Measured on Base from onchain state at 2026-09-07 12:27 UTC, and frozen at that block. Liquidity on these pools is placed by hand and moves within hours, so depth read a week from now will differ; the monitor carries the current numbers. Method and code below; every input is listed with where it came from. No lending market has listed these tokens. So this is a feasibility question rather than a measurement of anyone's live exposure: what could be got out of, at what cost, and what depth a market of a given size would need. Market parameters are assumed and swept across a range, because there is no market contract to read them from. ### When the feed is stale, and whether that is a weekend The reading most people arrive with is that a 24/5 oracle is a weekend problem: markets shut, the feed stops, nothing can be liquidated until Monday. It is the obvious inference. It is not what the windows say. Every completed window of six hours or more, split by when it fell. Six hours is the bar because a US regular session is six and a half, so it is the shortest gap that cannot be explained by a quiet afternoon. | | windows 6h+ | weekend | not a weekend | of those, in a live session | |---|---:|---:|---:|---:| | AAPLc | 32 | 7 | 25 | 20 | | AMZNc | 29 | 7 | 22 | 17 | | GOOGLc | 31 | 8 | 23 | 19 | | METAc | 31 | 8 | 23 | 13 | | MSFTc | 27 | 7 | 20 | 13 | | MSTRc | 5 | 4 | 1 | 0 | | NVDAc | 25 | 6 | 19 | 12 | | SNDKc | 4 | 4 | 0 | 0 | | SPCXc | 15 | 5 | 10 | 6 | | TSLAc | 22 | 6 | 16 | 11 | **159 of 221** windows across these 10 are not weekends, and 111 of those 159 overlapped a trading session that was open at the time. Across all 13 feeds, including the 3 with no pool, it is 175 of 251. That distinction decides what a fix would have to be. A weekend problem is fixed by extending hours. A deviation trigger with no off-hours heartbeat is not, because longer sessions do not add a heartbeat. The thin and clipped flags are deliberately not applied here. Both qualify a divergence, which is measured against a venue. This counts latency, which is a fact about the feed alone. ### Executable depth, by band Walked from the pool's initialised ticks rather than taken from total value locked or from reported volume. Each figure is the binding side: whichever half of the band runs out first. | | ±1% | ±2% | ±5% | as of | |---|---:|---:|---:|---| | AAPLc | $177,458 | $319,011 | $541,798 | 2026-09-07 12:22 UTC | | AMZNc | $35,387 | $38,021 | $43,324 | 2026-09-07 12:22 UTC | | GOOGLc | $315,727 | $479,432 | $681,041 | 2026-09-07 12:22 UTC | | METAc | $281,323 | $344,645 | $506,825 | 2026-09-07 12:22 UTC | | MSFTc | $3,131 | $5,039 | $12,957 | 2026-09-07 12:22 UTC | | MSTRc | $6,952 | $7,425 | $8,165 | 2026-09-07 12:22 UTC | | NVDAc | $394,124 | $510,781 | $797,945 | 2026-09-07 12:22 UTC | | SNDKc | $8,129 | $18,705 | $62,478 | 2026-09-07 12:22 UTC | | SPCXc | $21,482 | $30,055 | $69,731 | 2026-09-07 12:22 UTC | | TSLAc | $13,174 | $25,870 | $60,994 | 2026-09-07 12:24 UTC | ### Executable depth over time NVDAc across the weekend of 5 September 2026, ±5% band, lowest reading in each interval. The minimum rather than the average, because the question is whether the book went thin and for how long, and an average fills a trough back in with the hours either side of it. ``` Fri 00:00 ######################### $516,142 Fri 02:00 ######################## $488,096 Fri 04:00 ####################### $475,843 Fri 06:00 ################### $380,672 Fri 08:00 ###################### $440,647 Fri 10:00 #################### $413,067 Fri 12:00 ##################### $420,637 Fri 14:00 #################### $415,638 Fri 16:00 ######################### $508,365 Fri 18:00 ######################### $516,897 Fri 20:00 ##################################### $744,022 Fri 22:00 ##################################### $756,086 Sat 00:00 ###################################### $775,361 Sat 02:00 ####################################### $783,012 Sat 