Ledger 61340172
2026-02-22T00:01:24Z
Nine minutes earlier, with the ladder in place.
Back to KeelHistorical reading · not live
The February USTRY backtest.
Every UTC day of USTRY/USDC trading in February 2026, read from Horizon mainnet: what the trade stream can show about depth, and what it cannot.
The month in four numbers
Verdict for the month: unevaluated. Not safe and not unsafe. No single trade leg crossed even the smallest rung, so the stream never measured the depth it would take.
- trades in 28 UTC days
- 13,547
- 375,320.84 USDC traded, no genuine-trade rule applied
- largest move by one leg
- 0.392%
- On 22 Feb. The smallest rung Keel measures is 2%.
- days with a causal bound
- 0
- A bound one leg is known to have crossed, at any rung.
- days with an assuming bound
- 2
- Holds only if the book did not change between trades; the largest is 0.0000084 USDC.
Movement within a trade leg
Largest observed daily price span · USTRY / USDC
Observed movementIncident · 22 Feb× No within-leg observation
Day by day.
One row per UTC day, as the CSV carries it. Prices in USDC per USTRY to four decimals; hover a figure for its exact value.
| Day | Trades | Volume · USDC | Low | High | Largest move, one leg | Largest move, between legs | Assuming bound · USDC | Findings on that bound |
|---|---|---|---|---|---|---|---|---|
| 2026-02-01 | 239 | 16.08 | 1.0556 | 1.0565 | 0.0000999% | 0.09% | None | None |
| 2026-02-02 | 280 | 142.67 | 1.0556 | 1.0599 | 0.00125% | 0.31% | None | None |
| 2026-02-03 | 275 | 44.95 | 1.0553 | 1.0588 | 0.0444% | 0.189% | None | None |
| 2026-02-04 | 224 | 158.12 | 1.0559 | 1.0569 | 0.000246% | 0.09% | None | None |
| 2026-02-05 | 435 | 635.10 | 1.0555 | 1.0574 | 0.0394% | 0.18% | None | None |
| 2026-02-06 | 481 | 680.28 | 1.0555 | 1.0575 | 0.000255% | 0.18% | None | None |
| 2026-02-07 | 182 | 22.98 | 1.0558 | 1.0573 | 0.00105% | 0.137% | None | None |
| 2026-02-08 | 146 | 10,359.75 | 1.0560 | 1.0591 | 0.157% | 0.15% | None | None |
| 2026-02-09 | 188 | 8.99 | 1.0561 | 1.0573 | 0.000398% | 0.113% | None | None |
| 2026-02-10 | 183 | 98.81 | 1.0564 | 1.1111 | 0.00205% | 5.17% | ≤ 0.000001for a 5% move · 399 s gap | Thin depth at 5% |
| 2026-02-11 | 189 | 15.11 | 1.0565 | 1.0575 | No within-leg observation | 0.0948% | None | None |
| 2026-02-12 | 187 | 11.74 | 1.0561 | 1.0581 | No within-leg observation | 0.18% | None | None |
| 2026-02-13 | 191 | 167.67 | 1.0566 | 1.0579 | No within-leg observation | 0.095% | None | None |
| 2026-02-14 | 183 | 14.85 | 1.0567 | 1.0577 | No within-leg observation | 0.09% | None | None |
| 2026-02-15 | 275 | 28.11 | 1.0568 | 1.0578 | 0.000118% | 0.09% | None | None |
| 2026-02-16 | 144 | 61.87 | 1.0569 | 1.0579 | No within-leg observation | 0.09% | None | None |
| 2026-02-17 | 110 | 4.97 | 1.0570 | 1.0580 | No within-leg observation | 0.09% | None | None |
| 2026-02-18 | 148 | 6.35 | 1.0570 | 1.0580 | No within-leg observation | 0.09% | None | None |
| 2026-02-19 | 98 | 27.36 | 1.0572 | 1.0582 | 0.00000946% | 0.09% | None | None |
| 2026-02-20 | 150 | 140.57 | 1.0573 | 1.0588 | No within-leg observation | 0.136% | None | None |
| 2026-02-21 | 93 | 182.13 | 1.0574 | 1.0584 | No within-leg observation | 0.09% | None | None |
| 2026-02-22Incident | 2,715 | 326,472.18 | 0.9868 | 106.7373 | 0.392% | 9,994.06% | ≤ 0.0000084for a 50% move · 36 s gap | Thin depth at 5%Cheap to move the price |
| 2026-02-23 | 3,295 | 567.78 | 1.0459 | 1.0685 | 0.0221% | 1.19% | None | None |
| 2026-02-24 | 2,039 | 285.01 | 1.0440 | 1.0626 | 0.000122% | 1.75% | None | None |
| 2026-02-25 | 427 | 34,774.41 | 1.0550 | 1.0592 | 0.21% | 0.391% | None | None |
| 2026-02-26 | 174 | 365.12 | 1.0578 | 1.0588 | 0.0000953% | 0.0975% | None | None |
| 2026-02-27 | 172 | 9.53 | 1.0579 | 1.0589 | 0.000672% | 0.09% | None | None |
| 2026-02-28 | 324 | 18.36 | 1.0580 | 1.0589 | 0.000382% | 0.09% | None | None |
- — no leg with more than one fill that day. It is an absent observation, not a zero move.
- None no qualifying move that day. A bound is never carried forward from an earlier day.
- Assuming bound the most that could have been traded at that rung, if nothing was posted or cancelled in the gap between two trades.
Three ledgers, read by the engine.
The daily rows above come from the trade stream. These three are full risk results the engine rebuilt from the operation stream at single ledgers on the incident morning. Each opens the engine’s own answer, gaps included: a rebuilt book is too thin, never too deep, so its depth is a lower bound.
Ledger 61340262
2026-02-22T00:10:15Z
Inside the 79 second window after the ladder was withdrawn.
Ledger 61340263
2026-02-22T00:10:21Z
The next ledger, still inside that window.
How to read a bound.
Within one leg
One taker, one direction, filling from the best price outward. The span between the first and last fill was crossed by that leg, so the bound assumes nothing.
0 of 28 daysBetween two legs
From one leg’s last fill to the next leg’s first. Offers can be posted and cancelled in that gap, and a trade stream records none of it, so the bound holds only if the book stood still.
2 of 28 daysReproduce the rows
go run ./cmd/keel backtest \
-pairs scripts/record-pairs.example.json \
-from 2026-02-01 -to 2026-03-01 -mark 2026-02-22 \
-from-ledger 60977383 -out docs/evidencesRun from the engine repository. The CSV and the evidence notes linked above are the files this produced.
Evidence filed since
- The daily book rebuild, beyond the USTRY makerRead 17 September 2026. Whether the 00:10 UTC withdrawal-and-repost seen on the USTRY book is something other makers do, measured over the same February window.
- The first pass of the trade-derived metricsRead 18 September 2026. What volume-to-supply and time-since-last-genuine-trade cost to compute on public Horizon, and what they answer across the demonstration set.