# Michigan focal wallet: acquisition and reversal

The focal wallet acquired 128,682.920742 Michigan Democratic YES shares in five orders between 11:34:24 and 11:48:34 UTC on November 2, then sold 128,682.92 in two orders at 12:25:37 and 12:27:04. Its YES balance was zero immediately before the first buy and 0.000742 immediately after the last sale. It paid 106,044.525397 pUSD for the buys including fees and received 101,867.494520 pUSD from the sales after fees, a 4,177.030877 pUSD cash shortfall on these seven transactions. The tiny residual share remains. This is not a complete account profit or loss: it excludes gas, other markets, other assets, outside hedges and payments. The observation strengthens the prior account of two-sided trading: the previously selected buy and sales alone could not establish a near-complete reversal.

The exact transaction hashes, native block times, own-order log indexes, shares, gross collateral, fee fields and net wallet transfer amounts are in [seven_fills.csv](seven_fills.csv). The five gross buys were 105,278.659997 pUSD plus 765.865400 fees; gross sales were 102,696.643800 pUSD less 829.149280 fees. The wallet's pUSD transfer in each of the seven retained receipts agrees with the row's fee-adjusted cash amount. Neither a cash shortfall nor a trading pattern establishes motive, controller or political funding.

## Execution sequence and context

The 11:48:34 purchase consumed maker executions equivalent to 14,119.39 YES shares at 0.81 and 105,799.419633 at 0.82. The 12:25:37 sale executed 45,000 at 0.81. The 12:27:04 sale executed 30,455.67 at 0.80, 36,485.28 at 0.79 and 16,741.97 at 0.78. The counts sum exactly to their respective taker aggregate shares. `ladder2.py` converts complementary-token maker prices to YES equivalents and uses the maker legs in each own transaction, not extra independent orders. These are executed levels, not a historical order book or proof that this wallet caused the market move.

The retained one-minute samples show Michigan Democratic YES at 0.795 at 11:30:17, 0.805 at 11:48:16 (18 seconds before the large buy), 0.795 at 12:27:16 (12 seconds after the large sale), and 0.795 at 12:35:16. At approximately the same four times Michigan Republican YES was 0.205, 0.205, 0.175 and 0.185; Texas Democratic YES was 0.755 at all four; Texas Republican YES was 0.255, 0.245, 0.245 and 0.245. These are sparse sample points from the prior package's `raw/*_price_7.json`, not contemporaneous executable quotes. Republican and Democratic contracts are not exact complements because another party may win. No causal price effect is established.

The wallet also sold three other positions in South Korea-related markets at 11:32:06, 11:32:19 and 11:32:27, receiving 2,585.140090, 2,343.513060 and 1,201.816270 pUSD respectively. Receipt and native block files `raw/receipt_10.json` through `receipt_12.json` and `raw/block_10.json` through `block_12.json` support those specific sales. The earlier retained API activity includes June purchases in those markets; it does not establish a complete lifetime history or account age. This broader activity supplies context, not an explanation of the Michigan trading.

## Immediate collateral route

A bounded pUSD incoming-transfer scan from 11:34:24 November 1 to immediately before the first Michigan buy returned the three Korean sale proceeds; the dRPC pUSD balance at the beginning was 11.289539 and immediately before the first Michigan buy was 6,141.758959. During the sequence two mint events credited 1,000 pUSD at 11:44:15 and 98,999 pUSD at 11:48:13, the latter 21 seconds before the large buy. Their receipts show equal USDC.e flows from `0x6f4a53d86921d443c6c6d8e82a1a625120dda6cd` through the pUSD collateral contract. These are observed collateral transformations and do not identify the original source or owner of the underlying assets. Existing funds, platform infrastructure, and outside sources remain possible; the scan does not establish or exclude political funding. See `raw/fund_log.json`, `fund_during.json`, `fund_receipt_0.json`, `fund_receipt_1.json` and `pusd_log.json`. No identity hunt or upstream inference was performed.

## Coverage and limitations

The fixed interval is [09:00, 14:00) UTC November 2, within the earlier 18:00 cutoff. Native timestamp binary search located the first block at or after 09:00 as 96603029 and the first block at or after 14:00 as 96615029; log scans cover blocks 96603029–96615028 in contiguous leaves. Three exchange surfaces were queried for the OrderFilled topic with the wallet in both indexed maker and taker positions. Successful publicnode responses found seven Michigan and three Korea own aggregate orders; their counterparty leg echoes are not additional orders. The correct third exchange `0xe3333700ca9d93003f00f0f71f8515005f6c00aa` returned none. The first dRPC attempts for that surface failed with a free-plan historical-range error; successful publicnode retries supersede them. An earlier provisional scan used a wrong third address and is not evidence. Logs and exact filters are in `raw/chain_log.json` and `raw/extra_publicnode_log.json`.

Separately, 20 successful publicnode leaves queried the CTF ERC-1155 TransferSingle and TransferBatch events with the wallet in both sender and recipient indexed roles. The Michigan YES results were exactly the seven receipt transfers above; no Michigan NO transfer appeared. YES `balanceOf` at the block before 09:00 and immediately before the first buy returned zero; immediately after the final sale and at the block before 14:00 it returned 742 microshares. Michigan NO balance returned zero at all four anchors. Historical CTF code calls succeeded; a positive first-buy transition to 1,250 shares checks that the archive provider honored at least these historical tags. Publicnode archive balance calls failed with a personal-token requirement; dRPC returned the retained historical results. The `ctf_log.json`, `ctf_single_*`, `ctf_batch_*`, `drpc_balance_*`, `no_balance_*`, `extra_log.json` and `no_balance_log.json` preserve queries, responses and hashes. This is a bounded check of registered exchange and CTF surfaces; unseen alternate contracts, other assets, off-chain positions and payments are outside the claim. API ASC and DESC activity pulls each returned 20 records with the same transaction hashes, and the wallet/market `takerOnly=false` pull returned the seven rows, but API agreement alone is not treated as proof of completeness.

The lead is a new, source-reproducible near-round-trip and cash-flow observation, not an established campaign-funding, artificial-price or insider-trading story. No independent reviewer or blinded reconstruction was performed in this lane. No outreach, trading, publication, payment or account changes occurred.