Research · Build · Backtest · Deploy · Monitor
AI-Native Quant Infrastructure for Portfolio Managers
Describe a strategy in one sentence. HyperPM parses it into a specification, loads the matching strategy skill, writes the module, runs it through the static gate and a sandboxed backtest, and hands back the report. Analysis, tuning and data agents answer the follow-ups, and every version is saved to your library.
Built for practitioners
01
Independent PMs and small quant teams
A desk's data, backtests and strategy library without building the desk; hosted deployment and monitoring are on the roadmap.
02
Family offices and fund PMs
An institutional process for rule-based strategies, from hourly bars to daily rebalancing.
03
Research teams and developers
Code-first and API-first: edit the generated module or write your own against the SDK; the platform gates, backtests and versions it, and every run is reachable through the API.
What you can use today
Everything below is live now, not a roadmap.
Strategy agent
Start from a sentence or a template card; the board shows the result, code, parameters, specification, report, tuning, chart and every run.
→Strategy library
Save a version from the conversation; reopen it, backtest it again, keep its tuning history and restore an earlier version.
→Skill library
100 strategy families in twelve categories, each with its guide, a reference module, standard requests and the latest regression record.
→How it works
The six steps from a sentence to a backtest report, who does each, why the result can be trusted, and the full agent architecture.
→
From a sentence to a backtested strategy in six steps
Models do only what needs judgement; the platform's deterministic code does the rest.
- 01
State the request
user - 02
Parse it into a specification
platform (deterministic) - 03
Load the skill pack
platform (deterministic) - 04
Write the module
agent (model) - 05
Gates, then the backtest
platform (deterministic) - 06
Report, analysis, tuning, library
agents + platform
Nine steps, one platform
Data, backtesting, execution and risk are the product itself, not integrations.
4 steps live today5 on the roadmap
- 01live
Research
Ideas and hypotheses explored with the data agent: returns, volatility, correlations, charts.
- 02live
Data
Exchange bars and funding rates, cached and reproducible.
- 03live
Strategy Development
Code-first strategies with AI generation, review and explanation.
- 04live
Backtesting
One engine and one fill model, with fees, slippage and funding settled in the P&L.
- 05planned
Simulation
Paper trading on live data before any capital is committed.
- 06planned
Deployment
Hosted, tenant-isolated runs with a full audit trail.
- 07planned
Execution
A broker-agnostic gateway to brokers and exchanges.
- 08planned
Risk
Limits enforced before an order leaves the platform.
- 09planned
Monitoring
Live P&L, exposures and alerts, with AI-assisted repair.
Start with one sentence.
Pick a template card or type the entry logic; the specification, the code and the backtest report come back in the same conversation.