Skip to main content

On 8 September 2026, Robinhood announced it had taken equity stakes in Crypto.com and in OG.com, the prediction-markets exchange Crypto.com has spun out, and that it would begin routing retail event-contract volume through OG.com’s CFTC-regulated exchange and clearinghouse. Eligible US customers started seeing OG-backed contracts the same week. The stakes were priced in line with Citadel Securities’ July investment, which valued Crypto.com at $20 billion and OG.com at $5 billion standalone.

Every part of that is legal, disclosed, and unremarkable by the standards of American brokerage. It is also the fifty-fifth unasked half of fairness, and it is the cleanest example this series has had of a fairness question that lives nowhere near the odds.

Because here is the sentence nobody put in a headline. Robinhood now has a financial interest in one of the venues it decides to send your bet to, and you have no way of knowing which venue got your bet or what the others would have paid.

TL;DR

  • Robinhood took equity stakes in Crypto.com and its prediction-markets spinout OG.com on 8 September 2026, and will route retail event contracts through OG.com’s CFTC-regulated exchange and clearinghouse.
  • Event contracts made Robinhood $156 million in Q2 2026, more than tenfold year on year, ahead of both equities transaction revenue ($129 million) and crypto ($100 million). Routing is now the main business.
  • A checkable bet needs four things: the terms, the price, the venue, and the resolution. Robinhood publishes the first two. The venue is chosen for you, by a party that now earns differently depending on the choice.
  • Routing produces no artefact. Only the branch that was taken leaves a trace, so a stranger cannot re-run the decision, and disclosure is a filing rather than a receipt.
  • Satoshie’s answer is structural rather than virtuous: one published contract address, odds readable before the ticket is bought, Chainlink VRF request and fulfilment transaction ids published with every result, and no venue-selection layer to have an opinion.

What actually happened

The facts are dull, which is usually where the interesting part hides. Robinhood is adding OG.com to the list of venues that handle its event contracts, joining Kalshi, ForecastEX and Rothera according to reporting on the deal. OG.com runs a CFTC-regulated derivatives exchange and clearinghouse. Robinhood holds equity in it, and in its former parent. Crypto.com chief executive Kris Marszalek has said the platform intends to push beyond prediction markets into futures and perpetuals.

The financial context matters more than the equity does. Event contracts generated $156 million for Robinhood in the second quarter of 2026, up more than tenfold on the year, comfortably ahead of the $129 million it made on equities transactions and the $100 million it made on crypto. This is not a product experiment tucked into a tab. It is the largest line on the page, and Bernstein’s projections have the whole business at $1.7 billion by 2028.

So consider what Robinhood actually is in this arrangement. You open the app, you see a market, you press yes. Somewhere between that press and a contract existing, a decision is made about where the contract gets listed and cleared. That decision has always belonged to Robinhood. What changed on 8 September is that one of the destinations now sits on Robinhood’s own balance sheet.

Robinhood’s due, and it is real

Multi-venue routing is not a scandal, and pretending otherwise would be lazy. The case for this deal is genuinely strong.

A CFTC-regulated exchange and clearinghouse is a materially higher standard than the offshore norm this blog usually writes about. There is a rulebook, a regulator, and clearing that exists as a separate function rather than as a spreadsheet on the operator’s laptop. Spreading volume across four venues is more robust than depending on one, and that is not theoretical: New Jersey’s attorney general filed a 332-page certiorari petition on 2 September asking the US Supreme Court whether states were ever preempted from regulating these contracts, with Kalshi at the centre of it. A broker with a single venue partner has a single point of legal failure.

Brokers holding stakes in the venues they trade on is also ancient rather than novel. Exchanges were member-owned for most of their history, and the mechanism society settled on for that conflict is disclosure. Robinhood disclosed. Nobody has produced a single fill that was worse for it.

The objection is narrow, and it is not fraud. It is that disclosure is a filing, and a filing is not a receipt.

The fifty-fifth unasked half

Previous instalments of this series have gone after the outcome (who resolves it), the data (which source), the counterparty, the settlement, and the history. This one sits earlier than all of them, at the layer of selection.

A checkable bet needs four things: the terms, the price, the venue, and the resolution. Robinhood shows you the terms and shows you a price. The resolution belongs to whichever venue’s rulebook you ended up under. The venue itself is chosen for you, by a party with a balance-sheet preference among the options, and the choice leaves nothing behind that you could re-run.

