A5060426633- Stochastic processes and financial applications
- Financial Risk and Volatility Modeling
- Financial Markets and Investment Strategies
- Credit Risk and Financial Regulations
- Monetary Policy and Economic Impact
- Neural Networks and Applications
- Banking stability, regulation, efficiency
- Statistical Methods and Inference
- Complex Systems and Time Series Analysis
- Market Dynamics and Volatility
- Bayesian Methods and Mixture Models
- Stochastic Gradient Optimization Techniques
- Financial Distress and Bankruptcy Prediction
- Model Reduction and Neural Networks
- Insurance, Mortality, Demography, Risk Management
- Target Tracking and Data Fusion in Sensor Networks
- Advanced Memory and Neural Computing
- Housing Market and Economics
- Stock Market Forecasting Methods
- Economic theories and models
- Gaussian Processes and Bayesian Inference
- Advanced Neural Network Applications
- Capital Investment and Risk Analysis
- Control Systems and Identification
- Global Financial Crisis and Policies
École Supérieure des Sciences Économiques et Commerciales
2015-2024
CY Cergy Paris Université
2013-2024
Pázmány Péter Catholic University
2020-2023
The paper proposes a self-exciting asset pricing model that takes into account co-jumps between prices and volatility jump clustering. We employ Bayesian learning approach to implement real-time sequential analysis. find evidence of clustering since the 1987 market crash, its importance becomes more obvious at onset 2008 global financial crisis. also affects tail behaviors return distributions has important implications for risk management, forecasting, option pricing.
We propose a density-tempered marginalized sequential Monte Carlo (SMC) sampler, new class of samplers for full Bayesian inference general state-space models. The dynamic states are approximately out using particle filter, and the parameters sampled via sampler over bridge between prior posterior. Our approach delivers exact draws from joint posterior latent any given number state particles is thus easily parallelizable in implementation. also build into proposed method device that can...
Psoriasis is a chronic inflammatory disease with unmet medical needs. To clarify potential therapeutic targets, different animal models have been developed. In the current study, imiquimod-induced psoriasiform dermatitis was used for monitoring changes in skin thickness, transepidermal water loss, body weight, blood perfusion and drug permeability topical cream formulation of caffeine, both wild type knock out mice. Morphological characterization control diseased tissues performed by...
The forward-intensity model of Duan, {et al} (2012) is a parsimonious and practical way for predicting corporate defaults over multiple horizons. However, it has noticeable shortcoming because default correlations through intensities are conspicuously absent when the prediction horizon more than one data period. We propose new approach that builds in among individual obligors by conditioning all forward on future values some common variables, such as observed interest rate and/or latent...
We estimate and test long-run risk models using international macroeconomic financial data. The benchmark model features a representative agent who has recursive preferences with time preference shock, persistent component in expected consumption growth, stochastic volatility fundamentals characterized by an autoregressive gamma process. construct comprehensive data set quarterly frequency for 10 developed countries employ efficient likelihood-based Bayesian method that exploits up-to-date...
In state-space models, parameter learning is practically difficult and still an open issue. This paper proposes efficient simulation-based method. First, the approach breaks up interdependence of hidden states static parameters by marginalizing out using a particle filter. Second, it applies Bayesian resample-move to this marginalized system. The methodology generic needs little design effort. Different from batch estimation methods, provides posterior quantities necessary for full...
We develop a new model where the dynamic structure of asset price, after fundamental value is removed, subject to two different regimes. One regime reflects normal period price divided by dividend assumed follow mean-reverting process around stochastic long run mean. The second bubble with explosive behavior. Stochastic switches between regimes and non-constant probabilities exit from are both allowed. A Bayesian learning approach employed jointly estimate latent states parameters in real...
The popularity of convolutional neural networks and deep learning based approaches has increased continuously in the past years. These methods enabled solution various practical problems, but they still are not heavily exploited embedded domain, which requires low-power implementation these architectures. Cellular can provide an analogue power-efficient also enables exploitation non-Boolean, beyond CMOS elements such as memristive structures. In this paper we will demonstrate on simple...
This paper analyzes the interdealer-broker market for single-name Credit Default Swaps (CDSs) using a novel dataset from GFI trading platform. We find that CDSs exhibit reverse J-shaped patterns and quoting activity in U.S., U-shaped Europe Japan. also trade on average 3 times day, which is comparable to corporate bonds. To deal with scarcity of data absence best limits, we estimate state-space model bid ask quotes infer dynamics volatility transaction costs. The estimation uses new...
We propose a density-tempered marginalized sequential Monte Carlo (SMC) sampler, new class of samplers for full Bayesian inference general state-space models. The dynamic states are approximately out using particle filter, and the parameters sampled via sampler over bridge between prior posterior. Our approach delivers exact draws from joint posterior latent any given number state particles, thus easily parallelizable in implementation. also build into proposed method device that can...
Cellular neural networks were used with success in the past decades and helped laying foundations of net-work applications image processing. In last few years convolutional have appeared solution complex practical problems. Meanwhile programming templates cellular designed by analytical methods, gradient based optimization is applied popularly networks. this paper we will demonstrate how these methods can be exploited using they to implement classification feature extraction tasks, both...
The paper proposes a self-exciting asset pricing model that takes into account co-jumps between prices and volatility jump clustering. We employ Bayesian learning approach to implement real time sequential analysis. find evidence of clustering since the 1987 market crash, its importance becomes more obvious at onset 2008 global financial crisis. It is found affects tail behaviors return distributions has important implications for risk management, forecasting option pricing.
We investigate liquidity changes in the credit default swap (CDS) market around two events that increased transparency and standardization during Great Financial Crisis: dissemination of CDS positions starting November 2008, implementation Small Bang July 2009. build an econometric model based on bid ask quotes to measure thinly traded CDSs. find that, after release positions, market-wide deterioration is less important for banks, consistent with information revelation alleviating systemic...