What depth stands behind this price?

XAG against USDC · GB4I3O…YCLXAG

How much can this asset safely back?

The ceiling is the lower of two independent limits. Which one binds is the part a lender acts on, so both are shown.

Low
Medium
High
Critical — this asset

Criticalpartial confidence

A band marked partial is a floor. At least one high-severity check could not be evaluated, so the real band can only be worse than the one shown, never better.

0 USDC

Limited by manipulation cost

Liquidation term: 18.49 USDC (shortened; full value 18.491125167695211554811215537182337841175259251182 USDC)
Sell-side depth at the liquidation delta, times the haircut.
Manipulation termbinds: 0 USDC
Order-book-only manipulation cost at the critical delta, times the safety margin.

These bands come from thresholds that were chosen, not calibrated

The thresholds were chosen based on the magnitude of the Blend incident of February 2026 and on conservative judgement, not calibrated against a set of incidents. Every flag is reported separately so that consumers can apply their own thresholds.

Every threshold, as the engine serves it

Where this number comes from

Source: GET /asset/{assetId}/depth → maxSafeCollateral, maxSafeCollateralLiquidation, maxSafeCollateralManipulation
What it is: Awaiting methodology copy.
What it does not account for: Awaiting methodology copy.

What this computation could not do

maxReachablePrice and costToMaxReachablePrice are null because an active pool is present: under a constant product curve the price tends to infinity as the base reserve tends to zero, so every target is reachable and a highest price has no meaning

Served by the engine with this response and shown as written.

Where does the price come from?

Two venues can disagree. Both readings are shown, not only the one that won.

Mid price: 0.00014518 USDC (shortened; full value 0.0001451858784858 USDC)
Source: AMM pool spot
Spread: 68,877,118.7962 % (shortened; full value 68877118.79621994426204696766373919 %)
Undefined unless the price came from a two-sided book
Pool spot price: 0.00014518 USDC (shortened; full value 0.0001451858784858 USDC)
Reported whenever an active pool exists, whichever source won
Divergence between the two: 34,438,562.7139 % (shortened; full value 34438562.71394314021069199640508995 %)
PRICE_SOURCE_CONFLICT is triggered: the two sources disagree past the threshold and the pool was taken

Where this number comes from

Source: GET /asset/{assetId}/depth → midPrice, priceSource, spreadPct, poolSpotPrice, priceDivergencePct
What it is: Awaiting methodology copy.
What it does not account for: Awaiting methodology copy.

What volume can it absorb before the price moves?

Three rungs at 2, 5 and 10 per cent, in each direction. The buy side matters for oracle manipulation and the sell side for liquidation.

SPREAD_EXTREME is triggered, so this ladder is not meaningful

The spread is past the threshold at which the engine says the mid price, and every figure derived from it including these rungs, stops describing an executable market.

SDEX
AMM
Sell side, composition not reported

2% from mid

Buy side

7.21 USDC

Sell side

7.24 USDC

5% from mid

Buy side

17.90 USDC

Sell side

18.24 USDC

10% from mid

Buy side

35.38 USDC

Sell side

36.98 USDC

Bars share one scale. The venue split decomposes the buy side; the sell side is drawn as an outline because its composition is not reported.

Where this number comes from

Source: GET /asset/{assetId}/depth → depth[]
What it is: Awaiting methodology copy.
What it does not account for: Awaiting methodology copy.

What would it cost to move the price?

Read cost together with reachability: a figure on an unreachable rung says how far the book goes, not what the move costs.

Every target price below is measured from the same unreliable mid

SPREAD_EXTREME is triggered. A target price is the mid price moved by the delta, so when the mid sits between quotes that are far apart, the targets describe a market that is not there. Reachability and the furthest price the book reaches are the readings that still mean something here.

Cost to move the price by each delta, through all venues and through the order book alone, with whether the target is reachable at all.
Move	Target price	Cost, all venues	Cost, order book only
0.5%	0.000217 USDC (shortened; full value 0.0002177788177287 USDC)	162.94 USDC (shortened; full value 162.9423851330072171165893837138 USDC)	free to reach
1%	0.000290 USDC (shortened; full value 0.0002903717569716 USDC)	300.30 USDC (shortened; full value 300.3091700807444809679845905988 USDC)	free to reach
10%	0.001597 USDC (shortened; full value 0.0015970446633438 USDC)	1,679.57 USDC (shortened; full value 1679.577231113130117628270842899 USDC)	free to reach
100%	0.014663 USDC (shortened; full value 0.0146637737270658 USDC)	6,561.25 USDC (shortened; full value 6561.2545894014086755803131621557 USDC)	free to reach

Reachability on the all-venues ladder is unconditionally true whenever an active pool exists, because a constant product curve has no upper price bound. The order-book-only column is the one that answers whether a target is attainable, and it is the one Keel itself uses.

How far the book actually goes

Furthest price the book reaches: not computed
Null by structure, not by failure: an active pool means every target is reachable and a highest price has no meaning
Cost to walk it that far: not computed
Null for the same reason as the price above

Against genuine volume

Not reported for this asset. Without it there is no comparison between the cost of an attack and the genuine trading it would have to hide inside.

