what comes back when you sell

latest round block 93,836,017 · canon rhdepth-v2 · 36 rows · 2026-10-20 23:37 UTC
roundKeccak 0xa57fe1a575eb681ffbbb266933ee340d8ec33ee77c0fcfec60f33e40920fa02b
day 2026-10-20 · 48 rounds · blocks 93,024,975 .. 93,836,017
in the ledger this page does not ask — ask the ledger itself. Commits are batched, so the round above may not be on chain yet; these two always answer the same way, which is what makes them worth running:
    cast call 0x7f5446b920e09531f443ce951076cbaed09dfab6 'getRoundHash(uint64)(bytes32,uint8)' 61129566 --rpc-url https://rpc.mainnet.chain.robinhood.com
    cast call 0x7f5446b920e09531f443ce951076cbaed09dfab6 'getRoundHash(uint64)(bytes32,uint8)' 53808204 --rpc-url https://rpc.mainnet.chain.robinhood.com
    61,129,566 is a published round in the backfill — hash, canon 2. 53,808,204 is a published round that was quarantined (canon rhdepth-v1) and sits below the first committed block, so it returns canon 0 and always will. Canon 0 means absent, not broken.
generated 2026-10-21 06:05 UTC · regenerated daily with the data, so this page cannot be older than the day it shows

	$100	$1,000	$10,000	$100,000
SPY	99.90constant	99.90constant	99.87constant	99.7199.63–99.72
QQQ	6.49*constant	1.86constant	0.26constant	0.03constant
AAPL	99.7899.78–99.92	99.3499.28–99.87	99.1199.11–99.36	97.3097.28–97.36
GOOGL	99.30constant	99.27constant	98.99constant	96.2196.14–96.45
NVDA	99.8499.82–99.90	99.8199.47–99.87	99.5899.22–99.63	97.1996.84–97.23
TSLA	99.6999.39–99.81	99.5899.29–99.70	99.07constant	97.0196.98–97.04
RDDT	99.29constant	98.95constant	95.7595.70–95.93	27.4825.55–27.51
AMC	99.76constant	99.7499.69–99.75	99.5499.01–99.63	92.880.02–97.06
GME	95.29constant	46.5444.40–50.05	5.655.41–6.05	0.580.55–0.62

Recompute any of it. Every round is pinned to one block, canonically serialised and hashed. Replaying one needs an archive endpoint — a public RPC keeps only minutes of state history, and the verifier refuses explicitly rather than hanging when pointed at one.

python code/verify.py 93836017

This page is the dated series, regenerated daily; the report page is a different window. That page is a frozen snapshot of 12 rounds ending at block 54,270,401 (2026-09-04) and does not change after publication. A figure that differs between the two pages is a difference of window, not of method.

Known limitation, stated here rather than in a footnote elsewhere. Every row is the best route among the pools in the published pool table, and that table keeps the eight best pools per token. A pool created after it is invisible here.

* QQQ at $100. This cell is one of the numbered errors in the published archive. The pool table ends at block 53,983,886; a cheaper 0.05% QQQ/USDG pool was created at block 55,328,824, after that cutoff, so the series routes around it. Measured against the chain, this cell is about 95.6%. The damage is bounded and the bound is checkable: that pool returns no quote above about $300, so the $1,000 row and every deeper size are unaffected by it.
What is not established: whether other rows are affected by other post-cutoff pools. That is the post-cutoff pool scan, and the repository says so rather than implying this is the only one.

The failure is not one thing. Some tokens collapse at every size with almost no variance; some switch between two states with nothing in between; some slope down as size rises. The range above shows how far each cell moved within the day — it is not a characterisation of the token. A token that alternates between two states can look constant on any single day, and one that is quiet today can be wide tomorrow. Which shape a token has is a property of the series, not of one day; it is established in the report, not here.

data, method, verifier and error archive · every entry numbered, each carrying the side of publication it landed on — the count lives on the archive's own page, not here

Not the quoted price. The result of a full round trip, inside one pool.