Research Analysis 05 Pairs Trading Strategy Based On Github

Leo Migdal

-Dec 11, 2025, 11:33 PM

research analysis 05 pairs trading strategy based on github

There was an error while loading. Please reload this page. Pairs trading is a market neutral trading strategy and it belongs to statistical arbitrage. The basic idea is to select two stocks which move similarly, sell the high priced stock and buy the low priced stock where there is a price divergence between the pairs. https://www.quantconnect.com/terminal/processCache?request=embedded_backtest_beb5b38bb307c677d9611dc48bc38db9.html Before using pairs trading, we need to know the cointegration.

Cointegration is a statistical property of time series (that is a series of random variables) Correlation specify the co-movement of return, it is a short-term relationship Cointegration specify co-movement of price, it is a long-term relationship The primary goal in an investment endeavor is the implementation of strategies that minimize the risk while also maximizing the financial gain or return from the said investment. While there have been many popular strategies and techniques developed over the years that point towards the same goal, the 'Pairs-Trading' strategy is one that has been used to great extent in modern hedge-funds,... This strategy, often termed a statistical-arbitrage, relies on monitoring the correlation between a pair of stocks (known to be correlated).

A long position is opened on the stock that rises and a short position is opened on the stock that falls. The underlying assumption in pairs-trading is that pairs of stocks, that have historically shown similarities in their behavior will eventually converge in the long run, even if they diverge in the short term, allowing... In such a strategy, identification of correlated stocks and generation of pairs is of paramount importance. In this project, we employ unsupervised learning techniques that include Density-Based Spatial Cluster of Applications with Noise and K-Means Algorithm. Once, the relevant pairs have been identified, their price relations are extrapolated using supervised learning techniques such as Linear Regression. This overall methodology will help provide insight into the relations between various stocks and facilitate the generation of appropriate trading strategies for them.

The datasets are provided by Wharton Research Data Services (WRDS). We mainly obtained the daily stock files from file from CRSP and quarterly fundamentals from Compustats for our purpose. Initially, our dataset consists of stock price files from 3000 stocks which are constituents of Russell 3000. Those stocks' value and size are large enough to restore the whole market value, representing approximately 95% of the total market shares. We performed this pre-screening process to avoid the 'small-cap' trap in the market. Currently, there are more than 6000 active stocks in the U.S.

Stock Market but most of them are micro-valued. In reality, investors often cautiously avoid investing in those stocks, since trading, even a small number of shares might have unpredictable effects on their stock prices. We should keep this in mind when doing academic research. We set the sample period from 2010-01-01 to 2015-12-31 for training strategies and use sample period 2016-01-01 to 2019-12-31 for backtesting. In our next stage, we want to pre-select eligible stocks that enable us to sail through further steps. First, we removed stocks that were delisted, exchanged, or merged during our sample period since those stocks are no longer tradable.

Next, we removed stocks that have negative prices which will be problematic for further analysis. Stocks that are constantly trading at-low-volume also have to be removed since improper trading executions can largely change their stock prices and altered history. Finally, we remove stocks that have more than half missing prices, so that we have enough available data for imputation. A similar approach was performed on the financial fundamentals of datasets. In the end, there are 1795 eligible stocks for further analysis. In this step, we imputed the missing values in our preprocessed dataset.

We worked with the time series data and the financial ratios separately. We imputed both of them using means, although in a slightly different way. For the time series data of stock prices, missing values were replaced by the mean of all the available stock prices for that stock in the training period. Since the financial ratios individually have different bounds we imputed missing values in the financial ratios dataset with the average of all available data for the particular ratio. Statistical arbitrage of cointegrating currencies with pair trading where the signal for the next day is predicted using LSTM Financial analysis and demonstration of the classic algorithmic trading method, pair trading.

This analysis compares the portfolio's growth with the underlying assets value and volatility over time. Code for Hedging via Opinion-based Pair Trading Strategy A pair-trading algorithm using cointegration, linear regression, and Z-score-based entry/exit rules. The strategy, applied to validated stock pairs, achieved consistent portfolio growth from $24,050 to $25,489.50 over 2 years through trading simulation. Pair trading strategy integrates multiple components, including technical analysis indicators, machine learning models, and risk management techniques. Pairs trading is a form of statistical arbitrage that takes advantage of mean reversion or convergence in the prices of two instruments.

