Modelling the evolution of the New York Stock Exchange
This project handles with the 10-min NY Stock Exchange data between 2011 and 2014 and aims at processing the data and publish it for research purposes.
In this project we will study the finance data from New York Stock Exchange (NYSE) collected directly from Yahoo Finance. The data were collected from the website http://finance.yahoo.com/ in time-steps of 10 min, comprehending a portfolio of 1750 companies. Different properties, including volume and prices, are recorded from January 27th 2011 till April 6th 2014, a total of 976 days, corresponding to 17708 data points after filtering out weekends, holidays, after-hours, and nights.
Goal
The aim of the project is twofold:
- Fit a set of possible distribution models for the different finance properties, keeping track of the associated parameter values. Repeating this fit to each 10-min snapshot of the NYSE we will derive a time-series of the parameter values.
- Use these time series together with the original data to train a neural network for predicting the future values of the finance properties. The aim is to provide a computational framework for modelling non-stationary value distributions and establish a comparative analysis between different models and their associated prediction errors.
The aim is to provide a computational framework for modelling non-stationary value distributions and establish a comparative analysis between different models and their associated prediction errors.
Learning outcome
- Data processing and data modelling
Qualifications
- Some knowledge in Machine Learning and programming (e.g. python)
Supervisors
- Pedro Lind
