| Seminars
2005-2006 October
3, 2005 (Monday)
Designing electronic markets is a challenging task. The Market Engineering approach suggests a structured engineering process that addresses the complexities inherent to the design by dividing the design process into phases and recommending solution techniques to all individual tasks. The talk introduces the Market Engineering approach and the concept of a workbench CAME (computer aided market engineering) intended to support the development of electronic markets. CAME facilitates all design phases of the market engineering process, starting from the conceptual design which specifies the institutional rules of the market to the implementation and automatic tests by extensive simulations. It will be shown how to use the market platform meet2trade as a CAME tool for various real world applications such as introducing and analysing the impact of bundle trading, testing innovative order types in financial markets, or implementing and running a configurable forecasting market for the FIFA Soccer World Championship 2006. October
3, 2005 (Monday) University
of Karlsruhe (TH), Germany Posted Price Offers in Internet Auction Markets Recently,
internet marketplaces like eBay or Yahoo! have extended the
flexibility of the selling mechanisms available on their platforms.
The product features “Buy It Now” or In the presentation, a model of an auction with a posted price offer (APPO) is introduced and the equilibrium strategies of the seller and the bidders in an APPO are derived. It is shown, e.g., that by offering a posted price the seller can raise her expected revenues if bidders are risk averse. The presentation also reports on an experiment which compares the theoretic predictions with the behavior of students in the lab
Oct 19, 2005 Oct
26, 2005
Nov 2, 2005
Nov 9, 2005 The tapered Gaussian semiparametric estimator (GSE) and, more recently, the exact local (ELW) Whittle estimator are two major approaches in estimating the memory parameter d of a potential nonstationary/noninvertible process. Though the ELW estimator has uniform efficiency for all values of d, it pays a much higher computational cost. On the other hand, the tapered GSE has the trend invariant property and computational advantage, but the variance of estimates increases with the order of tapers. We introduce a class of shift invariant tapers and study the efficient properties of the new tapered GSE. These tapers, which are of orders (p,q), are a maximally efficient sub-class of tapers originally proposed by Chen (2001). If the choice of q=n^{kapa/2p}, a taper is also of degree (p,kapa) defined by Dahlhaus (1988). We investigate whether/how the new tapered GSE can reach the same efficiency as the non-tapered GSE for d in the range of (-0.5,0.5), i.e. a limiting distribution of! N(0,1/4). Furthermore, we conduct a simulation study to compare finite sample properties of the new tapered GSE with those of the ELW (Shimotsu & Phillips, 2005), specifically the modified version by Shimotsu (2005), the two-step feasible ELW estimation. With a tapered GSE estimate as the initial value, this modified version of ELW can be applied to a nonstationary process with unknown mean and trend and is asymptotically N(0,1/4), though it does not have the mean or trend invariant property and the same computational efficiency as the tapered GSE. November
14, 2005 (Monday) November 16, 2005 3:15 pm, Refreshments served at 3 pm South Hall 5607F Hyunyoung Choi, UCSB A Bayesian Methodology of Random Intervention Model with Panel Data: Impact Study on Interest Rate Futures Market A random intervention function model is proposed to model panel data with a common input with the assumption that the output time series reacts to the common input in a nonlinear fashion. Assuming a common distribution for the individual effects, the random coefficient function model allows individual subject variation in response to the input series. The baseline of output time series is assumed to be an ARIMA(p, d, q) model with parameters commonly distributed across the time series. The model is estimated in the Bayesian framework using Gibbs sampling and the Metropolis Hasting Algorithm, which also allows handling of missing data within each series. A simulation study is conducted to demonstrate the performance of MCMC estimation.
Any announcement from the Federal Reserve has a huge impact
on the interest rate markets. The press releases from the Federal
Open Market Committee (FOMC) are one of the major inputs to
the market. The liquidity and the versatility of the Eurodollar(ED)
market provide traders the opportunity to hedge their interest
risk through Eurodollar Futures. The ED market is highly correlated
to the US Government short term (T-bill) interest rate market.
