Real Time Math

It sells two on-premise numerical libraries — FinMath and FinAnalysis — for the .NET and Java runtimes. FinMath covers linear algebra, optimization, random-number generation, statistics, and option valuation; FinAnalysis layers on more than 40 technical indicators and 15 logical predicates for multi-series time-series analysis. Both are pitched as low-latency tools that avoid garbage-collection pauses during real-time model re-estimation.

The firm states its components are in production at high-frequency stat-arb and algorithmic trading organizations operating across global equities, futures, forex, and options. The website does not name specific clients, but the focus on zero-garbage .NET/Java interfaces suggests quant teams that need to co-locate numerical work inside an existing execution stack rather than calling a separate analytics server.

It wraps Intel’s Math Kernel Library and Integrated Performance Primitives, then adds engineering to minimize or eliminate garbage collection in the managed runtime. The same object-oriented API — identical in C# and Java — exposes tuning parameters for precision control while the underlying native code runs on extensively threaded, optimized Intel routines.

It is exclusively a software vendor. The company does not trade, manage capital, or run a platform. Its business model is licensing numerical libraries to trading organizations, and it offers a 30-day free trial on its website for evaluation.

The commercial wrapper focuses on two things that generic open-source libraries often sacrifice: deterministic low latency inside a managed runtime, and a unified API across .NET and Java that handles threading and memory layout. The firm also bundles a broad set of functions — from linear algebra to technical indicators to solvers — under a single license, arguing this reduces the integration and procurement overhead of assembling separate libraries.