The Bitcoin Credit Stack
rates are percent a year, continuously compounded · figures as of 2026-10-12 · read 2026-10-13 04:01 UTC

The bitcoin credit market, observed

What borrowing against bitcoin costs, in each currency it is borrowed in, and what stands behind it, across every venue this dataset covers — measured against published criteria, with no score and no ranking.

It also measures credit secured by ether, the same way and never added to bitcoin: choose Ether under “Secured by”.

Governed indices, free with history; methodology on each page.

Secured by
Borrowed in
BTC collateral
1.17m BTC
partial
1.14m BTC – 1.40m BTC
from 92 of 134 venues
Dollar debt against bitcoin
$33.23bn
partial
$32.43bn–$37.10bn
from 63 of 134 venues
Aggregate LTV
35.6%
partial
from 57 of 134 venues
Borrow rate (USD loans)
6.35%
partial
from 80 of 128 venues
beside it, not in it: corporate lending 11.42%
Lend rate (USD loans)
3.51%
partial
from 68 of 128 venues
Liquidates at
$46,204
partial
44.7% below $83,613
from 46 of 134 venues
Dollar loans only: other currencies are separate books, never added in

Debt and rates count loans made in US dollars and dollar stablecoins. Loans in euros, francs, yen, ether or bitcoin itself are separate books, shown by currency below, never added in.

What these figures cover — and what they leave out
  • BTC collateral — read for 92 of 134 venues; part of it is the supplied amount each pool reports, not what is pledged against a loan; 947,851 BTC of it is held by corporate issuers and reported by them — not pledged: there is no lien; about 27,865 BTC of it is estimated from desks' books and their cards' LTVs, within the range shown.
  • Dollar debt against bitcoin — debt against bitcoin read for 63 of 134 venues; 32 more lend from a pool or are financed without a lien — named, not added; part of it is partial (bitcoin-only borrowers in a pool, or a desk's bounded estimate); $26.33bn of it is corporate debt financed by bitcoin the issuers hold — no lien on it.
  • Aggregate LTV — both sides read for 57 of 134 venues.
  • Borrow rate (USD loans) — weighted by dollar debt for 80 of 128 venues; 5 posted rate(s) carry no weight and are not in it; the secured market's rate: corporate lending is not in it (no lien on the issuers' bitcoin), shown beside it.
  • Lend rate (USD loans) — weighted by dollar supply for 68 of 128 venues; the secured market's rate: corporate lending is not in it (no lien on the issuers' bitcoin), shown beside it.
  • Liquidates at — the priced books of 46 of 134 venues taken as one position — an aggregate, not a forecast.
  • Corporate lending, beside the rates — corporate lending, not bitcoin-backed credit: what the issuers pay on their listed debt and preferreds, which is what holders earn. Unsecured, with no lien on the issuers' bitcoin, so it is in neither the market's borrow rate nor its lend rate.
What “Liquidates at” is: one price for the whole book, not a forecast

"Liquidates at" is the bitcoin price at which this book, taken as one position, stops covering its debt at its liquidation thresholds — an aggregate, not a forecast: individual loans are levered differently and liquidate above it.

BTC/USDT hourly price from Binance public market data.

The market, by class
ClassBTC collateralshareDollar debt against bitcoinshareLTVBorrowLendLiquidates atThreshold (LTV)
Corporate ›947,851 BTC81%$26.33bn79%34.1%11.42%——
Algorithmic ›160,173 BTC14%$4.44bn13%45.1%4.88%3.35%$46,24245.0%–97.0% LTV
Posted card ›58,450 BTC5.0%$2.42bn7.3%44.3%14.39%6.14%—
Minted ›1,407 BTC0.1%$37.2m0.1%42.9%7.64%—$41,63094.0%–96.5% LTV
Auction ›0.4555 BTC<0.1%$17k<0.1%44.5%5.48%5.48%—
Outside the market — no rate class; shown with their own figures, never added to it:
Futures and brokers (no rate class) ›——————
Ordered by debt, never by a rate; a figure in italics covers part of a book

Ordered by debt against bitcoin, then collateral — never by a rate. Shares are of this market's total, and add up. A figure in italics covers only part of a book — hover for what it leaves out; a dash carries its reason. 32 lend from a pool against every collateral, report a whole book, or are financed without a lien: their figures are shown on their rows and not added ($16.06bn in all). /v1/market/tree

By the currency the loans are made in

Each currency is its own book, compared in dollars, never added across

Each currency is its own book. Sizes are dollar equivalents so they can be compared, and are added only within a currency, never across. Corporate debt is the issuers' filed debt in its own currency, with no lien on bitcoin.

