At 19:31 UTC, Sui activity looked less like a crowd than a tightly timed order-book session. The minute contained 1,344 eligible transactions. One sender, 0xcde6db…6c0e04, supplied 548 of them—40.8%—and was also the gas owner for those transactions. The minute's median gap between transactions was 0 ms, with a mean of 44.5 ms and a longest idle period of only 457 ms. That cadence is consistent with automated execution, but the evidence does not establish who controlled the address or whether it was a bot.
Among the 1,339 transactions with an asserted behavior, 350 were order-cancellation attempts, 340 were limit-order placements, and 221 updated prices. Together they account for 911 of 1,339 transactions, or 68.0%. The leading reusable explanation was an owner-proof DeepBook flow: 344 cancellation attempts and 288 placement attempts used that composition. Another 144 transactions recorded campaign activity, while 75 updated prices through a dual-expiry flow.
The repetition becomes clearer in the final seconds. At 7Wpvc4gNGtLBn6FXhYcy1EJiZRGdMdSgzXEqMxfKQPGq, the sender generated an owner proof and passed it to a limit-order call for 1,050.9 NS at a stated resting price of 0.01532 SUI per NS. That proves the submitted order size and price input; it does not prove that the order entered the book or remained there. A nearby transaction, 7JivsCCBWugDNKux8a8FnMTYmWKEpdcsYzubRgQXpVTN, generated the same kind of proof and submitted a cancellation for an order with 1,044.1 NS of resting size at 0.01542 SUI per NS. The evidence describes an attempted cancellation and explicitly does not establish that the resting size traded.
Failures were present, but they were not the defining feature of the minute: 13 of 1,344 eligible transactions failed, or 1.0%. Twelve of those failures stopped at two paired repayment functions—six at repay_loan_base and six at repay_loan_quote—with one more stopping at reward distribution. The failed transaction CWKiiPWBQ2mrtn3rwmRFPEbhUQsySHeqckD44Ex2HmXe shows the mechanism. It opened a short-lived SUI credit, moved through SUI/CETUS, CETUS/NAVX, and NAVX/SUI route calls, then stopped at repay_loan_base with Move abort code 401. Earlier application effects rolled back, so the reached route calls do not prove that an exchange completed. The sender still paid 0.000387844 SUI in gas.
The minute also carried substantial boundary movement, but it should not be read as profit. Sender-boundary SUI holding evidence recorded 655.48952879 SUI in and 9,843.325845551 SUI out across 451 transactions, for a net of −9,187.836316761 SUI. A separate gas record showed 0.087115682 SUI in and 1.46724715 SUI out across 1,064 transactions. Gas and holding records can describe the same underlying charge, so these figures must not be added; there are no prices or cross-asset valuations here. Of the 1,344 transactions, 278 used a different gas owner, another reminder that the sender and payer are not necessarily the same on-chain identity.
The minute's central story is therefore repetition with a narrow failure seam: one heavily represented sender repeatedly submitted order-management calls at near-continuous cadence, while a smaller set of multi-step route attempts consistently encountered repayment guards. The records show what was attempted and where execution stopped; they do not establish a shared operator, completed trades, or economic benefit.
For a closer look, start with 7Wpvc4gNGtLBn6FXhYcy1EJiZRGdMdSgzXEqMxfKQPGq, compare it with the cancellation attempt 7JivsCCBWugDNKux8a8FnMTYmWKEpdcsYzubRgQXpVTN, and then follow the failed route CWKiiPWBQ2mrtn3rwmRFPEbhUQsySHeqckD44Ex2HmXe or sender 0xcde6dbe01902be1f200ff03dbbd149e586847be8cee15235f82750d9b06c0e04.