Olaf Korn

Model selection in neural networks

In this article, we examine how model selection in neural networks can be guided by statistical procedures such as hypothesis tests, information criteria and cross validation. The application of these methods in neural network models is discussed, paying attention especially to the identification problems encountered. We then propose five specification strategies based on different statistical procedures and compare them in a simulation study. As the results of the study are promising, it is suggested that a statistical analysis should become an integral part of neural network modeling.

Market Depth and Order Size

An acoustic detector which includes at least one emitter unit with a solid tapered profile spike (C) associated with an element (X) for exciting the spike so as to propagate ultrasound waves in an antisymmetrical propagation mode in the spike. The device also emits waves into a surrounding gas. At least one receive unit includes a tapered profile solid spike associated with the detector for receiving the ultrasound waves.

Improving the pricing of options: a neural network approach

In this paper we apply statistical inference techniques to build neural network models which are able to explain the prices of call options written on the German stock index DAX. By testing for the explanatory power of several input variables serving as network inputs, some insight into the pricing process of the option market is obtained. The results indicate that statistical specification strategies lead to parsimonious networks which have a superior out-of-sample performance when compared to the Black/Scholes model. We further validate our results by providing plausible hedge parameters.

Bond portfolio optimization: A risk-return approach

In this article, the authors apply Markowitz’s approach of portfolio selection to government bond portfolios. As a main feature of the analysis, the term structure models is used to estimate expected returns, return variances, and covariances of different bonds. The authors’ empirical study for the German market shows that a small number of risky bonds is sufficient to reach very promising predicted risk-return profiles. If the number of risky bonds in the portfolio is not too large and the term structure model does not contain more than two factors, these predictions are confirmed by the realized risk-return profiles.

Portfolio Optimization Using Forward-Looking Information

We develop a new family of estimators of the covariance matrix that relies solely on forwardlooking information. It uses only current prices of plain-vanilla options. In an out-of-sample study we show that a minimum-variance strategy based on these fully-implied estimators outperforms several benchmark strategies, including various strategies based on historical estimates, index investing, and 1/N investing. The outperformance originates in crisis periods when information ow and information asymmetry are high. Although the historical benchmark strategies improve when more recent data is used, they never outperform fully-implied strategies. Thus, our results suggest that investors are better off relying on forward-looking information.

Volatility in Oilseeds and Vegetable Oils Markets: Drivers and Spillovers

Food price volatility has re-emerged as an important topic of political discussion since the food price crisis of 2007–2008. Different volatility drivers have been identified for different markets in the theoretical and empirical literature. However, there is no comprehensive analysis that considers a large number of potential drivers and investigates their joint effects in a dynamic model of interrelated markets. Our study provides such a volatility analysis for the oilseeds and vegetable oils markets. We use a common GARCH approach and a VAR model to identify volatility drivers and spillover effects. Our results show that exchange rate volatility is very important. However, the hotly debated financialisation of commodity markets is not found to be volatility increasing in our monthly data. Impulse response functions show strong spillover effects. Because many volatility drivers found to be important in other markets have no significant effect in our study, our results suggest that volatility drivers are market specific. This implies that any volatility-reducing policies need to be designed for the market in question.

Which Beta is Best?On the Information Content of Option-Implied Betas

Option-implied betas are a promising alternative to historical beta estimators, because they are inherently forward-looking and can incorporate new information immediately and fully. Recently, different implied beta estimators have been developed in previous literature, but very little is known about their properties and information content. This paper presents a first systematic comparison between six different implied beta estimators, which provides some guidance for applications and identifies directions for further improvements. The main results of the empirical study reveal that betas derived from implied variances are better predictors of realized betas than betas obtained from implied skewness, and that cross-sectional information from all stocks in the market improves beta estimation significantly. We also find that option-implied betas generally have a higher information content in periods of relatively high trading activity in options markets.

Hedging langfristiger Lieferverpflichtungen mit kurzfristigen Futures: möglich oder unmöglich?

SummarySince the crisis of the Metallgesellschaft AG in 1993, their roll over hedging strategy with short-term oil futures has been a subject of controversal debate. The main issues of this controversy are whether a roll over hedge is at all appropriate and what hedge-ratios should be used. In this paper we derive minimum risk hedging strategies based on theoretical pricing models. These models contain either the spot price of oil or both the spot price and the convenience yield as stochastic factors. In the empirical part of the paper we develop a data based simulation model, which is used to generate realistic paths of spot and futures prices. Based on these prices we compare the different strategies in respect to their ability to reduce risk. It turns out that only one of the hedging strategies achieves a risk reduction compared to an unhedged position.

Stock Illiquidity, Option Prices, and Option Returns

We provide evidence of a strong effect of the underlying stocks illiquidity on option prices by showing that the average absolute difference between historical and implied volatility increases with stock illiquidity. This pattern translates into significant excess returns of option trading strategies that are not explained by common risk factors. Simulation results show, however, that our results can be explained by the hedging costs of market makers who are net long in options on some underlyings and net short in options on other underlyings. Our empirical findings are robust with respect to the chosen illiquidity measure, the measure of option expensiveness, and the return period.

Forward-Looking Measures of Higher-Order Dependencies with an Application to Portfolio Selection

This paper provides implied measures of higher-order dependencies between assets. The measures exploit only forward-looking information from the options market and can be used to construct an implied estimator of the covariance, co-skewness, and co-kurtosis matrices of asset returns. We implement the estimator using a sample of US stocks. We show that the higher-order dependencies vary heavily over time and identify which driving them. Furthermore, we run a portfolio selection exercise and show that investors can benefit from the better out-of-sample performance of our estimator compared to various historical benchmark estimators. The benefit is up to seven percent per year.