As a lead manager in machine learning and software/algorithmic engineering, primarily in the domains of scientific computing and automated trading and based in Zurich, Switzerland, I implement, explore, and apply a multitude of modern computational tools for various tasks in machine learning and data science. This blog is intended as a portal to share many of the computational tools that I’ve built and used with success over the past few years, with an emphasis on real-time signal extraction and predictive analytics in large time series data, along with newly developed machine learning methodologies.
The recent explosion in the past few years in deep learning and data science fields have spawned a large domain of new applications of deep network architectures for generalized learning in very large high-dimensional data sets. Thus this blog also serves as a communication portal for a deep learning tool I recently developed called MDFA-DeepLearning, which proposes a fast real-time feature extraction technique coupled with a deep recurrent weighted average network or an LSTM, for adaptive learning in multivariate nonstationary time series. The package is available on my github page (link at bottom).
Finally, this blog also serves as a communication portal for the software packages and apps I have written, namely iMetrica and it’s newer version iMetricaFX, along with various tutorials in MDFA, deep learning, R, financial trading and other subjects. I built two automated financial trading platforms, one called TWS-iMetrica that currently trades on Interactive Brokers, and a more general MDFA-Tradengineer automated financial trading platform using MDFA coupled with DeepLearning, available soon for license and can connect to several different brokers. With nearly 5 years of algorithmic trading experience, my Sharpe ratios are typically between 2 and 5, depending on the year. Requests on performance track records can inquired via email. As the MDFA paradigm is quite generalized, I’m currently looking for opportunities to expand to many other markets, including commodity futures, equities, index futures, and fixed-income.
And now for the requisite educational background blurb: an M.Sc. and Ph.D. in Applied Mathematics and Scientific Computation from the University of Maryland, College Park, completing a dissertation on large scale computational hybrid atmospheric models under a graduate fellowship with NASA Goddard Space Flight Center. I then switched gears and headed in the direction of econometrics, predictive analytics, and time series analysis. As a post-doctoral fellow, I developed a software package in C, Fortran, and Java called iMetrica – an interactive approach to hybrid signal extraction and forecasting that focuses on graphical tools and interactive controls for building stochastic models for all sorts of economic and financial data, filters, signal extractions, and forecasting. A follow-up to iMetrica was completed and open-sourced as iMetricaFX, which is written entirely in JavaFX, and accompanied by MDFA-Toolkit and MDFA-DeepLearning packages for learning, analyzing, and computing in real-time data analysis in large time series.
If you like the results that you see in the articles presented in this blog, I’m always open for collaboration and/or consulting opportunities. My Github can be found here.