Where this number comes from

Source: GET /asset/{assetId}/depth → manipulationCostCombined[], manipulationCostOrderbookOnly[], oracleResistance
What it is: Awaiting methodology copy.
What it does not account for: Awaiting methodology copy.

Which checks fired, and which could not run?

An unevaluated check is not a check that came back clear. The two never share a treatment.

Triggered(7)

MANIPULATION_CHEAP
MANIPULATION_RATIO_LOW
PRICE_SOURCE_CONFLICT
SPREAD_EXTREME
HOLDER_CONCENTRATION_EXTREME
THIN_DEPTH_5PCT
HOLDER_CONCENTRATION_HIGH

Could not be evaluated(3)

NO_GENUINE_TRADE_30D
NO_GENUINE_TRADE_7D
WASH_TRADE_SUSPECTED

Where this number comes from

Source: GET /asset/{assetId}/depth → flags[], unevaluatedFlags[], bandConfidence
What it is: Awaiting methodology copy.
What it does not account for: Awaiting methodology copy.

Who holds it, and is it genuinely traded?

Concentration and wash trading bound how much of the depth above is real.

Largest holder: 89.0741 % (shortened; full value 89.07413303082708356530243253 %)
Top ten holders: 98.5935 % (shortened; full value 98.59350391178447912611475057 %)
Concentration index: 7,981.70 (shortened; full value 7981.7021001826631229899811217928966564672062866967130639)
Herfindahl–Hirschman, over the holder distribution
Volume excluded as not genuine: not computed
Of 30 day volume. A high share indicates suspected wash trading
Volume to supply, 30 day: not computed
Last genuine trade: not computed
No genuine trade was found in the window

Where this number comes from

Source: GET /asset/{assetId}/depth → holderTop1Pct, holderTop10Pct, holderHhi, volumeToSupply, lastGenuineTrade, tradesExcludedPct
What it is: Awaiting methodology copy.
What it does not account for: Awaiting methodology copy.

How has this moved over time?

A single reading cannot separate a thin asset from one that just got thin. This is a stored series, not the historical replay that health reports as unavailable — those are different endpoints.

WindowLast 24 hours Last 7 days Last 30 days

ResolutionHourly Daily

SourceHorizon Hubble Offers implied Trades implied

2 readings from horizon, with no gaps reported. One source per chart.

Band at each reading

Critical: 2 of 2

Which checks were firing

Whether a finding has been there all along or started recently. Only checks that fired at least once appear.

MANIPULATION_CHEAP

throughout

MANIPULATION_RATIO_LOW

throughout

PRICE_SOURCE_CONFLICT

throughout

SPREAD_EXTREME

throughout

HOLDER_CONCENTRATION_EXTREME

throughout

THIN_DEPTH_5PCT

throughout

HOLDER_CONCENTRATION_HIGH

throughout

Price

Lowest: 0.00014376 USDC (shortened; full value 0.0001437635657709 USDC)
Highest: 0.00014518 USDC (shortened; full value 0.0001451858784858 USDC)

2026-09-17T23:46:03Z — 2026-09-18T12:46:03Z

Mid price: 0.00014518 USDC (shortened; full value 0.0001451858784858 USDC)

Buy-side depth

What an order can absorb before the price moves against a buyer. The rungs are nested, so one scale covers all three, and the lines darken as the rung widens.

Lowest: 7.17 USDC (shortened; full value 7.1785694263304900717184302355 USDC)
Highest: 35.38 USDC (shortened; full value 35.3869260161322001195915968919 USDC)

2026-09-17T23:46:03Z — 2026-09-18T12:46:03ZOne scale across every line on this chart

2% from mid: 7.21 USDC (shortened; full value 7.2142122259954867500014788283 USDC)
5% from mid: 17.90 USDC (shortened; full value 17.9041891219707365058082521133 USDC)
10% from mid: 35.38 USDC (shortened; full value 35.3869260161322001195915968919 USDC)

Cost to move the price by half

The series carries one manipulation rung, at the half per cent delta the engine treats as critical. A reading the engine could not produce breaks the line rather than sitting at zero.

Lowest: 0 USDC
Highest: 0 USDC

2026-09-17T23:46:03Z — 2026-09-18T12:46:03Z

Manipulation cost, 0.5%: 0 USDC

Collateral ceiling

Lowest: 0 USDC
Highest: 0 USDC

2026-09-17T23:46:03Z — 2026-09-18T12:46:03Z

Max safe collateral: 0 USDC

Where this number comes from

Source: GET /asset/{assetId}/history → points[], gaps[], dataSource
What it is: Awaiting methodology copy.
What it does not account for: Awaiting methodology copy.

What was this computed from?

Ledger: 64,490,137
Closed at 2026-09-18T13:01:03Z
Computed at: 2026-09-18T13:01:12.725666Z
Data source: horizon
Methodology: 1.0.8-draft
The version that produced these figures, not the one this page was built against