Tid: 8 juni 2018 kl 17.00-17.35.

Föredragshållare: Sofie Eklund och Randa Estaifo

Title: Modeling implied correlation matrices using option prices

Abstract: In the process of calculating a fair value it is preferable to price the asset from observable market data. Some assets are valued using variables which can not be directly observed in the market but are instead implied from observable market data. One such variable is the correlation between assets. The purpose of this thesis is to model correlations between stocks based on observable market data. Three different approaches are used to construct implied correlation matrices on OMXS30. All matrices are constructed using implied volatilities from the option market. The methods are then compared in order to determine which method that generates the most reliable implied correlation matrix. This is done by looking at deviations from counterparty prices on basket options. The used basket options have two different types of underlying autocallable products; Phoenix Autocall and Autocall Uncapped. It was found that the method with an equicorrelation matrix had the smallest deviations from the counterparty price in a majority of the tested cases. Another result was that the implied correlation matrices performed better on the basket options with Autocall Uncapped than Phoenix Autocall as underlying. An interesting topic for further research is to examine other markets but also to study the methods when more than one market is considered.

The full report (pdf)

Till seminarielistan
To the list of seminars