Last updated: March 1, 2026
Ulvistium was founded in 2024 by a team of machine learning researchers, quantitative engineers, and portfolio managers united by one goal: build the first trading platform that learns continuously from real-world outcomes — not quarterly model reviews.
The system processes 16,000+ data points per second across three reinforcement-learning layers: signal generation, weight attribution, and outcome feedback. Order books from 35+ exchanges, interbank forex flows, macroeconomic releases, and earnings transcripts feed the engine in real time.
Funds are held in segregated accounts at regulated European brokers. AES-256 at rest, TLS 1.3 in transit, mandatory 2FA, and continuous penetration testing form the security foundation.