I have attached the instructions of the project. This is a matlab machine learning project for Machine Learning course. I need a report and presentation for this project as well.

I have attached the instructions of the project. This is a matlab machine learning project for Machine Learning course. I need a report and presentation for this project as well.


Course Project As explained in class, the course project will be an attempt to implement a pair trading strategy (Lecture 8). The strategy follows the classic paper by Avellaneda and Lee (2008) that I uploaded on the BB and S. Ganguli (2014) at https://swetava.files.wordpress.com/2015/08/swetava-ganguli-statistical-arbitrage-of-eigenportfolios.pdf. At the time when the paper was published, however, machine learning was not yet widely used in finance. The strategy involves the following steps: 1. Choose a pair of stocks in your sample of 10 stocks. You may start with the pair that has the highest correlation. 2. Write matlab code that moves a sliding window of a given length M through the time series. Within the time window implement a regression of one time series (F) on the other (S). In the simplest case of a linear least squares regression it will determine the coefficients: is the residual. We evaluate the process: Where and are returns of the two stocks and is the intercept. If and perfectly predicted the relationship, the residual would be zero. The residual is actually the P&L (profit and loss) of the portfolio with $1 invested in stock S and to stock F at day . The integral of this process is: X(k) s the P&L (Profit and Loss) between days 1 and (with daily rebalancing). 3. Use LSTM in matlab to learn forecasting of X(k) in the training window of the first 400 observations. Try to obtain the best training performance by varying the hyperparameters. 4. Predict X(k) out of time in the testing sample of 100 observations. 5. Try to formulate a trading strategy based on the observed behavior of X(k). 6. Repeat with other types of regression instead of linear. We will go through this again in class. I will also be monitoring the progress of groups week by week.