Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

ml-quant-trading

PyTorch factor-research stack with 213 factors and cost-aware backtests, plus a preview typed portfolio-decision adapter.

Projectbeta
View on GitHub ↗

Overview

ml-quant-trading (package mlquantx) is a reproducible PyTorch research stack for cross-sectional factor research: 213 mask-aware factors, public-data validation, cost-aware backtests and auditable Markdown and JSON reports. A deterministic demo runs the full pipeline on a CPU laptop in about 30-90 seconds with no market-data account.

The Jev connection is a preview typed portfolio-decision adapter documented in docs/systemone_adapter.md, along with a Chinese article on using Jev for portfolio decisions. The README excerpt does not spell out the adapter's question set, so the decision shown here is inferred. The project also has a recipe for running with DeepSeek Harness (DSH) and an agent benchmark challenge. It targets quant researchers who want an auditable baseline.

What the decision model decides

TypeQuestionAnswers
ChoicePortfolio action for a position or candidate (adapter preview)(inferred)—

How it works

  1. Load data and masks, compute 213 factors.
  2. Score stocks with a model and build portfolio weights.
  3. Run a cost-aware backtest and write reports.
  4. Optionally pass portfolio context to the typed decision adapter.

Strengths

  • Runs end to end on CPU with no API key.
  • 100+ tests and CI per the README.
  • Reproducibility contract for public reports.

Caveats

  • The System One adapter is a preview.
  • Research tool, not an audited trading strategy.

Getting started

python -m pip install mlquantx && mlquant demo

At a glance

Type
Project
Status
beta
Language
Python
Runtime
CLI, PyTorch
Has code
yes
Has tests
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

Where does Jev come in?

Through a preview typed portfolio-decision adapter documented in the repo's docs.

Do I need market data or keys for the demo?

No, the deterministic demo uses bundled data and needs no API key.

Is it a trading bot?

No, it is a research and backtesting stack.

Related projects

  • jev-trader

    Trade Arena proof of concept comparing trading strategies on 13 US stock funds, grown out of a pre-registered Jev study.

    ProjectprototypePython
  • jev-claim-vs-measured

    Reproducible test of viral Jev HFT claims: direction not calibrated, never under 100 ms, and no edge after trading fees.

    ServiceresearchPython
  • jev-trade-cc

    Point-in-time O'Neil momentum backtest where Jev picks entries and judges exits, with a jackknife showing no resolvable edge over rules.

    ProjectresearchPython
  • jev-typesafe-real-financial-use-cases

    Decision Lab: sixty finance demos asking Jev, OpenAI Decisions and twelve other decision models the same questions, graded.

    ProjectresearchJavaScript
  • trade-jev

    Backtest harness that tests Jev as a BUY/SELL/HOLD trader on NQ level-10 order-book data, with replay, viewer and paper-trading modes.

    ProjectresearchPython
  • jev-alpha-bench

    Study of whether Jev's reading of financial news predicts stock returns: strong same-day signal, no tradeable alpha.

    BenchmarkresearchGo