Algorithmic trading and investment
FIN7030
Overview
This module will introduce the modern practices in finance of using algorithms to extract computer age statistical inference. The purpose of this course is not to introduce students to the vast array of machine learning algorithms. The purpose is to introduce the emerging field of Financial Machine Learning (FML). as a complement to traditional financial research techniques.
This course presents machine learning as a non-linear extension of various topics in quantitative economics such as financial econometrics and dynamic programming, with an emphasis on novel algorithmic representations of data, regularization, and techniques for controlling the bias-variance tradeoff leading to improved out-of-sample forecasting.
Context is king in computer age statistical inference, and financial datasets used to solve modern investment problems offer unique challenges which are beyond many plug and play data science algorithms.
Efron and Hastie (2016) explain the challenges of computer age statistical inference as follows:
Broadly speaking, algorithms are what statisticians do while inference says why they do them. An energetic brand of the statistical enterprise has flourished in the new century, data science, emphasizing algorithmic thinking rather than its inferential justification.
The era of “Big Data” has provided a backdrop for the rapid expansion of immense computer-based processing algorithms, for instance, random forest for prediction. The importance of inferential arguments in support of the ML applications has emerged as an exciting (yet underdeveloped) field. This is particularly true for financial research questions where the complexity of the data story^[Or more formally the data generating process which underpins the sample] result in notoriously noise covariance matrices. A small percentage of information these matrices contain is signal, which is systemically suppressed by arbitrage forces. This course will introduce best practice techniques in financial data science which can help illicit economically meaningful signal and answer contemporary financial research questions.
How to get a top grade
I am passionate about student development. I use the latest knowledge transfer science to activate permanent changes in students' understanding. I achieve this through learning by growth rather than memory. This is especially important in maths, where modern nueroscience tells us that everyone has an innate ability to do well in math. In the below video, Professor Jo Boaler explains how to succeed in learning through growth.
Below is the grading system using this course, which is based on the standard postgraduate taught conceptual equivalent grading scheme of the School. To get an above-average mark students must show a maturity in their learning and understand far beyond rote memorising.
| Grade Range | What you need to demonstrate | What moves you up within-grade band |
|---|---|---|
| 80-100 | Thorough and systematic knowledge and understanding of the module content. A clear grasp of the issues involved, with evidence of innovative and the original use of learning resources. Knowledge beyond module content. Clear evidence of independent thought and originality. Methodological rigour. High critical judgement and a confident grasp of complex issues. | Originality of argument |
| 70-79 | Methodological rigour. Originality. Critical judgement. Evidence of use of additional learning resources. | Methodological rigour |
| 60-69 | Very good knowledge and understanding of module content. Well argued answers. Evidence of originality and critical judgement. Sound methodology. Critical judgement and some grasp of complex issues | Extent of use of additional or non-core learning resources |
| 50-59 | Good knowledge and understanding of the module content. Reasonably well-argued. Largely descriptive or narrative in focus. Methodological application is not consistent or thorough. | understanding of the main issues |
| 40-49 | Lacking methodological application. Adequately argued. Basic understanding and knowledge. Gaps or inaccuracies but not damaging. | Relevance of knowledge displayed |
| 0-39 | Little relevance material and/or inaccurate answer or incomplete. Disorganised and irrelevant material and misunderstanding. Minimal or no relevant material. | Strength of argument |
Learning Outcomes
- Understanding of the application of algorithms and machine learning to finance.
- Introducing to using algorithms to research contemporary finance problems.
- Introducing to cloud computing for finance.
- Learn to combine R+python in an agile, durable and credible way.
- Using state-of-the-art cloud computing solutions (Rstudio Server Pro).
- Develop independent problem-solving techniques.
- Learn the properties of algorithms through Monte Carlo simulations:
Fake it before you make it
- Introduction to the use of financial machine learning to * explains* modern phenomenon in finance.
- Understand how credible theory is needed to build successful algorithmic trading and investment strategies
- how can we use ML to build better financial theories?
