keelDashboard
February USTRY case study

Know how much a price can actually support.

Measure how much can trade on Stellar before the price moves, what it costs to move it, and how much collateral the market can support.

Open the dashboardSee the full response
Result
GA5ZSE…KZVN · Stellar

USDC / XLM

Volume the market can absorb before the price moves, and the collateral limit that follows from it.

SDEX 76.6% · AMM 23.4%
Horizon
LOWFull confidence

Executable depth XLM

RangeBuySell
±2%184,220.42176,905.22
±5% (selected)441,038.99428,771.00
±10%852,119.77839,004.55
Max safe collateral419,502.28XLM

Rounded to 2 decimals. Exact amounts in the sample response.

Reading live engine status…

Compare Stellar markets.

Executable depth, collateral capacity, and the findings that deserve a closer look. Amounts in XLM, rounded to two decimals.

LOWFull confidence
GA5ZSE…KZVN
5% buy depth
441,038.99XLM
Max safe collateral
419,502.28XLM
Price source
Order book
Triggered flags
None

USDC / XLM

No flags triggered. The reference price comes from the order book.

Every low-risk market
HIGHFull confidence
GDAHBQ…2FJ5
5% buy depth
2,710.99XLM
Max safe collateral
2,677.94XLM
Price source
Pool
Triggered flags
4

RWAX / XLM

4 flags triggered, including thin depth at 5%.

Every high-risk market
CRITICALFull confidence
GBTVRM…UJZE
5% buy depth
0.00XLM
Max safe collateral
Not available
Price source
None
Triggered flags
6

USTRY / XLM

6 flags triggered, including no executable price.

Every critical market

From Stellar liquidity to risk calculations.

Keel combines orderbook offers and pool reserves, then calculates depth, price targets, and collateral limits.

  • Input: SDEX, Orderbook liquidity. The engine absorbs asks to the target, feeding executable depth (fromSdex), manipulation cost (orderbook-only), safe collateral (manipulation term), flags & confidence (checks SPREAD_EXTREME).
  • Input: AMM pools, Pool reserves. The engine swaps along the pool curve, feeding executable depth (fromAmm), manipulation cost (combined), safe collateral (liquidation term), flags & confidence (checks PRICE_SOURCE_CONFLICT).
  • Input: Market observations, Trades & supporting data. The engine keeps genuine trades in the window, feeding manipulation cost (vs genuine volume), flags & confidence (checks WASH_TRADE_SUSPECTED).
ProcessKeel engineDepth simulation & risk rules
  • Executable depth
  • Manipulation cost
  • Safe collateral
  • Flags & confidence

Why this market is marked critical.

What triggered a finding, and what the available data could not establish.

CRITICAL · Partial confidence

196.08%between bid and ask

4 triggered · 6 not evaluated

The reference price is unreliable.

The best bid and the best ask sit almost two hundred per cent apart, so the midpoint between them is a number no one can trade at. Six more checks could not be evaluated at all, which is why this band is a floor rather than a verdict.

See the full finding

Cost to move the priceOrder book · USDC

+50%
0.00USDCReachable at zero cost
+100%
130.06USDCNot reachable; book exhausted
+1000%
130.06USDCNot reachable; book exhausted
+10000%
130.06USDCNot reachable; book exhausted

And moving it costs nothing.

Lifting this price by half takes no capital at all. Past that the book is exhausted, so the figures beside those rungs are where the market ran out — not the price of getting there.

See every rung

The February USTRY incident.

Twenty-eight UTC days of trading, read from Horizon, and what the trade stream can and cannot prove about depth.

Movement within a trade leg

Largest observed daily price span · USTRY / USDC

Historical observations
Observed trade-leg price movement in February 2026Daily maximum price movement, from zero to 0.4 percent. The largest observed movement occurs on February 22, the incident date. Missing observations are shown as gaps, not zeros. Exact values are available in the table below.0.4%0.3%0.2%0.1%0%Feb 22 · incidentFeb 1Feb 8Feb 15Feb 22Feb 28

Observed movementIncident · Feb 22
Inspect exact observations
Historical trade observations. This is not an executable-depth series.
Date (UTC)Maximum within-leg move (%)
2026-02-010.000099874100140462623987
2026-02-020.0012501192043795093533
2026-02-030.044369570178304953945
2026-02-040.00024588175848873264216
2026-02-050.039443018858765913246
2026-02-060.00025493099904198230138
2026-02-070.0010478926788688783779
2026-02-080.15739375921253033859
2026-02-090.00039777968990199122701
2026-02-100.0020516161771109487971
2026-02-11Not available
2026-02-12Not available
2026-02-13Not available
2026-02-14Not available
2026-02-150.00011782723236872660389
2026-02-16Not available
2026-02-17Not available
2026-02-18Not available
2026-02-190.000009458948164964055997
2026-02-20Not available
2026-02-21Not available
2026-02-220.39196451642944242464
2026-02-230.022140556194779885722
2026-02-240.00012169280219998152551
2026-02-250.21018917025323734376
2026-02-260.000095314130116958650073
2026-02-270.00067170373157718006942
2026-02-280.0003824054379989333424

First 5 of 28 UTC days

The first 5 days of the daily record. The full table carries every day and every bound.
DayTradesVolume · USDCLargest move, one legFindings
2026-02-0123916.080.0000999%None
2026-02-02280142.670.00125%None
2026-02-0327544.950.0444%None
2026-02-04224158.120.000246%None
2026-02-05435635.100.0394%None

Every day, as it was recorded.

One row per UTC day: how many trades cleared, what they moved the price by, and which checks that movement was enough to fire. A day with nothing to report says so rather than showing a zero.

Inspect every day

Four readings for the month

13,547
trades in 28 UTC days
375,320.84 USDC traded, no genuine-trade rule applied
0.392%
largest move by one leg
0
days with a causal bound
2
days with an assuming bound

What the month came to.

The smallest rung Keel measures is 2%, and nothing one leg did in February reached it. The bounds that do exist hold only if the book did not change between two trades, which is a weaker claim and is labelled as one.

Open the backtest

Questions about Keel.

How is Keel different from a price feed?

A price feed reports a price. Keel measures the volume available near that price, the cost and reachability of price targets, and the resulting collateral constraints.

Is the data on this page live?

The product previews use unchanged examples from the backend API contract. The February chart uses archived trade observations. Both are labeled; neither is a live market recommendation.

What does partial confidence mean?

Some HIGH or CRITICAL checks could not be evaluated. The reported band is a floor: missing evidence could make the finding worse. It does not mean the asset is clear.

Does Keel need a wallet?

No. Keel is read-only. It never signs or submits a Stellar transaction. You can inspect the sample data and contract without connecting a wallet.

Does the backtest prove Keel would have prevented the incident?

No. An event marker is historical context, not proof of prediction or prevention. The published February trade analysis did not establish an actionable advance warning.

Can I use the result as a financial guarantee?

No. Keel is a proof of concept. Its thresholds are chosen rather than empirically calibrated, and its results should not be the sole basis for a financial decision.