That is the shape of every post in this series. Never “they cheated”. Always: there is a step in the process whose correctness you are asked to take on trust, because the step produces no evidence.

A fair market you were never routed to is not a fair market you got.

The two halves you cannot check

One: which venue took your order, and why that one. Even where the confirmation names the venue, naming is not verification. Routing is a decision about alternatives, and only the branch that was taken leaves a trace. Checking a routing decision requires both branches, and the untaken branch was never written down anywhere you can reach. You are not being asked to trust a number. You are being asked to trust a counterfactual.

Two: what the other venues were showing at that instant. In equities there is at least a consolidated tape and a best-execution obligation with data underneath it. Event contracts spread across four venues have no consolidated anything, and the contracts themselves are not fungible. The “same” market on two venues can differ in resolution wording, settlement timing, fee schedule and depth, so even a straight price comparison is not clean. You cannot be shortchanged by a number you were never shown, which is a comforting sentence right up until you notice it is doing no work.

What this looks like in a casino

Strip the regulatory clothing off and the structure is one every gambler should recognise.

A platform offers a coinflip. You press flip. Behind the front end, the platform decides which contract instance runs it, which liquidity pool backs it, which house wallet takes the other side. Two instances, identical published rules, different bankrolls and different economics for the operator. The platform has a preference. You get a receipt for the flip you played, and absolutely nothing about the flip you would have played.

Notice that the randomness never entered this. The generator can be flawless. Provable fairness answers the question “was this outcome produced honestly”, and it says precisely nothing about the question “was this the game I should have been in”. That is the same gap the fourth instalment found in the front end, arriving from a different direction: the proof covers the event, and somebody else picked which event you were standing in front of.

How Satoshie handles it: no router to trust

Our answer is structural, not moral.

  • One contract, one published address. The game you press is the game that runs, because there is no venue-selection layer. There is one venue and its address is readable on Basescan.
  • Odds readable before the ticket is bought. Ticket price is constant, the entrant list is on-chain, and the draw conditions are deployed rather than described.
  • Chainlink VRF for randomness, with the paperwork attached. The VRF request and fulfilment transaction ids are published with the result, so verification is something a stranger can perform without asking us for anything.
  • Coinflip escrows, resolves and pays in a single transaction. There is no interval during which your stake is a loose position waiting on somebody’s operational choice.
  • We do not take one side on some games and a different side on others, because there is no “some games”.

None of that makes us better people. A routing layer is simply a place where discretion can live, and the cheapest way to make discretion checkable is to not have any.

Honest limits, three of them

We have a routing-shaped surface too, and it is called Base. Satoshie settles on Base, where a sequencer decides the order in which transactions land. Ordering is a routing decision wearing different clothes. We did not choose Base’s sequencer policy, we cannot verify it per transaction, and the honest description is that the risk moved one storey up rather than disappearing.

The front end is still a front end. We publish the contract address, and whether the button you pressed actually called that address is something you can confirm on Basescan afterwards and something almost nobody will. We are not exempt from our own fourth instalment.

Satoshie is not a prediction market, and that is a limit rather than a boast. We do not resolve real-world events, so we never had a venue-selection problem in the first place. Holding up our clean record here against Robinhood’s is comparing a problem we solved to a problem we declined to have. The comparison is only worth anything if you are choosing between them, which is the point, but it is not a virtue.

Three questions worth asking, wherever you are betting

  1. Do you know which venue your contract was listed and cleared on, and can you establish it from your own records rather than from the broker’s marketing?
  2. Does the party choosing the venue earn differently depending on which one it picks, and is that disclosed once as a filing or attached to your trade every time?
  3. If two venues showed different prices or different resolution wording for the “same” market, who eats the difference, and what would you compare your fill against?

The odds were never the interesting part. The interesting part is who decided where you were standing when those odds were quoted. Robinhood published the market. It did not publish the fork.

Satoshie runs provably fair raffles and coinflip on Base, using Chainlink VRF, with one contract address and no routing layer to take an interest in the result.

📷 Photo by Mihai Lăzăr on Unsplash

Valentina Ní Críonna

Author Valentina Ní Críonna

More posts by Valentina Ní Críonna