THE DOSSIER · one of 50

A flash-route address caught in a failure loop

0x9cad98bde3e40d10fec68a6d6de179f53b2fcdce339519db9599fb8fe2b7f6c2
2,376 transactions in the window · 2,072 stopped early · browse every one →

For roughly ten and a half hours, this address repeatedly assembled multi-hop, flash-funded routes—and most of them failed. Across its complete frozen-history record, 2,072 of 2,376 eligible transactions failed (87.2%). Only 72 of the 2,368 transactions with a whole-transaction interpretation were assigned the execute-swap-route behavior. The rest were understood only as connected call sequences, without a stronger claim about the transaction’s purpose.

The evidence is consistent with repeated automated route attempts, but it does not establish who operated the address, whether it was a bot, or whether any route was profitable.

A rhythm of bursts and retries

The complete-history rhythm is strikingly mechanical. The median gap was 0 milliseconds, reflecting batched submissions; the mean gap was 15.8 seconds and the 90th-percentile gap was 32.0 seconds. The longest idle period was about 12 minutes 36 seconds. Activity peaked at 143 transactions in the minute beginning 17:01 UTC.

The longest consecutive failure run was 66 transactions. Activity persisted from 2026-07-19T13:02:05.393Z through 2026-07-19T23:28:31.072Z, with no meaningful change in the basic repertoire: the named behavior remained route execution, while most surrounding transactions stayed at the level of linked calls.

What the routes attempted

The complete-history repertoire contains 28 reusable route patterns and 769 distinct ordered call shapes. Their shared structure is temporary liquidity followed by several venue or router legs, balance handling, and flash-loan return or repayment steps. The authored explanations mention Cetus, Bluefin, pool adapters, threshold transfers, and router swaps, but they explicitly stop short of asserting an exchange outcome.

One concrete example is ATh1r9qq. Its nine-call route opened short-lived flash credit, passed SUI through a WAL/SUI leg, used a returned-output router swap and a threshold helper, then closed and repaid the flash position. The transaction reported no execution error, but its own detail says that structural completion does not establish the economic size or outcome of the swaps.

The failure mode is equally concrete. In 2vVQw5S3, the route moved through NAVX, BUCK, USDC, and SUI handling before stopping with InsufficientCoinBalance. Application effects were rolled back, so the reached calls prove an attempted route—not a completed exchange. The transaction’s reported sender movement was only −0.000837566 SUI in the holding record, with the same amount appearing separately as gas.

The latest 500 rows are even harsher

The detailed ledger and prose stream is capped at the latest 500 classifiable transactions, covering 2026-07-19T20:48:30.288Z to 2026-07-19T23:28:31.072Z. In that narrower window, 429 of 500 transactions failed (85.8%). Only 10 received a complete purpose-level route explanation; the other 490 were described as connected call compositions without a whole-transaction purpose claim.

Among the 429 failures, no specific abort location was recorded for 424. Across the complete history, the corresponding figure was 2,058 of 2,072 failures (99.3%). The named sites were rare: five failures at bank::burn_btoken, five at price_oracle::get_valid_price, and four at pyth::check_price_is_fresh.

What actually moved

The complete-history movement record is narrow and bounded:

  • Gas: 2.285150772 SUI out.
  • Sender holding record: 2.267991345 SUI out.
  • Sender holding record: 2.037224 DEEP in across 18 transactions.
  • Sender holding record: 0.000337 USDC in across one transaction.

Gas and holding records are separate evidence categories and must not be added together. These movements also do not establish profit, an exchange rate, or the outcome of any individual route. Every transaction used the sender itself as gas owner; none was sponsored.

This address is therefore best read as a persistent route-attempt identity: highly repetitive, bursty, and unusually failure-heavy, with a small set of structurally understood flash-funded routes inside a much larger field of unresolved call compositions. The evidence does not identify the operator or establish that the attempted swaps settled economically.

Further reading

  • Address: 0x9cad98bde3e40d10fec68a6d6de179f53b2fcdce339519db9599fb8fe2b7f6c2
  • Successful structural route example: ATh1r9qq
  • Insufficient-balance failure: 2vVQw5S3
  • Multi-hop route with SUI/USDC/SCA legs: 6CYp1zro
  • Representative pattern: sui/execute-swap-route/packed-e118-c6822ce160a6-40b2f87809c7/v1

Dossiers are AI-authored readings of one address's stored window. They describe observed behavior — never operators, identity, or profit the evidence doesn't establish.

everything this address did → all dossiers the shelf