Settlement · architecture proposal
There is no contradiction. The argument against payment for order flow is an argument about who the customer is, not about block times. Robinhood Chain is the best-capitalised place in the world to settle tokenised equity claims, and their event-contract book is the deepest retail-facing source of market-implied probability that has ever existed.
We are not better at the feature. We are differently paid. That statement survives intact on their chain.
Settlement rails of the first onchain stock options exchange
An onchain equity options exchange has to settle somewhere. This is the case for settling on Robinhood Chain: tokenised equities as the underlying, cheap anchoring, and a real market implied reference from the deepest retail event contract book that exists. Open the terminal.
What Robinhood Chain actually is
Settles to Ethereum using blob data availability; ETH for gas. Stock Tokens plug directly into DeFi protocols, which is the property that matters here.
Four things the chain does for the protocol
01. ANCHOR
The protocol already anchors only the epoch Merkle root, hourly, O(1) per epoch, never per prior. On a 100ms-block L2 that cost rounds to zero, so the anchoring cadence stops being an economic decision.
A root on a public chain converts "trust our log" into "recompute it yourself."
02. RESOLVE
Stock Tokens are onchain, 24/7, and redeemable 1:1. That sidesteps the ugliest part of resolution: a token with a canonical, published redemption ratio makes corporate-action adjustment policy explicit onchain rather than a footnote each adapter re-litigates.
It also kills the "official close" ambiguity, there is no close on a 24/7 book, only a block height.
03. REFERENCE
The scoring rule that gets ranked and sold is skillScore = 1 − B / B_market. It needs a real B_market. The composer currently uses a hardcoded 0.54, a known gap, logged as open question #6.
Robinhood's event contracts traded 13.6 billion contracts in Q2 2026 at an average 126M/day. That is a liquid, continuously-priced market-implied probability on exactly the kind of binary proposition the grammar expresses. Wire the reference to it and the hardcode disappears.
04. BOND
Forecaster bonding against one's own calibration, adapter and resolver bonds, buyer-side staking for query priority, all slashable, all naturally contracts. No emissions: staking rewards come from protocol fee revenue only, and slashed tokens are burned.
Records stay soulbound and off-market. Reputation is never a transferable token, on this chain or any other.
Why the reference matters more than the anchoring
skillScore = 1 − B / B_market B_market today 0.54, hardcoded constant open question #6 B_market wired implied probability from the event-contract book, at the timestamp of commitment ────────────────────────────────────── why this is the hard part the reference must be observed at commit time, not at resolution. a reference read afterwards is contaminated by the outcome, the same error the protocol exists to prevent in its users.
A market reference drawn from a thin book is worse than no reference: it manufactures apparent skill out of illiquidity. Robinhood's event-contract volumes are the opposite problem, deep enough that beating them is genuinely hard, which is exactly what makes the resulting score worth selling.
That difficulty is already visible in the offline harness. The synthetic market reference there is itself a skilled forecaster, and on one seed in six the engine still loses to it.
Read this before getting excited
Wiring a real reference makes the score honest, not better. It is entirely possible that against a real, liquid market-implied probability the engine's measured skill goes to zero or negative. That is the point of building the gate. See the roadmap's kill criteria.
The two problems with this plan
The protocol's single load-bearing claim is that control of resolution must be distributed before governance is, if one entity controls resolution, it controls every score, and no amount of token voting fixes that. Anchoring an unforgeable record to a chain sequenced by a single broker-dealer sits awkwardly against that.
Resolution: Robinhood Chain is an anchor, never the root of trust. Roots reach Ethereum through blob DA, the log is independently recomputable from public data, and the Stage 0 gate, anyone can recompute every score without us, is unchanged. If the chain vanished, the record would survive. An anchor you can replace is not a dependency.
The entire teardown is a US argument: US options PFOF crossing $1B in a quarter, a US SEC settlement, US retail being sold as uninformed flow. Stock Tokens are available to Robinhood users in the EEA.
No resolution offered. The cohort the thesis is about cannot currently hold the instrument the adapter resolves against. Either the protocol resolves US claims through conventional adapters and uses the chain purely for anchoring and bonding, or it follows the instrument to the EEA and accepts a different first market. This is unresolved and should not be smoothed over in a pitch.
And a third, about incentives
Taking the market reference from a venue that also profits from the flow being referenced is a conflict that must be disclosed and monitored, not waved away. The mitigation is redundancy: a reference is only admissible if k-of-n independent sources agree within tolerance, the same standard the resolution ladder already applies. A single-source reference. Robinhood's or anyone else's, is a single point of manipulation.
Sequencing
Contracts are cheap to write and expensive to be wrong about. The engine has to earn its place on real data first, a beautifully anchored record of a model that does not work is still a model that does not work.
| Piece | Depends on | Phase |
|---|---|---|
| Read-only reference feed | Nothing, can be prototyped now against public prints | −1 |
| Epoch anchoring contract | A record worth anchoring | 1 |
| rhc.stocktoken adapter | Adapter SDK, shadow period, ratification | 1–2 |
| Resolver & adapter bonds | Multi-operator resolution existing at all | 2 |
| $PRIOR on Orbit | Profitability without it; counsel per jurisdiction | 4 |
Regulatory note, carried forward unchanged
Deploying on a chain operated by a registered broker-dealer increases regulatory surface, it does not launder it. Applying Howey without flinching: investment of money, common enterprise, expectation of profit, substantially from the efforts of others, $PRIOR is likely a security in the United States. Utility does not defeat Howey. Where no compliant structure exists, ship without the token there.