04:00 #################################### $720,111 Sat 06:00 #################################### $721,063 Sat 08:00 #################################### $727,788 Sat 10:00 #################################### $731,769 Sat 12:00 ################################## $691,353 Sat 14:00 ####################### $472,296 Sat 16:00 # $27,468 Sat 18:00 ############################## $603,326 Sat 20:00 ################################## $688,401 Sat 22:00 ################################## $684,600 Sun 00:00 ##### $109,746 Sun 02:00 ## $52,779 Sun 04:00 ################################ $649,847 Sun 06:00 ############################# $591,030 Sun 08:00 ############################ $559,558 Sun 10:00 ########################### $554,167 Sun 12:00 ################################ $648,663 Sun 14:00 ################################### $709,689 Sun 16:00 ####################################### $794,353 Sun 18:00 ####################################### $780,169 Sun 20:00 ############################# $593,386 Sun 22:00 ############################## $613,892 Mon 00:00 ############################### $633,643 Mon 02:00 ########################## $536,737 Mon 04:00 ############################## $601,443 Mon 06:00 ################# $357,110 Mon 08:00 ####### $145,906 Mon 10:00 ############################### $633,178 Mon 12:00 ######################################## $796,243 Mon 14:00 not observed Mon 16:00 not observed Mon 18:00 not observed Mon 20:00 not observed Mon 22:00 not observed source: stalewatch.com/research/exit-capacity | Aerodrome Slipstream tick book, read onchain ``` Peak $796,243, trough $27,468, over 2026-09-04 00:00 UTC to 2026-09-08 00:00 UTC. 5 of 48 intervals carry no reading and are left blank rather than filled in. ### What an exit costs The price impact of selling that much of the token, simulated against the tick liquidity actually read, with the pool fee included. A liquidation is an exit, so this is the cost the bonus has to cover. "beyond the read" means the size runs past the range of the book we fetched. We did not observe that liquidity, so its cost is withheld rather than extrapolated. | | $100,000 | $250,000 | $500,000 | $1,000,000 | $2,000,000 | |---|---:|---:|---:|---:|---:| | AAPLc | 21 bps | 71 bps | 156 bps | beyond the read | beyond the read | | AMZNc | beyond the read | beyond the read | beyond the read | beyond the read | beyond the read | | GOOGLc | 11 bps | 34 bps | 75 bps | beyond the read | beyond the read | | METAc | 7 bps | 14 bps | 112 bps | beyond the read | beyond the read | | MSFTc | beyond the read | beyond the read | beyond the read | beyond the read | beyond the read | | MSTRc | beyond the read | beyond the read | beyond the read | beyond the read | beyond the read | | NVDAc | 8 bps | 32 bps | 66 bps | 138 bps | beyond the read | | SNDKc | beyond the read | beyond the read | beyond the read | beyond the read | beyond the read | | SPCXc | beyond the read | beyond the read | beyond the read | beyond the read | beyond the read | | TSLAc | beyond the read | beyond the read | beyond the read | beyond the read | beyond the read | ### How far the published value drifts before it is corrected Measured, not taken from the documented update threshold. Each figure is how far the oracle feed moved when it resumed after a latency window: the distance the market had already travelled, revealed at the moment the feed agreed. A liquidator is paid in collateral valued at the published number and sells at the market number, so this distance comes out of the bonus before the book is reached. The counts are small and the percentiles inherit that. With this many observations a p99 is the largest one seen, not a p99. | | windows | p50 | p95 | p99 | |---|---:|---:|---:|---:| | AAPLc | 49 | 7 bps | 59 bps | 465 bps | | AMZNc | 5 | 9 bps | 750 bps | 750 bps | | GOOGLc | 43 | 14 bps | 497 bps | 746 bps | | METAc | 61 | 8 bps | 72 bps | 88 bps | | MSFTc | 1 | 790 bps | 790 