← Back to Projects
Open source from day one. Track progress and contribute. View on GitHub →

Probabull is a belief and risk engine, not a hype machine. Most stock apps chase price targets and hot takes. This one gives you probability distributions instead of single numbers, evidence you can actually test, and reliability scores that tell you when the system just doesn't know.

The point is honest uncertainty. Forecasts come with calibrated confidence intervals, regime awareness, and clear triggers for what would flip the view. When conditions are too messy for a good call, it says so. Paper trading with guardrails that build habits, not gambling instincts.

Built in the open for researchers, developers, and curious minds who want to understand how modern quant systems actually work. Check out the implementation plan for the full technical blueprint.

What Makes This Different

Not another app claiming to predict stocks. Credibility comes from being honest about what you don't know, not from flashy accuracy numbers.

Development Updates

January 24, 2026

Project Kickoff

Starting Probabull for real. Working on the initial architecture, making sure the deep learning, reliability, and LLM pieces can plug together cleanly. Building this in the open because why not. Read the technical plan.

Architecture Planning

More Updates Coming Soon

Check back for regular development logs, technical deep dives, and progress reports.

Prediction Targets

Multi Task, Multi Horizon

No raw price predictions. Instead: return distributions and risk events that actually matter.

Build Roadmap

Phase 1: MVP

  • Deep baseline with quantile outputs
  • Confidence thermostat with abstentions
  • Evidence timeline
  • Thesis Translator for core claim types
  • Simple regime detector

Phase 2: Credibility

  • Mixture of experts regime gating
  • Calibration dashboards
  • Conformal intervals
  • Red team analog search
  • Thesis tracking alerts

Phase 3: Platform

  • Public API with billing and SDKs
  • Skill tree learning mode
  • Expanded data sources
  • Options features like IV and skew
  • Community contributions

Want the API specs and repo structure? Check the implementation plan.