CF969-7-SP-CO Big Data For Computational Finance Assignment

Please refer to the Student’s handbook on the School’s Policy on Plagiarism and Late Submission.

All deliverables below must be uploaded on FASER by the deadline as independent items, i.e., not bundled together in a zip file.

Part I (55%): Report on Machine Learning in Finance

You are asked to write a report on one (1) recent research paper in applications of machine learning for computational finance from the following list:

• “FinRL: Deep Reinforcement Learning Framework to Automate Trading in Quantitative Finance” by X.-Y.Liu et al. Available at: https://arxiv.org/abs/2111.09395
• “Trading with the Momentum Transformer: An Intelligent and Interpretable Architecture” by K. Wood, S. Giegerich, S. Roberts, and S. Zohren. Available at: https://arxiv.org/abs/2112.08534
• “Event prediction within directional change framework using a CNN-LSTM model” by A. Rostamian and J. O’Hara. Available at: https://repository.essex.ac.uk/33313/1/Published_Version.pdf
• “Ascertaining price formation in cryptocurrency markets with Deep Learning” by F. Fang, W. Chung, C. Ventre, M. Basios, Leslie Kanthan, L. Lid, and F. Wu. Available at https://arxiv.org/abs/2003.00803
• “The Efficient Hedging Frontier with Deep Neural Networks” by Z. Gong, C. Ventre, and J. O’Hara. Available at: https://arxiv.org/pdf/2104.05280.pdf
• “ROLAND: Graph Learning Framework for Dynamic Graphs” by J. You, T. Du, and J. Leskovec. Available at: https://arxiv.org/abs/2208.07239
• “Learning Mutual Fund Categorization using Natural Language Processing” by D. Vamvourellis, M.A. Toth, D. Desai, D. Mehta, and S. Pasquali. Available at: https://arxiv.org/pdf/2207.04959.pdf
• “Multivariate Realized Volatility Forecasting with Graph Neural Network” by Q. Chen and C.-Y. Robert. Available at: https://arxiv.org/abs/2112.09015
• “Visual time series forecasting: an image-driven approach” by S. Sood, Z. Zeng, N. Cohen, T. Balch, and M. Veloso. Available at: https://arxiv.org/abs/2011.09052
• “Explainable deep reinforcement learning for portfolio management: an empirical approach” by M. Guan and X.-Y. Liu. Available at: https://arxiv.org/abs/2111.03995

A report of at most 1000 words must be written. The report should summarise and evaluate the article. A good report should address satisfactorily the following questions, namely:

• What is the paper about (i.e., what is the topic)?
• How do the authors approach the problem? I.e., what is the method they use?
• What are strong points in the paper, in your view? Provide arguments.
• What are weak points in the paper, in your view? Provide arguments.

The report will be assessed attending factors such as its contents and the connection to the key questions outlined above, presentation, organisation, clarity, soundness of arguments, etc.

Deliverable for Part I: The report as a pdf file

Part II (45%): Optimization and Machine Learning in Finance – Software

Part IIA (20% of the total mark)

Consider a scenario where an investor has £20,000 to invest in a combination of the following:

• Stock XYZ sells today at £15 per share.
• A European call option to buy 100 shares of stock XYZ at £10 per share exactly six months from today sells for £1,000.
• In addition, a 6-month riskless zero-coupon bond with £50 face value sells for £40.

The investor has decided to limit the number of call options that they buy to at most 50. The investor considers three scenarios for the price of stock XYZ six months from today: the price will be the same as today, the price will go up to £35, or drop to £7. The investor’s best estimate is that each of these scenarios is equally likely.

(13% of the total mark) Formulate and solve a linear program to determine the portfolio of stocks, bonds, and options that maximises expected profit.

(7% of the total mark) Suppose that the investor wants a profit of at least £1,000 in any of the three scenarios for the price of XYZ six months from today. Formulate and solve a linear program that will maximise the investor’s expected profit under this additional constraint.

You can use any solver (e.g., Excel solver, Gurobi, gurobipy, linprog) for solving the linear programs.

Deliverables for Part IIA:

• A document presenting the model formulation, explaining the reasoning behind it, as well as what is the answer and its interpretation. It should be either a pdf file or a Jupyter notebook file.
• The source code. E.g., an Excel file in case you are using the Excel solver, a Jupyter notebook file in case you are using linprog or gurobipy, the Gurobi file in case you are using Gurobi.Part IIB (25% of the total mark)
  Consider the .csv file available on CF969-7-SP Moodle page at the “Assessment Information” tab. It contains 1700 observations of 26 financial and accounting metrics for a set of firms in several different industries. For each observation, the last column denotes the rating according to Moody’s, while the second-to-last column denotes whether the assets are of investment grade or not; ratings in the set {Aaa, Aa1, Aa2, Aa3, A1, A2, A3, Baa1, Baa2, Baa3} are in an investment grade.

You are asked to implement

• a Convolutional Neural Networks based approach to classify the firm’s rating into one of the rating categories and predict if it is in an investment grade
• an LSTM-based approach to classify the firm’s rating into one of the rating categories and predict if it is in an investment grade and discuss how and why were the parameters and the NN architecture selected at each model and also what the results demonstrate for the effectiveness and suitability of each approach on this problem.

To do so, you should split the dataset in a training set and a test set in a 80%:20% ratio. Any language/software we covered during the modules is acceptable; you are also welcome to use a different one that you might have learnt independently or in some other module. Your submission will be assessed attending factors such as contents, clarity, explanations, etc. of the Jupyter-like notebook and correctness, techniques and style of the software.

Deliverables for Part IIB:

• The source code (and any executables, if applicable) of your software in Part IIB
• A document presenting and discussing the selection of parameters as well as the results. It should be either a pdf file or a Jupyter notebook file.

Are you struggling to keep up with the demands of your academic journey? Don't worry, we've got your back! Exam Question Bank is your trusted partner in achieving academic excellence for all kind of technical and non-technical subjects.

Our comprehensive range of academic services is designed to cater to students at every level. Whether you're a high school student, a college undergraduate, or pursuing advanced studies, we have the expertise and resources to support you.

To connect with expert and ask your query click here Exam Question Bank