Research-stage memo: publication-status statements below describe when this research memo was prepared. The article itself records the current publication status.

Papeasy expanded reconciliation and publication assessment — October 21, 2026

Main finding
The premise of 393 supplier wallets was incorrect. All 393 non-exchange incoming-transfer counterparties have now been classified using the ERC1155 event's operator field. There are 87 collector-pull suppliers and 306 counterparties to the collector's ordinary exchange purchases, with no overlap. The 87 direct suppliers exactly equal its supplier cash-return recipients. Summing all 393 wallets as a single operation would import unrelated trading.

The collector plus all 87 actual suppliers were reconciled over identical windows. Their rolling-window collateral trading cash surplus is $21,718.250959 before the gas-token purchase, or $21,599.994599 after that cash outlay. Adding the bounded change in the captured CTF claims gives $21,367.245053 to $21,764.280638. This is an explicit collateral-and-captured-CTF calculation, not certified total profit after every asset, liability, gas expense or external position transfer. It is nowhere near the captured $1,404,347.9109331253 ranking. The ranking formula itself remains unresolved.

Classification is established, not inferred from wallet size
CTF operator = collector 0x111f73e91f85b6fe4de1ddec3de2fe32122e355b: 87 source addresses, 165,792 UTC-day-window movements, 4,471,248.128126 shares.
CTF operator = exchange 0xe111180000d2663c0091e4f400237545b87b996b: 6,892 movements, 29,351.817623 shares, matching the collector's own 6,892 indexed BUY rows and $14,327.456941 purchase cash. These involve 306 other addresses plus the exchange itself.
The original 37-wallet sample is fully contained in the 87 direct suppliers. The 393 count remains a valid count of non-exchange source counterparties, but the earlier supplier label is superseded. An on-chain operator identifies the transfer route; it is not proof of common beneficial ownership.

Aligned rolling 30-day cash bridge
Window: September 21 12:36:21 UTC to October 21 12:36:21 UTC; blocks (94,192,732, 95,920,541].
Collector redemption receipts                         $2,799,738.899231
Actual supplier exchange payments                    -2,778,881.886831
Collector exchange payments                             -14,327.456941
Supplier plus collector rebates                          +15,188.695500
Collateral trading cash surplus                           21,718.250959
Gas-token purchase cash outlay                              -118.256360
After that cash outlay                                    21,599.994599

The API's supplier BUY cash totals $2,778,828.612831. The chain ledger exposes another $53.274 in two purchases in transaction 0x74c357e27256af94a14e7f3ce76d76e4ae9f8db81bd75cb04bac709f29efaee2: $30.75 for 75 shares and $22.524 for 56.31 shares. They are missing from those wallets' cursor-complete activity responses. These are receipt-confirmed purchases, not assumed fees. The bridge uses actual chain payments, so this API omission is included.

Cash closure and internal-flow removal
Complete token-transfer extraction for every direct supplier, plus the existing collector ledgers, reconciles each wallet's pUSD, USDC.e and native USDC at three opening boundaries and the fixed close: 792 checks, zero nonzero residuals.
The 88-address group's rolling collateral cash is $213,413.023938 at opening and $201,413.018537 at close, a $12,000.005401 decrease. Transfers within the group cancel. There are $33,600 of other outward user/Safe transfers and $118.256360 spent acquiring native gas tokens. Adding the $33,600 treated as capital withdrawals to the cash decrease gives the $21,599.994599 subtotal. Counting wrapping mints as new income or counting supplier cash returns as extra expenses would double-count internal conversion or redistribution.

Opening and closing claims
The 87 suppliers' current OPEN inventories and complete fixed-window activity contribute 56,264 token IDs across 516,740 wallet/token pairs. Canonical payout queries, archival balances, and the complete collector-delivery census are used rather than trusting current API values. The delivery census adds 47 missing wallet/token pairs, including two token IDs omitted from that metadata universe; their balances are zero at both examined boundaries.
Supplier opening resolved claims: $1,614.376378; unresolved claims range from $0 to $348.532635 at dollar par.
Supplier closing resolved claims: $1,730.159467; unresolved claims range from $0 to $48.502950 at dollar par.
The collector's $9.368821 claims are unchanged and add zero period return.
Therefore captured claim-value change lies between -$232.749546 and +$164.286039. This is a conservative 0-to-par range for unresolved binary claims, not a recovered historical market mark. The resulting collateral-and-claim return range is $21,367.245053 to $21,764.280638 after the gas-asset purchase cash outlay.
This does not independently enumerate lifetime holdings of all 87 wallets or every CTF contract. Assets absent from current OPEN metadata, indexed activity and collector-delivery records, external position transfers, liabilities, and other assets require separate treatment before certifying full economic profit. The gas purchase is an asset conversion; deducting its cash cost is a stated policy, not a measurement that all purchased gas was consumed.

