Products‎ > ‎

DiFMan™

Digital Fund  Management

DiFMan™ is a production quality implementation of FrANN designed to utilize large publicly available data sets to locate periodic stock arbitrage opportunities. DifMan™ has as its background work we began for the NSA in seeking user intent using obvious and non-obvious relationships in large amounts of handwritten / video / audio / structured and unstructured data. The same techniques are being applied here to generate the cognitive framework required to locate arbitrage opportunities in the market. Additional information about how the system was designed and built can be found on the FrANN™ page where DifMan™ is used as our FrANN™ development example.

Please note - DifMan™ is a research project into the suitability of using a FrANN™ to seek obvious and non-obvious relationships in large arbitrary datasets which consist of structured and unstructured data. While created in an 'investment scenario' the subscriptions are reports on the AI's findings and should not be considered investment advice.

Our initial DifMan™ system will not work in real time. It will perform simulated periodic re-balancing of a portfolio to maximize returns and issue printed findings at the end of an appropriate trading period. This periodically generated report is completely generated by the system, the purpose of which is to demonstrate learned goal seeking and natural language communications capabilities.

A second phase of DifMan™ will add the ability to interface with online real-time trading systems to autonomously generate buy / sell orders. A final iteration ideally would be given control of a hedge fund and allowed to invest at will. Phases 2 and 3 have legal issues which have not yet been resolved. Stay Tuned!