2211 07080 Designing Efficient Pair-Trading Strategies Using Cointegration for the Indian Stock Market

Statistically, correlation is measured from -1.00 (where one asset goes up and the other tends to go down) to +1.00 (highly correlated pairs where the two are expected to move in the same direction). Let’s take a deeper look into the taxonomy of pairs trading strategies. Time was exactly what was needed for the ice to melt and give pairs trading its long-awaited new beginning.

• The input of backtesting is the z-score history generated in the ‘trading strategy’ part and the price history.
• We offer spread bets and CFDs on over 330 forex pairs, as well as over 8500 stocks, ETFs and other financial assets that can be used as hedging tools.
• In the case of EWA and ENZL, the two assets are not priced the same.
• Not all of them are satisfying and, rather, some even would suffer significant losses over the testing period.
• Second, while historical trends can be accurate, past prices are not always indicative of future trends.

There can be many ways of defining take profits depending on your risk appetite and backtesting results. Running 1,000 simulations of random trades where each simulation starts with $100,000 and trades for four years gives us the following risk profile. Brokerage services are provided by Alpaca Securities LLC (alpaca.markets), member FINRA/SIPC.

An Example Using Options

Multiple comparisons bias is simply the fact that there is an increased chance to incorrectly generate a significant p-value when many tests are run, because we are running a lot of tests. If 100 tests are run on random data, we should expect to see 5 p-values below 0.05. If you are comparing n securities for co-integration, you will perform n(n-1)/2 comparisons, and you should expect to see many incorrectly significant p-values, which will increase as you increase. To avoid this, pick a small number of pairs you have reason to suspect might be cointegrated and test each individually. During the pair selection process, various distance metrics like pearson’s correlation, distance correlation, angular distance, and so on, are leveraged to identify the co-moving securities. In the trading period, simple nonparametric threshold rules are used to trigger the trading signals.

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 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.

Step 5: Model Selection

Having such a distribution ~ N(0, 1) is very useful for creating threshold levels. The monte carlo simulation shows an average drawdown of 20% if we risk 4% per trade. Again, we have no losing year and the winning percentage as well as the average payout ratio are both consistent across the complete validation set. That means we can https://bigbostrade.com/ use our full account size on our long leg and rely on leverage to fund the short leg. This way, if the spread goes against us by 2%, we can close the trade and the loss will equate to 2% of our account size. If we only wanted to risk 1% of our account size, then we would have allocated half our account balance to the long leg.

By default, the Alpaca API will return today’s candle even though it is not closed. We are only interested in the closing prices and want to be able to compare https://day-trading.info/ the closing prices between EWA and ENZL. We can use the concat() function from the Pandas library to create a new DataFrame with only this data.

What is pairs trading?

To be able to identify these threshold levels, a statistical construct called z-score is widely used in pairs trading. The main difference from the Dickey Fuller Test is that the Augmented Dickey Fuller test can also be applied to a large sized set of time series models. The large https://forexhistory.info/ sized time series models can be more complicated and hence, the DF test was modified into the ADF Test. Given a normal distribution of raw data points, the z-score is calculated so that the new distribution is a normal distribution with a mean of 0 and a standard deviation of 1.

• Here we check to see our running PnL and exit the trade if either the stop loss or take profit gets hit.
• Meanwhile, corn futures and soybean meal futures compete as substitutes to nourish cattle.
• If they do not form a trend, that means the spread moves around 0 randomly and is stationary.
• In this paper, we consider optimal pairs trading strategies in terms of static optimality and dynamic optimality under mean–variance criterion.

We scan through a list of securities and test for cointegration between all pairs. It returns a cointegration test score matrix, a p-value matrix, and any pairs for which the p-value was less than 0.05. This method is prone to multiple comparison bias and in practice the securities should be subject to a second verification step. A good example of this is our informative article on The Correct Vectorized Backtest Methodology for Pairs Trading. In the article, we highlight common pitfalls and how to correctly generate an equity curve / PnL plot. The mutual mispricing between the two assets is represented by the value of the spread.

Code, Data and Media Associated with this Article

We started by downloading historical stock data for 10 different companies and identifying cointegrated pairs of stocks. In this article, we will take you on a journey to discover how to build a pair trading strategy in Python. We’ll start by using the yfinance library to download stock price data and the statsmodels library to perform cointegration analysis and linear regression.

This price ratio is sometimes called “relative performance” (not to be confused with the relative strength index, something completely different). The center white line represents the mean price ratio over the past two years. The yellow and red lines represent one and two standard deviations from the mean ratio, respectively. The risk of loss on a short sale is potentially unlimited because there is no limit to the price increase of a security. There is no guarantee the brokerage firm can continue to maintain a short position for an unlimited time period. Your position may be closed out by the firm without regard to your profit or loss.