bps | 790 bps | | MSTRc | 3 | 92 bps | 928 bps | 928 bps | | NVDAc | 69 | 12 bps | 118 bps | 627 bps | | SNDKc | 2 | 27 bps | 449 bps | 449 bps | | SPCXc | 5 | 792 bps | 955 bps | 955 bps | | TSLAc | 5 | 233 bps | 930 bps | 930 bps | ### The largest position that could be exited profitably A liquidation pays while the bonus exceeds the cost of getting out: worth doing while bonus > price impact + gas + oracle error Below, for NVDAc, against the book read 2026-09-07 12:22 UTC and the error band measured above over 69 windows. The bonus is swept because there is no market setting one. | bonus | oracle error | budget for impact | largest clip | absorbed in a response window | |---:|---|---:|---:|---:| | 500 bps | p50 (12 bps) | 488 bps | $1,596,495 | $1,596,495 | | 500 bps | p95 (118 bps) | 382 bps | $1,596,495 | $1,596,495 | | 500 bps | p99 (627 bps) | -127 bps | nothing pays | not measured | | 750 bps | p50 (12 bps) | 738 bps | $1,596,495 | $1,596,495 | | 750 bps | p95 (118 bps) | 632 bps | $1,596,495 | $1,596,495 | | 750 bps | p99 (627 bps) | 123 bps | $918,433 | $918,433 | | 1000 bps | p50 (12 bps) | 988 bps | $1,596,495 | $1,596,495 | | 1000 bps | p95 (118 bps) | 882 bps | $1,596,495 | $1,596,495 | | 1000 bps | p99 (627 bps) | 373 bps | $1,596,495 | $1,596,495 | | 1500 bps | p50 (12 bps) | 1488 bps | $1,596,495 | $1,596,495 | | 1500 bps | p95 (118 bps) | 1382 bps | $1,596,495 | $1,596,495 | | 1500 bps | p99 (627 bps) | 873 bps | $1,596,495 | $1,596,495 | The clip size stops growing well before the bonus runs out. It is capped by the book itself: past roughly $1,600,000 there is no more of the quote asset inside the range we read. That ceiling, rather than the bonus, is what sets capacity here. ### The question in the other direction A prospective lender does not ask how deep the pool is. They ask how large a market this could carry, and what would have to change. Below: the largest Friday-to-Monday gap a market of a given size could take at a given loan-to-value before what must be liquidated exceeds what NVDAc can absorb today. The capacity used is $1,596,495, from the 500 bps bonus and p95 oracle error row above. | collateral | 50% LTV | 60% LTV | 70% LTV | 75% LTV | 80% LTV | |---:|---:|---:|---:|---:|---:| | $5,000,000 | any | any | 94% | 89% | 84% | | $25,000,000 | 63% | 53% | 43% | 38% | 33% | | $100,000,000 | 53% | 43% | 33% | 28% | 23% | Read down a column: the same book carries a smaller shock as the market grows. Read across a row: a lower loan-to-value buys headroom that depth does not have to provide. Across the gap range swept here, nothing goes underwater at any of these loan-to-values: 80% leaves 20% of headroom and the widest gap swept is 12%. On this model the weekend gap is not the binding constraint. What binds is whether positions already close to their threshold can be exited at all, which is the capacity table above rather than this one. The gap shock is a swept range and not yet a measurement: the underlying distribution has not been computed. Read the rows as a sensitivity rather than as a probability. Source: not measured (not yet computed). ### A second look at the weekend, from before this series existed The depth series above starts when it was built, so its first weekend is its only weekend. Every stored swap also carries the active liquidity at the traded tick, which reaches back across earlier weekends. This is **not** executable depth. Active liquidity is what is live inside the current tick range only, and it is weighted by the trades that happened rather than by time, so the two columns describe different populations. It cannot be compared with the depth table above. | | weekday trades | weekend trades | weekend median vs weekday | |---|---:|---:|---:| | AAPLc | 75,128 | 30,087 | 2.1x | | AMZNc | 7,853 | 19,475 | 6.6x | | GOOGLc | 107,857 | 33,698 | 3.7x | | METAc | 110,627 | 20,986 | 4.9x | | MSFTc | 3,609 | 14,317 | 8.4x | | MSTRc | 4,316 | 3,201 | 