We learn predictive models of price in financial data streams.
We build very high-performance, automated trading systems that house these predictive models.
We operate these trading systems in U.S. stock markets, where we compete with other
market makers for each trade, narrowing spreads and dampening volatility.
We are 17 computer scientists, engineers, and mathematicians, mostly PhDs, most
with roots at Bell / AT&T Laboratories. None of us has a background in banking or trading.
Founded in 2007, we have one team at one site a half hour outside NYC.
We discover predictive signals in financial data streams using machine learning, statistical analysis, visualization, and other quantitative techniques. The streams are large (billions of events per day) and extremely noisy, but yield to sustained analysis coupled with occasional bursts of inspiration.
Our C++/Linux systems are distributed and have latency requirements specified in nanoseconds. Our input streams are big and bursty. Safety is paramount. This is not for the faint of heart.