Black-Box Optimization for Privacy and Security
SigOpt is a black-box optimization solution, meaning that it is able to effectively solve optimization problems without needing specific information about your data or model. This makes our optimization solution great for optimizing trading strategies, simulations, and other highly guarded models. The solution is also highly modular, allowing it to seamlessly bolt into any model, platform, or infrastructure.
Data, Model, and Infrastructure Agnostic
SigOpt works with any data, model, or infrastructure, for security, privacy, and ease of use.
SigOpt never touches or sees your data, enabling you to retain ownership and control over your most valuable resource.
SigOpt seamlessly bolts on top of your model via our numerous client libraries or REST API.
Model training and evaluation occurs on customers’ own infrastructure.
Advanced Features to Amplify Results
Numerous advances features built into our optimization solution enable practitioners to model more effectively and solve fundamentally new problems.
- Multimetric Optimization enables optimization toward two metrics simultaneously, like both trading strategy return and risk.
- High Parallelism allows you to evaluate models in parallel and obtain results faster, especially if you have access to significant computational resources.
- Multisolution Optimization will locate multiple distinct optima which can be compared against each other or ensembled together.
Modular and Scalable
SigOpt’s modular design allows it to bolt onto any model directly, or integrate seamlessly into any modeling platform. Algorithmic trading and simulation firms may have numerous teams running experiments and require high optimization throughput. Once integrated, our optimization solution abstracts all the complexities of infrastructure behind a simple cloud API and scales to over thousands of concurrent optimization experiments.