Finding Alphas, 2nd Edition

6Data and Alpha Design

By Weijia Li

Data plays a central role in alpha design. First, basic data, such as the price and volume of a security, is necessary to run a backtesting simulation. No matter what kind of alpha idea one wants to backtest, this basic information is required to calculate performance statistics like return, Sharpe ratio, and turnover. Without these statistics, we will never know whether an alpha idea is good. Second, data itself can inspire alpha ideas – every alpha idea is associated with some sort of data. By observing the price–volume plots of some stocks, for example, we may find repeating patterns in the history that can be used to make predictions for the future. If we also have access to company earnings data, one idea would be to trade stocks based on fluctuations in earnings.

In this chapter, we will discuss the effective use of data in alpha design. Usually, finding data is the first step in alpha research. After the data is obtained, some sanity checks should be made to verify the data's usability. Then we may start alpha research.

HOW WE FIND DATA FOR ALPHAS

Finding new data is a critical skill for an alpha researcher. We prefer alphas with good performance and low correlation; a new dataset can serve both purposes. Sometimes we can get signals from one set of data, but they may not be strong enough even after we have tried our best to improve them. If we can get another set of data and look at companies from a different angle, we may improve ...