744.8x | | NVDAc | 185,554 | 46,908 | 2.1x | | SNDKc | 1,325 | 2,773 | 4.0x | | SPCXc | 13,626 | 12,636 | 1093.4x | | TSLAc | 3,969 | 2,878 | 2833.0x | On this measure liquidity behind a trade is higher at the weekend, not lower, in every pool with a venue. That runs against the expectation the exercise started from. It is a weaker measure than the depth series and the populations differ, so it is offered as a reason to take the direct measurement rather than as a finding. ### What this assumes, and where it would be wrong Everything above rests on choices that are defensible and are not the only ones available. They are listed here rather than left for a reader to find, because a model whose weak points have to be discovered is a model nobody can use. **One venue.** Depth is the Aerodrome Slipstream pool and nothing else. A liquidator could route to another Base venue, bridge out, or hedge on a centralised venue or a perp. Those routes exist and are not measured here. They are also not atomic: they carry inventory risk across exactly the interval in which the position is underwater, which is the interval this is about. Read these figures as the atomic onchain exit, and therefore as a floor rather than a ceiling. **The book that was actually read.** The walk fetches the initialised ticks around the current price, which here spans a 29% fall. An exhausted result means the sale ran the price down that far without filling. It is not a claim that nothing exists beyond it. **Oracle drift charged as a cost, every time.** In practice the drift is two-sided: about as often the stale price favours the liquidator. A liquidator who prices it will decline the adverse case, and a declined liquidation is exactly the outcome this is about, so the adverse side is the one that decides feasibility. Anyone modelling expected profit rather than whether a position clears should use the signed distribution instead. **Every borrower at the same loan-to-value.** The survivable-gap table treats a whole market as sitting at the stated loan-to-value. Real books hold a distribution, so the true answer depends on how much sits near the threshold, which cannot be read from outside. **Assumed parameters, stated.** Close factor 50%, 3 clips inside the response window, and gas charged against a $100,000 reference clip. None of these is readable anywhere, because no market has set them. **Small samples.** Roughly three weeks of feed history and days of depth history. The percentiles inherit that, and every table carries its own denominator. **What would change the conclusion.** If a liquidator can reliably exit off this pool at size, the capacity figures are a floor and the real number is higher. And if the oracle gains an off-hours heartbeat, the drift distribution changes and most of this stops applying. ### Method Depth is walked from the pool's initialised ticks: every tick in the band is a step change in liquidity and each sub-range is integrated at its own. Not total value locked, not reported volume. Price impact is the same walk run backwards, with the pool fee charged on the way in. A band or a clip reaching past the range of ticks actually read is refused rather than answered. An unread tick and a tick with nothing on it are indistinguishable once they reach the walk, and the resulting error runs toward reporting a thin book as a deep one. Oracle error is measured from how far each feed moved when it resumed after a latency window. No market calendar, session label or time of day enters it, so it qualifies latency independently rather than restating it. Market parameters are assumed and swept: no lending market has listed these tokens, so there is nothing to read them from. Every table above is a surface, not a point on one. Arithmetic is integer throughout. Values are reproducible from public chain state on Base. An em dash means a value we could not source. It is never a zero and never a carried-forward reading.