Explore quantitative research
Move from a market hypothesis to a testable strategy with AI assistance.
Turn a research question into conditions you can test.
Pyon combines an AI research workspace with a visual strategy builder. Ask questions about an asset or sector, inspect the supporting analysis, and turn a trading hypothesis into conditions you can test. Historical aggregate prices are adjusted for splits and dividends, while coverage depends on the chosen symbols and data provider.
For model-driven work, ML Lab supports training several statistical and machine-learning model types. Models used in a strategy backtest can be replayed with walk-forward predictions when a prediction series is available. A two-parameter sweep in the builder can compare nearby settings on a return, Sharpe, and trade-count grid.
AI research workspace
Explore an asset or market thesis and inspect the generated analysis before acting on it.
Adjusted price data
Use historical price aggregates adjusted for splits and dividends. Coverage varies by symbol and timeframe.
ML Lab
Train supported models and replay their predictions walk-forward in a strategy backtest when available.
Parameter sweep
Compare two strategy settings on a grid of backtest results.
Frequently asked questions
What can the AI research assistant do?
It can generate analysis for a company, sector, or asset. Treat its output as a starting point to inspect and verify, not as investment advice or a validated signal.
Is the historical data point-in-time and survivorship-free?
Pyon requests adjusted historical price bars for selected symbols. The strategy backtester does not construct a point-in-time universe or include delisted constituents automatically.
Can I train an ML model?
ML Lab supports several model families, including gradient boosting and neural networks. A trained model can feed an ml_predict node in the visual strategy builder.
Does Pyon include a Python notebook?
There is no integrated Python notebook or public Python SDK in the current product. Use the AI research workspace, ML Lab, and visual builder for the supported workflow.