Market structureAug 13, 2026 · 6 min read

There is no tape

Robinhood Chain has six venues trading the same 200+ instruments and no canonical record of what any of them printed. In equities, the consolidated tape is the thing everyone cites. Here, asking what something traded at has no single answer.

RobinhoodRPC Research

Engineering & market structure

In US equities the consolidated tape is so foundational it is easy to forget it was built. Every participant cites the same record, which is the only reason a market fragmented across a dozen venues is tractable at all.

Robinhood Chain has the fragmentation and not the tape.

Six places the same stock trades

VenueTypeWhat it contributes
UniswapPublic AMMPrimary public liquidity
RialtopropAMMProp trading venue
LighterPerpetualsLeveraged exposure
ArcusPerpetualsLeveraged exposure
MorphoLendingLiquidations print as trades
RFQ aggregatorsRouted0x, 1inch, LiFi fills
Named in Robinhood Chain's own ecosystem documentation as of 2026-08-27.
Fig. 1: Cumulative share of notional as venues are added, illustrative. A single-venue view captures well under half the activity; the shape of the curve is the argument for a consolidated record.

Ask what AAPL traded at and the honest answer today is a question: on which venue, adjusted how, in which session? Every provider on this chain will sell you decoded logs per venue. None will tell you the answer.

What makes a tape a tape

A feed and a tape are different objects. The difference is a canonical sequence.

  • One monotonic sequence across every venue, so two participants replaying the same range get identical order. Without it there is nothing to cite.
  • One price convention. Every print carries USD per token and USD per underlying share, with the multiplier that converts between them, so nobody has to guess which they were handed.
  • Context that travels with the print: which session, whether the underlying market was open, and the premium at execution — snapshotted, not recomputed later from current data.
  • Reconstructed trades, not raw transfers. A stock leg and a cash leg sharing a transaction hash are one trade, and the tape should say so.

Why sequencing is the hard part

Assigning a canonical order is a deliberate bottleneck: it cannot be done concurrently. That is a real engineering cost, and it is exactly why the property is valuable — a tape that can be produced in parallel by anyone is not canonical.

It also means the tape compounds. History cannot be backfilled faster than it happened, and the correctness of past records cannot be reconstructed once the multiplier and session context that surrounded them is lost. A tape started today is worth more in a year than one started in a year.

Who actually needs this

Anyone quoting across venues, because no single venue's price is the market. Anyone reporting performance, because a return computed from one venue's prints is a return on that venue rather than on the instrument. And anyone doing surveillance or compliance, because a record that cannot be replayed identically is not evidence of anything.

Every input required to build it is public. That is the striking part: this is not gated by data access or a partnership. It is unbuilt because it is composition work on a two-month-old chain, and the incumbents' products were designed for chains where instruments do not split.

This post contains no measured production statistics. The venue coverage chart shows an illustrative shape and is labelled as such. The venue list is taken from Robinhood Chain's published ecosystem documentation as of 2026-08-27; the properties described as necessary for a consolidated tape follow from how consolidated tapes work in equities markets generally.

Built with the captured historical data.

Queries used
Notional by venue, session-awaresql
SELECT    venue,    session,    countDistinct(transaction_hash)    AS trades,    sum(size_usd)                      AS notional,    avgWeighted(premium_bps, size_usd) AS vw_premium_bpsFROM tapeWHERE token_symbol = 'AAPL'  AND block_time >= now() - INTERVAL 1 DAYGROUP BY venue, sessionORDER BY notional DESC;
The research letter

One post a month, built from the archive.

Market-structure analysis and engineering notes, each with the queries to reproduce it. No product announcements, no schedule padding.