea. We will also introduce some important concepts on evaluating the performance of an alpha.

Before we talk more about alpha development and design, let's study a simple example to get a better understanding of what an alpha looks like.

Let's say we have $1 million in capital and want to invest continuously in a portfolio consisting of two stocks: Alphabet (GOOG) and Apple (AAPL). We need to know how to allocate our capital between these two stocks. If we do a daily rebalancing of our portfolio, we need to predict the next few days' return of each stock. How do we do this?

There are a lot of things that can affect the stocks' prices, such as trader behavior, price trends, news, fundamental corporate change, and a change in holdings by big institutions or corporate insiders – officers, directors, or shareholders with more than 10% of a class of the company's registered equity securities. To make things simple, we can deconstruct the prediction process into two steps: first, we predict the stock returns of each instrument, using a single factor like news or price trends; second, we aggregate all the different predictions.

Let's try to develop an alpha using recent price trends, employing available ...