Self Study
Much of the content for this course is self-contained within the lecture and online canvas notes. Where you find a gap in your background knowledge, you may also wish to consult one of the following texts and the relevant papers referenced in the course plan.
Core Reading
- López de Prado, Marcos. 2020. “Machine Learning for Asset Managers.” In Elements in Quantitative Finance. Cambridge University Press.
Advanced Reading
- ——. 2018. Advances in Financial Machine Learning. John Wiley & Sons.
- Efron, Bradley, and Trevor Hastie. 2016. Computer Age Statistical Inference. Cambridge University Press.
- Dempster, M.A.H., Juho Kanniainen, John Keane, Erik Vynckier. 2018. High-Performance Computing in Finance: Problems, Methods, and Solutions. Cambridge University Press.
- Dixon, Matthew F., Igor Halperin, and Paul Bilokon. 2020. Machine Learning in Finance: From Theory to Practice. Springer International Publishing.
Course plan
| Topic | Week | Learning outcome | Book chapters | Papers to read |
|---|---|---|---|---|
| Why to study financial machine learning? | 1 | The current state-of-play in the quant finance industry. The common fallacy around ML in finance. The ML alternatives to classical approach for scientific theory investigation | Book 1 Chapter 1 | (Easley et al., 2020; Wasserstein et al., 2019; Chen, 2020) |
| High-performance cloud computing in finance | 2 | Financial problems are becoming increasingly computationally expensive. Thus there has been a rapid expansion in HPC solutions in the finance industry. This includes new numerical methods, HPC systems, and cloud-based deployments. This topic will be an overview of the current landscape and an introduction to the Management schools high performance cloud-computing resource | Book 4 | |
| Denoising and detoning | 3 | Covariance matrices in financial are ill-conditioned as a result of a small number of independent observation used to estimate a larger number of parameters. Using ML algorithms based on the Marcenko-Pastur theorem to denoise and denote financial covariance samples. | Book 1 Chapter 2 | |
| Distance metrics | 4 | Financial correlation has critical limitations as a measure of codependence in financial research problems. The topic explores the use of Shannon’s entropy theory using ML algorithms to capture important non-linear features of financial data. | Book 1 Chapter 3 | |
| Optimal clustering | 5 | A common goal in financial research problems is to separate output group entities by maximising intragroup similarities or minimising intergroup similarities. Using the idea of distance students will be introduced ML techniques to optimise the number and composition of clusters in financial problems. | Book 1 Chapter 4 | |
| Financial labels | 6 | In supervised learning, financial researchers need to carefully ponder how they label their data, as labels determine the task that the algorithm is going to learn. All academic studies in finance use fixed-horizon labelling. Return computed on fixed-time labels^[For example daily returns] exhibit substantial heteroskedasticity and dismiss all information in the intermediate returns. Students will be introduced to alternative labelling methods that overcome these limitations, including the triple-barrier, trend-scanning, and meta-labelling methods. | Book 1 Chapter 5 | |
| Explainable machine learning | 7 | Traditional approach to financial research use the classical econometric approach which combines various specification guesses about functional form and explanatory variables and hunts for statistical significance misusing p-values^[This problem is so widespread that the American Statistical Association has discouraged their application as a measure of statistical significance (Wasserstein,2019)]. Students will be introduced to several glass-box ML supervised learning techniques to extract robust inference of financial phenomenon. Topics will include, Mean-Decrease Accuracy, Shapley Values and Accumulated Local Effects | Book 1 Chapter 6 | Apley et al., (2020), Strumbelj and Kononenko (2014), Molnar (2019) |
| Testing set overfitting | 8 & 9 | Understand the dangers of backtesting using historical data and some alternatives to control for the pitfalls. Backtesting using combinatorically purged cross-validation (CPCV), synthetic data. Learn about important backtesting statistics | Book 1 Chapter 8, Book 2 Chapters 12-15 | Prado(2019) |
| Round up | 10 | Review of course material and exam tips |
Assessment
Critical essay (30%)
- 30% Critical assessment essay on the following statement (1500 word limit excluding references):
Financial machine learning are black-box prediction engines and offer little benefit to researching phenomenon in finance beyond traditional econometric techniques
Section in the essay
- Introduction: Set the statement in the context of modern financial research practices
- Critical literature: Use high-quality research to illustrate the pros and cons of the statement.
