Here is an overview of my research background and interests.
Currently I am researching with Qiwei Yao some nonparametric and semiparametric methods for dimension reduction in the context of conditional probability density estimation. Our goal is to find effective ways to relieve the problem of "curse of dimensionality". Developing an efficient and robust dimension reduction model may lead to further research into financial application of conditional probabilities for highly-dimensional data, e.g., estimation of conditional quantiles or higher moment, time series forecasting, asset pricing, risk management, among others.
During my (second) Master studies in Econometrics and Mathematical Economics I had the opportunity to delve deeper into the field of econometrics and financial economics. My final project was about modelling and testing an Intertemporal CAPM-based model that allows for "momentum trading" and other market effects. Somewhat not surprisingly, my empirical results suggest that the CAPM model may be too limited and that relation between expected returns and risk is highly nonlinear. At this stage I seek to continue exploring some more realistic asset-pricing models and flexible financial econometrics models, and I believe there is still much ground for further research and collaboration between different disciplines to expand our understanding in the field.
In my master thesis in Tel-Aviv University I worked with Offer Lieberman and Issac Meilijson on parametric inference for stationary Gaussian time-series with long-memory. Many economics time-series and the volatility of some financial markets seem to exhibit the property of long-memory, making the research on the topic particularly relevant to econometrics. There is also ample evidence that Stationary time series with long memory occur in other fields as diverse as hydrology, astronomy, biology, computer networks, chemistry, agriculture and geophysics. The goal of my research was to develop asymptotic theory for Maximum Likelihood Estimation of the the model parameters (e.g. the memory parameter, the ARMA parameters and the variance) without a priori knowledge of the true memory type (long, short or intermediate). Based on my thesis I have submitted (joint with Offer Lieberman and Judith Rousseau) a paper for publication, and I intend to continue working in this area.
Studying in Tel-Aviv University, I took several courses on Statistical Learning, Data-Mining and Neural Computation. Later on, working in Segmarketing, I had the opportunity to apply some of these methods to data-bases of large retail chains with the aim of finding patterns and predicting consumers' behaviours. I find this field to be most fascinating for modelling financial data, and it seems to me that this area has not yet received the academic attention it deserves from econometricians and financial mathematicians.