The simplest variation of this strategy involves taking long and short positions simultaneously on a pair of cointegrated instruments. But more generally, you can construct a spread from any linear combination of the two instruments with the approach in this tutorial — even going long simultaneously on two instruments. We’ll make use of the following external libraries: Most existing literature shows the application of this strategy to equities, however the strategy can be applied to any asset class. We’ll demonstrate the pairs trading strategy on the two most liquid crude oil products traded on CME and ICE, the WTI crude oil futures (CL) and Brent crude oil futures (BRN) contracts respectively. Group: Krishna Sardana, Sanjot Singh, Saloni Shah, Sofia Fernandez Chacin, and Monica De Armas

Pairs trading is an algorithmic trading strategy where trading occurs based on the idea that pairs of securities will always revert back to their historical spread. In order to make a profit, investors go long and short during the period when the spread between these two stocks diverges. When the spread returns to its historical mean, at least one of the positions will result in a profit. Trading on pairs of stocks in the same industry is the beneficial because they will react to similar market factors, therefore having similar price variations. Once investors identify a pair of stocks, they verify the prices of the stocks are cointegrated before trading them. Investors can trade using various statistical thresholds that will allow them to figure out if the current spread is near the historical mean spread or not.

We aim to employ Machine Learning to find the ideal pair of stocks and predict their future spread which we can trade upon. Predictive models in the stock market are mainly used to learn about the upcoming fluctuation of different securities. Attaining an accurate projection leads to a potential increase in return for investors. Machine Learning is useful in this context because there’s a large range of methods that have the potential to discover patterns in datasets, leading to important resolutions. The dataset that we chose was the stock price data from the S&P 500 from 2013 to 2018. This includes the open and close prices of the day, the high and low prices of the day, and the volume traded.

However, the column that we were interested in was the close prices as we decided they would be the best data to trade upon for the algorithm. We started by pre-processing the data by taking the raw dataset which had vertically stacked datapoints as shown: This was not the way we wanted the data to be formatted as we wanted to have each row show the close prices of different stocks over the five years. Therefore, we processed the data and were able to put each individual stock in time series order as shown: Quantitative analysis, strategies and backtests Educational notebooks on quantitative finance, algorithmic trading, financial modelling and investment strategy

This repository contains three ways to obtain arbitrage which are Dual Listing, Options and Statistical Arbitrage. These are projects in collaboration with Optiver and have been peer-reviewed by staff members of Optiver. A curated list of awesome algorithmic trading frameworks, libraries, software and resources This project involves using a combination of statistics along with financial thoery to demonstrate a popular trading strategy used in equity markets: Pairs Trading. Notebook released under the Creative Commons Attribution 4.0 License. Pairs trading is a nice example of a strategy based on mathematical analysis.

The principle is as follows. Let's say you have a pair of securities X and Y that have some underlying economic link. An example might be two companies that manufacture the same product, or two companies in one supply chain. We'll start by constructing an artificial example. We model X's daily returns by drawing from a normal distribution. Then we perform a cumulative sum to get the value of X on each day.

Now we generate Y. Remember that Y is supposed to have a deep economic link to X, so the price of Y should vary pretty similarly. We model this by taking X, shifting it up and adding some random noise drawn from a normal distribution. We've constructed an example of two cointegrated series. Cointegration is a "different" form of correlation (very loosely speaking). The spread between two cointegrated timeseries will vary around a mean.

The expected value of the spread over time must converge to the mean for pairs trading work work. Another way to think about this is that cointegrated timeseries might not necessarily follow a similar path to a same destination, but they both end up at this destination. A research-grade quantitative finance project implementing a cointegration-based pairs trading strategy on S&P 500 Financial Sector stocks (2018-2024). Pairs Trading is a market-neutral statistical arbitrage strategy that: After running the pipeline, results will be available in: Two time series are cointegrated if they maintain a stable long-run relationship, even though each series individually may be non-stationary.

Research Analysis 05 Pairs Trading Strategy Based On Github

People Also Search

There Was An Error While Loading. Please Reload This Page.

Cointegration Is A Statistical Property Of Time Series (that Is

A Long Position Is Opened On The Stock That Rises

The Datasets Are Provided By Wharton Research Data Services (WRDS).

Stock Market But Most Of Them Are Micro-valued. In Reality,