Oil price volatility forecasts: What do users need to know?

Date and time
Wednesday, 13 March 2019
14.00 - 15.00

Room N410
Kedleston Road, Derby Campus
Kedleston Road
DE22 1GB

The Enterprise Evaluation and Applied Economics Cluster has organised a seminar series for Spring 2019, showcasing the research being undertaken by the Cluster, as part of the Centre for Business Improvement. The series brings together prominent economists to present their latest work in the field, and would be of interest to staff and students across the University.

We're delighted to welcome Professor George Filis, Professor of Financial Economics at Bournemouth University, as our second presenter.


Oil price volatility forecasts: What do users need to know?


Oil price volatility forecasting is of major importance due to the financialisation of the oil market over the last 10 years or so and the fact that the oil market participants (e.g. oil-intensive industries, policy makers, portfolio traders) form decisions based on such forecasts. The current practice focuses on predicting the conditional oil price volatility (using daily, weekly or monthly frequency) or the realized volatility (using intraday day). Even more, we observe that the bulk literature evaluates these forecasts using statistical loss-functions, such as the Mean Squared Error. Nevertheless, we maintain that in order to make informed decisions, oil volatility users need to need to employ loss functions that reflect the purpose of the oil price volatility forecast, i.e. they need to use objective-based evaluation functions. Thus, the aim of this paper is to forecast the WTI oil price implied volatility (OVX), focusing on several intraday realized volatility measures of the WTI oil prices. We are particularly interested on volatility forecasts that can be utilised by oil traders, portfolio and risk managers. Thus, forecasts are evaluated based on the after-trading cost profitability of four common trading decisions, namely (i) trading OVX based on the OVX forecasts, (ii) trading OVX based on the different realized volatility forecasts of the WTI crude oil price, (iii) trading straddles in USO based on OVX forecasts and finally (iv) trading the USO underlying price based on oil price volatility forecasts. We evaluate the after-cost profitability of each forecasting model for 1-trading-day up to 66-trading-days ahead. Our findings show quite clearly that objective-based evaluation functions are indeed useful compared to stand alone statistical loss functions, given that depending for which trading strategy the volatility forecast is used, different models and realized volatility measures are useful. The results remain robust under different market conditions.

Book your place

* Indicates a required field

Fill out my online form.