To estimate the impact associated with the FOMC announcements,
the random intervention model is used for an empirical study
on the Interest February 8, 2006 3:15 pm, Refreshments served at 3 pm South Hall 5607F Jan Bjornstad, Visiting Professor at UCSB and Director of Research for Statistical Methods, Central Bureau of Statistics, Norway. Likelihood theory for prediction The talk deals with a theoretical foundation for prediction problems in various models, covering usual parametric models as well as Bayesian models and empirical Bayes models. A generalized likelihood function is defined together with a generalized likelihood principle that satisfies the fundamental equivalence theorem with principles of sufficiency and conditionality. The generalized likelihood is the product of prior information of the predictand and the evidential likelihood in the data for the predictand or equivalently the product of the predictive distribution for and the likelihood function of the model parameters February 15, 2006 3:15 pm, Refreshments served at 3 pm South Hall 5607F Kenneth J. Hochberg, Bar-Ilan University, Israel HIERARCHICALLY
STRUCTURED BRANCHING-DIFFUSING SYSTEMS March
8, 2006 Logic Regression was recently introduced as a novel regression method and classification tool. This adaptive methodology is based on new predictors being generated as Boolean combinations from binary covariates, and hence models with high order interactions can be explored. We review the methodology and highlight the differences to related tools such as CART and MARS, show some public health related case studies, and report on some recent developments in the methodology (such as measures of variable importance obtained from MCMC based algorithms) and the software (in particular, the LogicReg R package). March
14, 2006 Obviously, the etiology of disease may explain why heart disease, with its intermediate endpoints such as serum cholesterol, has confirmed links to nutrition while the evidence is mixed with cancer. I will focus instead on a basic question of study design: how do we measure what we eat? Try this out: how many days per year do you eat pizza? I am going to review the accumulating evidence that suggests that with complex, subtle disease such as cancer, with no good intermediate endpoints such as serum cholesterol for heart disease, finding links between disease and nutrient intakes will be the exception rather than the rule, simply because of the way diet is measured. The talk will touch briefly on the recently-completed Women’s Health Initiative Dietary Intervention Study, and describe statistical reasons why this study might have been difficult to interpret even had it achieved statistical significance.
March 15, 2006 SOBEL LECTURE April
5 Proteins, like all machines, are complex, three-dimensional structures in which form intimately defines function. The manufacturing machinery in your cells, however, initially synthesizes proteins as random coil polymers that only then spontaneously, rapidly and efficiently fold into well-defined,, functional structures. Across the universe of simple, single domain proteins, the fastest performs this folding dance a million times more rapidly than the slowest. What accounts for this dramatic range of kinetic behaviors? In this talk I cast an experimentalist's critical eye on recent theories of protein folding kinetics that address this fundamental biophysical issue.
April 12: John Boscardin, UCLA Biostatistics
Biostatisticians are frequently asked to perform inference for
data sets with multivariate repeated or longitudinal measurements.
Investigators typically will ask questions such as "are
measures X and Y correlated" or "when did measures
X and Y exceed clinically important thresholds". Answering
these seemingly simple questions is not entirely straightforward;
we have developed a variety of Bayesian statistical models to
address them. Two settings will be presented: (i) modeling the
dependence of a mixture of continuous, ordinal, and categorical
repeated measures, April
26
It is often of interest in the agricultural seed industry and in other contexts to attempt to determine the ancestry of a particular organism. A probabilistic approach has been proposed by a group at Pioneer Hi-Bred International, Inc. The work presented here extends their approach for estimating the ancestry probability, the probability that an inbred line is an ancestor of a given hybrid, to account for genotyping errors. The effect of such errors on ancestry probability estimates is evaluated through simulation. The simulation study shows that if misclassification is ignored, then ancestry probabilities may be slightly overestimated. The sensitivity of ancestry probability calculations to the assumed genotyping error rate is assessed and likelihood-based approaches to simultaneously inferring the error rate and the ancestry probabilities are considered. (This is joint work with Hongmei Zhang, University of West Florida.)
