Advanced Pairs Trading Lecture Videos
Join the Reading Group and Community: Stay up to date with the latest developments in Financial Machine Learning!
LEARN MORE ABOUT PAIRS TRADING STRATEGIES WITH “THE DEFINITIVE GUIDE TO PAIRS TRADING”
ArbitrageLab is a python library filled with algorithms from the best academic journals and graduate-level textbooks, which focuses on the branch of statistical arbitrage known as pairs trading.
This playlist is a series of lecture videos that explore advanced topics and highlight how your team can compete with the world’s best hedge funds!
The Distance Approach
From at least four distinct branches of research in the pairs trading literature, the distance approach is the most cited. The popularity of the approach is mainly attributed to its simplicity. The downside, however, is that the dependencies found by this approach may be spurious. We will cover the upsides of this approach and possible improvements.
Measures of Codependance
We explore various measures of codependence and how these algorithms are used in pairs trading (distance approach), clustering, portfolio optimization, and risk management.
The presentation is largely based on the lecture notes of Prof. Marcos Lopez de Prado at Cornell University.
This functionality is also available via the MlFinLab python library.
The Cointegration Approach and Min Profit Optimization
Cointegration is one of the most important statistical arbitrage strategies for pairs and multi-asset trading.
In this talk, we cover the basic concepts of cointegration, the simulation of cointegrated pairs using a stationary AR(1) process, the application of mean first-passage time of an AR(1) process to optimize cointegrated pairs trading boundaries and frequency, and the numerical algorithm to generate the trading signal.
The Principal Component Analysis (PCA) Approach
In this video, we’ll go over the approach proposed in the paper “Statistical arbitrage in the US equities market”(2010) by Avellaneda, M. and Lee, J.-H. This strategy produced an annualized Sharpe ratio of 1.44 over the period from 1997 to 2007. We will briefly cover the theory behind the trading strategy, show its use examples with the ArbitrageLab package, discuss upsides, downsides, and variations of the strategy.
Machine Learning for Pairs Selection
Learn how machine learning is used to find viable securities to trade in a pairs trading setting.
Classical statistical arbitrage research has grown quite generously in the past few years. Yet attempts to solutions using machine learning have been far and few between; this presentation will cover some of the latest approaches, key takeaways, and tips to avoid common pitfalls.
Based on Sarmento, S.M. and Horta, N., 2020. Enhancing a Pairs Trading strategy with the application of Machine Learning. Expert Systems with Applications, 158, p.113490.
Literature Review: Machine Learning to Model Spreads
Modeling financial time series is a notoriously hard problem. Fortunately, because of the hedged out / dampened factor exposures of a pairs trading strategy, the spreads may be easier to model than standard closing prices.
This video reviews the academic literature around machine learning for modeling spreads and particularly the work of Dr. Christian L Dunis and co-authors.
Sparse Mean Reversion Portfolio Selection
Assets that exhibit significant mean-reversion are difficult to find in efficient markets. As a result, investors focus on creating long-short asset baskets to form a mean-reverting portfolio whose aggregate value shows profitable mean-reversion.
This video demonstrates three different approaches to constructing mean-reverting multi-asset portfolios that require trading as few assets as possible. Such sparse portfolios have shown significant advantages in lowering transaction costs, improving P&L interpretability, and capturing meaningful statistical arbitrage opportunities.
Introduction to the Copula Approach
The concept of copula has been widely used in risk management and CDO pricing since the 90s. However, applications for capturing pairs trading arbitrage is relatively novel.
For this talk, we will go through the basic concepts of copula for a pairs trading application, underlying assumptions, trading signal generation, and some common fallacies. We will cover two commonly adopted approaches: using prices series and returns series.
Variations on the Copula Based Mispricing Index Strategy
This is the 2nd part of the copula-based pairs trading strategy. We dive into what the mispricing index is, its pros and cons, what can be changed for better performance. Then we analyze and compare a few other simple variants popular in the literature for this approach and mention the potential issues to address.
Introduction to the Vine Copula Trading Strategy
Vine copula is used for modeling the dependency structure of multiple random variables. Now we venture beyond pairs trading: We will use this advanced tool for statistical arbitrage over a cohort of stocks. In the presentation, we will spend most of our time understanding the vine copula concept, and the workflow used to generate trading signals.
Optimal Trading Rules for Pairs Trading
Presentation of the Ornstein Uhlenbeck submodule of ArbitrageLab, which allows the user both to create an optimal mean-reverting portfolio and to find the optimal timing of trades using the properties of the Ornstein-Uhlenbeck process.
The module is based on the work of Professor. Tim Leung and Xin Lee: “Optimal Mean reversion Trading: Mathematical Analysis and Practical Applications”.
