¹û¶³´«Ã½Â鶹Éç

Skip to main content

Stock price and volatility prediction using news analytics

Completed

The aim of this project was to enhance the predictive ability of existing financial time series models by incorporating information from news data.

The idea fits into the 'explainable AI' paradigm, where we combine a mathematical model with a clear economic interpretation for financial time series data with an intuitively interpretable, semi-parametric map for information extracted from news analytics. The project was carried out in collaboration with , which provided doctoral student funding as well as financial news data access.

The following research contributions were achieved:

1. A new static function was proposed for modelling the impact of news sentiment on asset price volatility. This captures the essential features of the impact of news on volatility in a transparent fashion while being easy to calibrate.​

2. This function was used to enhance the predictive ability of GARCH(1,1) model, with stock-specific news was used as input. Exhaustive empirical testing was carried out to compare the modified GARCH model with the traditional GARCH model as well as exponential GARCH model, with 9 datasets each for 12 different stocks from two different stock markets. The ability to predict one day ahead realized volatility of all these models was compared using 2 different error measures. The results convincingly demonstrated that modelling the impact of news data on volatility using our method significantly improves volatility prediction.

3. The same functional form was used to modify volatility structure in a Kalman filter based, vector valued time series model for crude oil. This model uses the logarithm of crude oil futures prices daily as measurements and estimates the underlying crude price as a latent variable. For this application, global macroeconomic news data was used as exogenous input impacting volatility. In this case as well, it was found that using news data to modify volatility function improves the predictive ability of the model, when measured for 12 different futures prices and using two different error measures.

Publications

1. Zryan Sadik, Paresh Date and Gautam Mitra, News augmented GARCH(1,1) model for volatility prediction, IMA Journal of Management Mathematics, Volume 30, pages 165-185, 2019.

2. Zryan Sadik, Paresh Date and Gautam Mitra, Forecasting crude oil futures prices using global macroeconomic news sentiment, IMA Journal of Management Mathematics, volume 31, pages 191-215, 2020. This paper won the


Meet the Principal Investigator(s) for the project

Dr Paresh Date
Dr Paresh Date - I am an electrical engineer by background, with expertise spanning nonlinear state estimation (with applications in tracking and in finance), financial portfolio optimization and financial derivative pricing. My book 'Nonlinear estimation: methods and applications with deterministic sample points' was published by Taylor and Francis in 2019. Every home should have a copy.

Related Research Group(s)

maths

Financial Mathematics and Operational Research - Developing advanced mathematical, probabilistic and optimisation approaches to problems in financial mathematics and operational research.


Partnering with confidence

Organisations interested in our research can partner with us with confidence backed by an external and independent benchmark: The Knowledge Exchange Framework. Read more.


Project last modified 14/07/2021