- Experimental evidence^[optional section]: You are free to use experimental evidence using simulated data and code.
- Your conclusions: summarise and state your critical assessment of the statement using the scientific evidence that you have present in the previous sections. Students should use an RMarkdown report to produce an HTML or pdf essay. Due end of week 5 submitted electronically via TurnitinUK. The lecturer revise the right to orally exam students after each assessment if he suspects foul-play.
Computer-based practical test (70%)
The assessment will be a mixture of computational, theoretical and inferential questions based on all the course material and readings. The assessment will be an open-book test, run on the Schools high-performance cloud computing server. Exams will be proctored using Rstudio Server user logs to ensure fairness. The lecturer revise the right to orally exam students after each assessment if he suspects foul-play.
Assessment protocols and learning tips
In both cases, it is important to learn how to read and critique academic papers. This is a learning process which requires practice. This link provide an excellent guide.
References
Apley, Daniel W., and Jingyu Zhu. 2020. “Visualizing the Effects of Predictor Variables in Black Box Supervised Learning Models.” Journal of the RoyalStatistical Society. Series B, Statistical Methodology 82 (4): 1059–86.
Athey, Susan. 2017. “Beyond Prediction: Using Big Data for Policy Problems.” Science 355 (6324): 483–85.
Chen, Andrew Y. (2019). “The Limits of p-Hacking: a Thought Experiment,” Finance and Economics Discussion Series 2019-016. Washington: Board of Governors of the Federal Reserve System, https://doi.org/10.17016/FEDS.2019.016.
Easley, David, Marcos López de Prado, Maureen O’Hara, and Zhibai Zhang. 2020. “Microstructure in the Machine Age.” The Review of Financial Studies, July. https://doi.org/10.1093/rfs/hhaa078.
Easley, David, Marcos M. López de Prado, and Maureen O’Hara. 2012. “Flow Toxicity and Liquidity in a High-Frequency World.” The Review of Financial Studies 25 (5): 1457–93.
Efron, B., and R. Tibshirani. 1991. “Statistical Data Analysis in the Computer Age.” Science 253 (5018): 390–95.
Efron, Bradley, and Trevor Hastie. 2016. Computer Age Statistical Inference. Cambridge University Press.
Jacquier, Eric, and Nicholas Polson. 2011. “Bayesian Methods In Finance.” In The Oxford Handbook of Bayesian Econometrics, edited by John Geweke, Gary Koop, and Herman Van Dijk. Oxford University Press.
López de Prado, Marcos. 2018. Advances in Financial Machine Learning. John Wiley & Sons.
——. 2019. “A Data Science Solution to the Multiple-Testing Crisis in Financial Research.” The Journal of Financial Data Science, February. https://doi.org/10.3905/jfds.2019.1.099.
——. 2020. “Machine Learning for Asset Managers.” In Elements in Quantitative Finance. Cambridge University Press.
Molnar, C. (2019): “Interpretable Machine Learning: A Guide for Making Black-Box Models Explainable.” Available at https://christophm.github.io/interpretable-ml-book/
Štrumbelj, Erik, and Igor Kononenko. 2014. “Explaining Prediction Models and Individual Predictions with Feature Contributions.” Knowledge and Information Systems 41 (3): 647–65.
Wasserstein, Ronald L., Allen L. Schirm, and Nicole A. Lazar. 2019. “Moving to a World Beyond ‘p < 0.05.’” The American Statistician 73 (sup1): 1–19.
Athey, S., & Imbens, G. W. (2019). Machine Learning Methods That Economists Should Know about. Annual Review of Economics, 11, 685–725. https://doi.org/10.1146/annurev-economics-080217-053433
