Research directions and product hypotheses.

Measure how well each valuation method anticipates later financing, tender, secondary-market, and IPO reference values.

• Define comparable reference outcomes and preserve the model version used for every forecast.
• Compare IPO-ladder and threshold-curve errors out of sample rather than assuming equal accuracy.
• Replace judgment-based weights only after the evidence is large enough to support calibration.

Make past estimates reproducible and show how market inputs, assumptions, and Rai outputs changed over time.

• Retain source quotes, fitted probabilities, model assumptions, and calculation versions together.
• Increase observation frequency when infrastructure and provider terms permit it.
• Publish enough history to distinguish an actual trend from a small number of snapshots.

Expand coverage when a company has a complete, verifiable set of markets that maps to a documented Rai method.

• Prefer complete outcome ladders or coherent threshold sets over isolated headline contracts.
• Require canonical source rules, usable quotes, and explicit open-tail assumptions before publication.
• Track new private-company markets without lowering the validation standard to increase the ticker count.

Revisit Kalshi only after its data-access, display, attribution, redistribution, and commercial-use requirements are clear for Rai.

• Obtain written permission or rely on published terms that clearly cover the intended use.
• Map contracts only when their resolution rules and valuation targets are comparable to an existing method.
• Avoid counting the same underlying prediction twice when venues share information or nearly identical contracts.

Test inputs that do not originate from the same prediction-market information set.

• Evaluate financing rounds, tender offers, secondary transactions, revenue, and comparable-company evidence.
• Define a separate schema and uncertainty model for each evidence family.
• Combine families only when their valuation basis and calculation date are comparable.

Explore a company-specific instrument whose reference value follows a transparent basket of prediction-market positions and Rai’s published valuation rules.

• Specify collateral, position sizing, rebalancing, settlement, fees, and a reproducible reference net asset value.
• Measure tracking error between the prediction-market position basket and the Rai valuation estimate.
• Resolve securities, derivatives, market-access, custody, liquidity, counterparty, tax, and investor-protection requirements with qualified counsel and licensed partners.
• An OpenAI Rai Stone would reference a model and position basket; it would not represent shares or ownership in OpenAI.