Products
3 available · 5 in development
Software for numerical computing and quantitative trading
Available
- 01 Numerical library for Java SuanShu SuanShu is an object-oriented, high performance, extensively tested, and professionally documented math library. Linear Algebra · Calculus · Optimization · Statistics · Time Series Analysis · Extreme Value Theory · Differential Equation · Digital Signal Processing Read more→
- 02 Quantitative trading toolbox AlgoQuant AlgoQuant is a toolbox for building automatic trading bots. It has a large collection of tools, mathematical modules and algorithms that you can use to create your quantitative trading strategies. Historical Data & Processing · Strategy Optimization · Backtesting · Monte Carlo Simulation & Bootstrapping · Sensitivity Analysis · Portfolio Optimization & Asset Allocation Read more→
- 03 Models for AlgoQuant Quantitative Trading Strategies We have implemented a collection of trading strategies, utility models and algorithms which can be purchased separately. Please note that you need to have AlgoQuant, the quantitative trading research platform, to run these models. Covariance Selection · Mean Reversion · Portfolio Optimization · Trend Following/Momentum Read more→
In development
New products for the age of machine learning, built on the same numerical foundations. Not yet released.
- 04 Precision-aware linear algebra In development Mantissa A linear algebra engine that chooses floating-point precision from the data instead of defaulting to double: low-precision factorizations, refinement in higher precision, and stochastic rounding for long accumulations. Precision budgeting from sample size · Mixed-precision Cholesky and least squares · Iterative refinement · Stochastic rounding · Error reports per solve Read more→
- 05 Certified machine-learning surrogates In development Certus Neural surrogate models for pricing and risk that come with computable error bounds, so that a fast learned model can be trusted, or refused, input by input. Surrogates for pricing and Greeks · A-posteriori error bounds · Automatic fallback to the reference method · Validation reports Read more→
- 06 Learned solvers with classical guarantees In development Operator Neural operators and learned preconditioners for PDE pricing and model calibration, built to accelerate classical numerical methods rather than replace them. Neural operators for pricing PDEs · Learned preconditioners · Calibration of stochastic volatility models · Classical solver as ground truth Read more→
- 07 Overfitting control for ML trading models In development Tempo A training toolkit that treats training time as the regularization parameter it is: effective degrees of freedom, out-of-time stopping and honest validation for machine-learning trading models. Training-time and step-size diagnostics · Effective degrees of freedom · Out-of-time early stopping · Regime-aware re-validation Read more→
- 08 AI research assistant for AlgoQuant In development AlgoQuant Agent A research agent that turns a trading idea written in plain language into an AlgoQuant strategy, a backtest and a validation report, with guardrails against the overfitting that automated search invites. Idea to AlgoQuant strategy code · Automated backtests and sensitivity analysis · Multiple-testing aware reporting · Human review before anything runs live Read more→