“Truth - more precisely, an accurate understanding of reality - is the essential foundation for producing good outcomes.”
Convinced that artificial intelligence will profoundly change the world of finance, Rubinstein & Schmiedel pushes the boundaries of AI in the field of financial market forecasting. We develop and operate fully automated, high-precision trading systems capable of creating systematic and sustainable returns - so far unmatched by professional traders and rule-based algorithms.
Rubinstein & Schmiedel was founded in 2019 after more than six years of fundamental research and development. The result: a novel AI which extracts information and learns complex strategies without human input solely based on data. We are 20+ experts in Switzerland and Germany with research backgrounds in AI, Software Engineering, Data Science, Big Data, and High-Performance Computing.
We have created an end-to-end, fully automated software solution covering data acquisition, signal generation, back-testing, trade execution, portfolio and risk management. In 2019, we launched our first 24/7 actively managed investment solution for an exclusive group of professional investors. Focusing on the emerging class of digital assets, we have been providing our partners with impressive returns while minimizing downside risk – in an investment universe known for its high volatility and difficult market conditions.
We are looking to extend our range of solutions, targeting new clients, and connecting traditional asset classes to our unique AI framework.
Our AI models are not limited by human experience. They rely solely on available market data and on individual investment goals to determine a “super-human” optimal trading strategy.
For our forecasting systems, we go beyond traditional optimization, machine learning approaches, and develop all algorithms from scratch, including hand-crafted compute-kernels for utmost performance.
Price charts are a superposition of many independent signals. Our AI system enables us to decompose market data into its underlying components (specific news, market makers, crowd behavior, etc.) to make much more precise predictions.
Our next-gen deep networks require a different way of training compared to traditional approaches (e.g. SGD). Inspired by aerospace research, we have developed novel optimization frameworks for this purpose.
(Deep) Neural Networks were not designed for market data initially. We have developed a novel deep network architecture with specialized neurons that are highly optimized for financial applications, handle the noise component well and don’t require labelled data.
Different parts of our machine learning, analytics, and trading pipeline have been designed to run on GPU hardware for maximum performance. Custom CUDA kernels were developed to improve AI training speed and model response times.
Large amounts of market data and news along with algorithmically enriched meta-information are collected to get a complete snapshot of the financial world at any point in time to ensure maximum reliability and a predictable live performance of all developed trading systems.
All storage and software services are deployed on geographically redundant premises and are configured for high availability. NIST-compliant encryption is used for all communication channels. Our custom-built hardware is locked and biometrically secured inside specialized data centers.
The products, services, information and/or materials contained within these web pages may not be available for residents of certain jurisdictions. Please consult the restrictions relating to the products or services in question for further information. Find our data privacy notice.here.