代写 math matlab python statistic COMP0040 – Assignment 2

COMP0040 – Assignment 2
Scaling Laws, Dependency and Causality
1 Overview
This assignment is focused on (i) Scaling and Multiscaling, and (ii) Dependency and Causality. It is part of feedback for the preparation of final report.
2 Data
The data set is the same as for Assignment 1. You can also include other currencies and other market signals to enrich your analysis.
3 Task
3.1 Scaling laws
You are asked to investigate the scaling properties of your time-series. Explore how the statistical properties of financial returns change at different time scales (e.g. from hourly, to daily, to weekly). Look at the scaling laws of different assets and discuss the consequences for portfolio formation.
You are expected to investigate the following points:
1. stationarity;
2. deviations from random walk (e.g., autocorrelations); 3. scaling of the return distribution;
4. tail exponent, Hurst exponent, and their relationship; 5. modeling, forecasting and validation.
3.2 Dependency and causality
The task is: (i) to quantify dependencies between different asset returns, and (ii) to investigate the effects of textual sentiment on asset returns. You are expected to compare linear, non-linear, and information-theoretic measures. Specifically, you should investigate differences between these measures and validate your results (e.g., using permutation tests). Discuss the consequences of your findings for portfolio formation.
1

4 Written report
Please note that this is an open question: the detailed approach, the choice of measures and the structure of the report are left to the students. Hints: Follow the lecture notes of Part II and III, and consider which method, which technique and which test can be used and why. Consider carefully on which data the analysis should be done. Consider if you want to analyze only a subset and why. Briefly describe and justify every assumption or choice. To perform the analyses upload the data on a system (such as Matlab, Python, R, or even Excel). A function for the generalized Hurst exponent in Matlab can be found here: Generalized hurst exponent.
Submit a brief written report (maximum 10 pages including Figures and Tables – do NOT submit your codes) containing the introduction, the justification of the approach, the pre- sentation of the results, the discussion of the results and conclusions. You need to submit to Moodle before the deadline on Sunday 17/03/2019 at 23:55. Cleary write your stu- dent number on the heading of first page. File name format must be: StudentNumber.pdf.
The first page must be the self marking sheet filled with all information. The reports must be brought in the class at the lecture the same Monday (18/03/2019) with all sheets stapled together, and this is should be identical with the one you uploaded to Moodle. Assignments not delivered in time and not delivered in the right format will be marked with zero mark.
5 Class discussion and oral presentation
Your work will be discussed in class in groups of students. Each group will give an oral presentation of the main results and methods at lecture on the following Monday 18/03/2018.
6 Marking
This assignment is worth 17% of the total exam mark. The marking will be based on the following criteria:
1. Justification of the choice of measures, assumptions and approaches; 2. Clarity of presentations and explanations;
3. Validity of the results;
4. Consistency of language and mathematical notation;
5. Scientific soundness; 6. Originality.
2