A review of spectral estimation of non-stationary but harmonizable processes will be given. Given a single realization of the process, periodogram like and consistent estimators are proposed for spectral mass estimation when the spectral support of the process consists of lines. Such a process can arise in signals of a moving source from array data or multipath signals with Doppler stretch from a single receiver. Such processes also include periodically correlated (or cyclostationary) and almost periodically correlated processes as special cases.
May
22 (Monday)
We
study the portfolio selection problem in the presence of the
option to exert costly effort for more income. As expected,
investors who have the flexibility to generate future income
are willing to take more risk. Additionally, we find that portfolio
allocation is not monotonically increasing with time. We also
study the dependence of the portfolio allocation on other parameters
like interest rate, market price of risk, individual's productivity
and employment constraints.
May 31 June 7, 2006 3:15 PM, Refreshments served at 3 PM South Hall 5607F Nan Chen, Columbia University, IEOR A tale of two simulations We discuss Monte Carlo methods for two problems central to the pricing and hedging of derivative securities: (i) calculation of "Greeks" (price sensitivities) and (ii) valuation of American options. Both are based on joint work with Paul Glasserman. Malliavin Greeks without Malliavin calculus: Standard methods for estimating sensitivities differentiate paths or differentiate measures. A recent line of work derives estimators using Malliavin calculus. In implementing a discrete-time approximation of a continuous-time model, one may discretize first and then differentiate or vice-versa. In several important cases the first route produces the same estimators produced by Malliavin calculus, but using only traditional elementary techniques.
Additive and multiplicative duality: Pricing an American option
entails solving an optimal stopping problem. Dual formulations
replace maximization over stopping times with minimization over
martingales. We compare duals based on additive and multiplicative
decompositions of positive supermartingales. We establish an
equivalence in the quality of the bounds achieved by the two
methods, but show that the variance of the multiplicative method
is typically much larger. June
12 (Monday) Abstract: We study the calibration of volatility surfaces in a stochastic volatility environment. Starting with a stochastic volatility model for asset prices, we cast the volatility estimation problem as a variational one and we derive an HJB equation for the volatility surface. We incorporate uncertainty in market prices and we study the asymptotics of the resulting HJB equation in the fast mean-reversion regime. We present numerical solutions from our estimation scheme and find certain parameters of the volatility surface to be both stable in time and stable with respect to our iteration procedure.
June 21, 2006 (Wednesday) Option pricing, Hedging, and efficient Monte Carlo methods Under stochastic volatility models, we propose a new and generic control variates method to efficiently evaluate financial derivatives by means of Monte Carlo simulation. These controls are local martingales and are closely related to some imperfect hedging strategies. Therefore, reduced variances obtained from these Monte Carlo methods are not only measurements of computing efficiency but also represent risks associated to some particular trading strategy in the incomplete market. An asymptotic result is obtained to characterize the variance for rescaled stochastic volatility models. The analysis is done through a combination of perturbation techniques and an averaging effect. Several numerical examples, including some lower and upper American option prices, computed by Monte Carlo and/or Quasi-Monte Carlo methods are presented. September
29, 2006 (Friday) Young Shin Kim (University of Karlsruhe, Germany) In this talk, we introduce a variant of tempered stable distribution named modified tempered stable (MTS) distribution, and apply it to an option price model based on a GARCH asset return process. The GARCH option price model with the MTS innovation allows the description of some stylized phenomena in empirical observation of financial market such as volatility clustering, skewness and heavy tail of the return distribution. We estimate the model parameters with S&P 500 index data, and use the parameters to calculate the S&P 500 option prices with Monte Carlo simulation. Finally, we compare the model prices to the market prices. |
||
Site
Map | UCSB website
| Home Statistics & Applied Probability University of California Santa Barbara, California 93106-3110 (805) 893-2129 South Hall 5607A |