Introduction
Pairs trading is a statistical arbitrage or mean reversion strategy. This strategy first involves identifying a pair of historically cointegrated assets through hypothesis testing, such as with the augmented Dickey-Fuller test. This is done because if two assets are cointegrated, then their spreads are stationary or integrated of order 0, and thus mean reverting. Then, when a temporary deviation occurs in the spread of these two assets, we simultaneously open both a short position on the asset that’s trading higher and a long position on the asset that’s trading lower. This strategy generates profit when the spread between the two assets reverts back to their historical mean. Theoretically, pairs trading is considered a market-neutral trading strategy, in that it can profit regardless of the overall direction of the market, inefficiencies such as slippage and transaction costs aside. This is because the two positions, one short and one long, serve to hedge against each other.
Goals
The goals of this project are three-fold. First, we will conduct a thorough literature review to understand the statistical concepts underlying pairs trading, such as correlation, cointegration, and hypothesis testing thereof. Next, we will develop an event-driven backtesting library to test different types of pairs trading strategies with real-world data. Finally, we will identify assets with historical cointegration.
Team Members