Qelly Live Lab
AI estimates probability. Markets set price. Qelly decides what belongs in the portfolio.
Free: live performance curve + aggregate activity. Live Lab: final decisions, probability movement, exposure, risk policy, weekly attribution.
* Convex solver in production. Experimental D-Wave hybrid runs published on /research.
How we measure skill
Once the research portfolio reaches n ≥ 30 settled forecasts — target: post-World Cup group stage — Qelly publishes the four metrics that separate skill from variance. Until then, structural metrics only.
Qelly is not a pick feed. Every recommendation is logged as a portfolio decision: market price, Qelly probability, sizing rule, exposure limits, thesis, and later attribution. The decision log is the audit object.
Optimizer backends
Every position-sizing call is tagged with the optimizer that produced it. classical 21 · quantum (CPU) 1
Convex solver in production. Experimental D-Wave hybrid runs in /research.
Mathematical definitions, sample-size thresholds, optimizer-backend tagging, open caveats.
Event markets reward consistent sizing, probability estimates tracked for calibration, and disciplined exit rules. Qelly automates the math. You and your venue accounts handle the rest.
You can't watch hundreds of thin markets across venues at once — by the time you see the mispricing, it's gone.
Sizing by gut feel produces correlated overexposure on losers and timid sizing on winners.
Manually rebalancing across venues is slow, error-prone, and ignores correlation between events.
Without a decision log you can't separate skill from variance, or fix the rules that lost you money.
Discover
The engine ingests event markets across venues and surfaces probability-vs-price gaps with explicit confidence bands.
Size
Fractional Kelly and a hybrid quantum-classical optimizer propose position sizes within the research risk policy:
Kelly ×0.25 · 3% max per position · 10% per event · 5% daily-loss circuit breaker
Gate
The beta logs paper decisions and blocks live execution. Future venue actions require user confirmation, eligibility checks, and approved integrations.
Audit
Every probability estimate, sizing call, and final decision is logged with the inputs that produced it. Exportable for attribution or post-mortem.
Start with public proof, then move into paper portfolios and user-confirmed research workflows.
Paper-first - engine sizes and logs
The engine continuously discovers candidate positions in the topics you follow, sizes them with fractional Kelly under your risk caps, and records paper decisions while live execution remains disabled during beta. You set the policy; the engine works within it.
Your input
Topics & risk policy
Engine does
Analysis, sizing, decision logs
Included on
All plans, paper only
Public proof is free. Live Lab unlocks the real-time portfolio intelligence layer. Personal paper portfolios and private research plans come after the first beta cohort.
Subscription pricing is monthly analytics access only. Qelly does not custody funds, does not accept user bankroll deposits, and does not charge performance fees during beta.