The open archive for STFC research publications

Full Record Details

Persistent URL http://purl.org/net/epubs/work/34490
Record Status Checked
Record Id 34490
Title Confidence and Prediction in Generalised Non Linear Models: An application to Option Pricing
Abstract There has been a considerable research effort to improve upon the Black-Scholes option pricing model. However, various authors [see e.g. Lajbcygier (1999a)] find generalisations of Black-Scholes and other modern parametric option pricing models, use implausible, inconsistent, implied parameters, and do not out-perform simpler approaches. Non-parametric and computational methods provide an alternative approach to option pricing. These include data intensive model-free approaches, that are often generalisations of better known non-linear regression techniques. This paper describes a robust method for determining prediction intervals for a broad class of such generalised non linear regression techniques. The method is demonstrated by application to a standard synthetic example. It is then applied to obtain prediction intervals for pricing options, using data from LIFFE for the 'ESX' European style FTSE 100 index options. The method uses standard regression procedures to determine local error bars. It is appropriate where errors have heteroskedastistic disturbances.
Organisation CCLRC
Keywords Engineering , Confidence Interval , Financial Options , Prediction Interval , Pricing
Funding Information
Related Research Object(s):
Licence Information:
Language English (EN)
Type Details URI(s) Local file(s) Year
Report Money Macro and Finance Research Group : Advances in Econometrics and Finance (MMF), London, UK, CICM Discussion Paper 03-6, Editor: N Sarantis. MMFPaperforIntCapMrkts.DOC