Other aligned time boundaries
UTC-day month (from block 94,162,484): cash trading surplus $23,940.750287 before the gas-asset purchase, $23,822.493927 afterward.
Calendar month (from block 94,738,440): cash trading surplus $11,862.876182.
These are identical address scopes and closing blocks. Different UTC/day/month conventions do not create a cash result near $1.40M. Within the captured nonnegative CTF asset universe, even adding all ending supplier claim value at its par upper bound while assuming no opening supplier assets would add at most $1,778.662417. Exact starting marks for these alternative windows were not reconstructed; no point estimate is claimed for their full return.

What happens to the earlier $921K gap?
It does not close. That comparison mixed 37 suppliers' interior-days costs with all collector receipts and a static ending-inventory figure. The inventory correction was only +$925.155473. Aligning those 37 suppliers adds $76,526.531964 in BUY costs; including the other 50 actual suppliers adds another $386,030.738439, partly offset by $2,035.9234 in rebates. Two omitted indexed purchases add $53.274 in costs. Replacing static ending inventory with the actual bounded change produces the $21K range above.
The remaining difference from the ranking is approximately $1,382,583.63 to $1,382,980.67 in this explicit scope. It is a metric/scope discrepancy, not unlocated collateral. No missing $921K pile of cash or claims was found. See bridge_waterfall.json for exact arithmetic connecting the earlier subtotal to the new one.

Additional candidate test
The indexed supplier BUY costs split by terminal outcome into $1,727,768.674661 on winning tokens, $1,051,035.951620 on losing tokens, and $23.986550 unresolved. Subtracting only winning-side cost from gross also does not reproduce $1.40M. A fitted deduction or guessed half multiplier is not accepted as the formula.

Publication assessment
There is enough evidence for a dated October update now. Preserve the September observations as historical evidence and state that their fitted payout relationship has not been validated as the current formula. Correct the 393-supplier label; show the 87-supplier aligned cash bridge, its stock-value range and explicit limits; distinguish the official marked-equity/net-flow definition from the native profile ledger. Do not call the $1.40M exact gross payouts or certified profit net of supplier purchase costs. Do not call the $21K range unconditional operation-wide profit or imply that the ranking provider has disclosed the exact adjustment.
The missing backend inputs remain the actual ranking valuation timestamps/blocks, opening and closing marked asset rows, cash and noncash capital-flow classifications, wallet scope and any legacy adjustments. Repeated unchanged rankings despite changing receipts also leave snapshot staleness relevant.

What was and was not completed
All 393 incoming counterparties were classified; all 393 current inventory pulls completed (572,437 position rows). Complete indexed activity and token cash ledgers cover all 87 actual suppliers. Broader exchange-counterparty trading histories were initially collected, then stopped after the scope correction; one unrelated high-volume counterparty had not finished. No complete 393-wallet economic profit claim is made. The expanded ordinary-counterparty captures are retained as research evidence, not combined into the supplier bridge.
No article was published, and the canonical site source remains unchanged. A full proposed HTML update and patch are included for review.

Evidence map
RECONCILIATION.json: exact windows, cash equations and claim-value ranges.
corrected_scope.json / operator_summary.json / counterparty_operator_census.json: transfer-operator attribution.
incoming_shares_with_operator.json: incoming movements with operators, block, token and transaction identity.
direct_cash_coverage.json / all_cash_proofs.json: complete supplier transfer ranges and 792 state checks.
cohort_external_cash.json: internal transfers removed, external counterparties retained.
aligned_activity*.json: exact-second/block-aligned indexed activity, including original 37-wallet comparison.
exchange_payment_activity_differences.json and additional_exchange_receipt_decoded.json: two missing indexed purchases.
all_inventory_marks.json / all_inventory_positive_candidates.json / all_payout_states.json: historical claim states and bounded values.
delivered_token_inventory_supplement.json: 47 missing wallet/token pairs independently checked.
bridge_waterfall.json: exact bridge from the earlier mixed-window subtotal.
supplier_buy_cost_by_terminal_outcome.json: winning/losing acquisition-cost test.
raw/, chain_*/raw/, manifests and scripts: source captures and reproduction. API keys are not stored in manifests.