The core of our strategy is our in-house prediction engine, SavantLDF. Our system combines natural-language analytics with our proprietary deep neural model to recognize, extract, and exploit meaningful signals that best reflects short-term market dynamics. MIT's open-source Conceptnet5 ontology serves as the basis for Savant's associative memory, providing a common-sense foundation on which the system can build and understand more sophisticated relationships between market-relevant words and phrases.

Savant's core advantage is not merely computational power but flexibility of intelligence. While the system builds its predictive models from a variety of content sources, it determines which analytic strategy - or combination of strategies - can best suit each content source and adapts accordingly. After all, data is not monolithic and the particular sentiment-driven approach that can best recognize signals from a financial Twitter feed may not be the ideal model for analyzing a 10-K report.