Aether Quant

A research toolkit for systematic trading, and the self-adapting trading system built on it.

The aether_quant library is pure Python on numpy and pandas. It offers:

• signal evaluation with HAC t-statistics and bootstrap intervals
• position sizing, from volatility targeting and fractional Kelly to the live path's bounded sizing
• dollar, sector, beta and factor neutrality
• composable cost models with the fees a live path pays
• a cost-aware backtest()
• the overfitting tests: PBO, the probabilistic and deflated Sharpe, the minimum track record and the haircut Sharpe
• pre-registered research rounds and generations that hand over capital by evidence

Every function in it is one the trading system itself uses, so a research result and a live decision come from the same code.

pip install aether-quant
aq hello

That downloads a few years of daily bars for 11 coins from Coinbase's public candles, runs the project's shipped rule through a cost-aware backtest, and prints a report with benchmarks and overfitting checks. It needs no key, no Docker and no Lean. Start with Hello world, then the tutorials.

No verified edge

The shipped book is a long-only crypto trend sleeve, on paper only. No pre-registered test has found a verified edge: 217 trials so far, no survivor, and the holdouts unread. The toolkit is the point. It is how this project judges itself, and how you can judge your own signals.

Install	Getting started: the light install, the extras, and the full system
Learn	Tutorials (they open in Colab) and the guide
Look up	The API reference, generated from the code
Go live	Paper to live: the measured checklist, secrets, accounts, users and the field test
License	The library is Apache-2.0; the trading system is PolyForm Noncommercial