Quantitative Asset Management
AI Alpha Signals
Our Deep Learning models and advanced Artificial Intelligence tools monitor social media squawk for the next best signal! Stockpulse’s outstanding coverage goes beyond equities, indices, and commodities to other asset classes such as FX.
We use Deep Learning on large historical datasets of more than 12 years. This delivers significant results and is crucial for successful pattern recognition in the data. We use Tensorflow and have built our own cloud based on the strongest Nvidia video cards. We can leverage several hundreds of gigabytes of GPU power.
AI Alpha Signals are available on different time scales, ranging from daily to weekly, up to monthly, even quarterly. Signals can be long/short or on a long-only basis. This is usually a highly individual process and depends on the internal requirements of each client. We develop in mutual cooperation, signals within the required parameters which we adjust accordingly.
We maintain very comprehensive backtesting and simulation engine which is used for internal optimizations for our Deep Learning processes as well as the testing of new strategies and application of signals.
60k+
Equities
13 years+
Historical Data
200M+
Diversified Sources
6B+
Historical Messages
Key benefits
Using Deep Learning and Tensorflow for pattern recognition
Long/Short and Long-only signals
Adjusted to clients requirements
Signals available for many different time scales, daily, weekly, monthly, quarterly
Comprehensive backtesting and simulation engine
Very long data history for significant results (more than 12 years, one of the longest dataset in the industry)
Check out our white papers about our AI signals based on Deep Learning
Deep Learning for Equities
Deep Learning on Emotional Data for Daily Long / Short Decisions: In the past few years, machine learning has become a popular application in various fields. One of the core concepts used are neural networks. A neural network is a computational system that loosely models the human brain in order to solve classification problems. Chaining multiple neural nets is called “deep learning”, with “deep” referring to the potentially large number of layers in these kind of models. Our aim of using deep learning is to recognize hidden patterns which potentially influence the oscillation of equity prices.
Deep Learning for FX
Sentiment Analysis for Foreign Exchange Trading: This text discusses the implementation of sentiment analysis into the production of foreign exchange trading strategies. Statistical significance testing for financial performance measures is used to compare generic foreign exchange strategies against sentiment analysis-strategies generated by Stockpulse. The strategy generated by Stockpulse outperforms the generic strategies across all presented performance measures.