Borrowed inSecured by bitcoinVenuesCorporate (no lien)Assets
US dollars$4.48bn51$26.33bnCRVUSD, USDC, GHO, USDT0, USDT, USDTB
Not dollars, not yet split$136m25—
Euros$30.2m4$977mEURC, EURCV
Swiss francs$15.9m1—ZCHF
Bitcoin (against bitcoin)$4.2m4—WBTC, cbBTC
Ether$709k4—ETH, WETH
Peg not established$104k1—JupUSD, WSOL
Yen——$200m
How each currency’s book is read, and what the desks do not state

On-chain books as each venue's best observer reads them today. CeFi desks state what they lend in, not how much in each: of 24 desks and brokers, no live one states bitcoin loans. “Not dollars, not yet split” is what bitcoin-only pool borrowers owe outside dollars, which the account read does not yet split by currency.

algorithmic minted auction posted cardsolid = observed · dashed = fitted · shaded = p10–p90 of the pool rates behind the mean
0%3.5%7%10.5%14%Rate (% a year, continuously compounded)−7d−14d−30d−90d−1y−2y+7d+14d+30d+90d+1y+2yTODAYrealised — trailing window, not a maturitycontracted — the term the borrower agreedalgorithmicauctionmintedposted cardtreasuries
Started 90d ago · what it cost
algorithmic4.57%
n=793 pools · 80 venues · 90d observed · $13.71bn debt · p10–p90 3.48–5.96% over 525 pools
auction4.97%
cleared — one rate, both legs · n=1 pools · 1 venue · 85d observed
minted7.78%
n=23 pools · 3 venues · 90d observed · $643m debt · p10–p90 4.85–9.08% over 19 pools
posted card11.75%
n=7 pools · 7 venues · 84d observed
treasuries3.80%
3.84% bond-equivalent as quoted · locked at origination — the 3m Treasury quoted 2026-07-14
90dhorizon
Agreed today for 90d
algorithmicno term agreed
auction5.58%
5.61% APR as quoted · fitted to n=20 auctions · 2 counterparties · 0 external
mintedno term agreed
posted card10.37%
10.50% APR as quoted, the median card · 1 desk · posted 2026-07-21 · administered, not cleared · 10.37–10.50% cc over the candidate compounding bases
treasuries4.19%
4.23% bond-equivalent as quoted · US Treasury constant-maturity, 2026-10-08
Left of today is what a borrower who started then has paid since — an averaging window, never a maturity. Right of today is the term a borrower agrees now. Nothing is drawn across zero: the two are different quantities and the step between them is a reading, not a slope. Solid where observed, dashed and lighter where fitted. Rates are percent a year, continuously compounded; each point also carries the rate as quoted, in its own unit. get_indicator({ id: "benchmark-curves" })

No borrow leg is observable for: corporate. The two legs are not symmetric — a class with no depositor has no lend rate, and a corporate security has a yield to its holder rather than a borrowing cost anyone can take. A zero would be a different claim.

Source: Board of Governors of the Federal Reserve System (H.15, H.4.1, H.6).

Data provided for free by IEX. By accessing or using IEX Historical Data, you agree to the IEX Historical Data Terms of Use. IEX

Collecting — the corporate term instruments have maturities but no observable yield — 15 convertible notes, 3 credit facilities and 1 bond mature between 2026 and 2032; 0 of the 19 carry a yield.
On the auction layer. Fitted by log_quadratic to n=20 observations across 7 tenor bins from 2 counterparties, 0 of them external · r² 0.52.
How these loans are backed2026-10-13 · /v1/market/debt-by-collateral · get_debt_by_collateral()
$26.33bn
$16.36bn
financed by bitcoin$26.33bn53%no lien on the coins — the issuers hold 947,851 BTC against it, 2.8x the registered debt (a ceiling, see the caveat)a pool$16.36bn33%basket only — no dollar is assignable to a collateral. Of it, $1,794m is owed by accounts posting bitcoin among other collaterala named wrapper$2.58bn5.2%exact — the market IS a (collateral, loan) paira custodial desk$2.42bn4.9%estimate · $1.62bn–$6.24bn, of which $785m published or filed. bitcoin held in custody at the desk — 7 of 13 desks counted, 3 of them from a published or filed book and the rest bounded from stated inputsa pool, bitcoin-only accounts$1.90bn3.8%exact — dollar loans to 19,306 accounts at 27 pooled venues whose only collateral is bitcoin, read one by one from the pools' contracts; the $136m they borrowed in other currencies stays in the pool segmenta vault$2.2m<0.1%exact — debt and collateral are held per vault typeescrowed at auction$17k<0.1%exact — the collateral is escrowed by the protocol against the loan
The largest segment, opened: what the pooled reserves hold — collateral, not debt, and not assignable to any one loan
other $19.12bn
dollar $13.16bn
ether $9.58bn
bitcoin $8.02bn
What the bars can tie to a specific coin, and what they cannot

Only the named wrapper, the vault, the auction and — inside the pools — the accounts whose only collateral is bitcoin tie a dollar of debt to a specific coin. The rest of a pool lends against its whole basket, so no dollar of that debt belongs to any collateral — the second bar is what the pools hold, never an assignment. The corporate layer is financed BY bitcoin, not secured on it: no lien exists, and the issuers hold 947,851 BTC against it, 2.8× the registered debt — a ceiling, since senior convertible notes are not in that denominator.

The custodial desks — 13 of them, and how each figure was arrived at · $2.42bn midpoint, $1.62bn–$6.24bn
DeskBookHowAssuranceCollateralIn the bar
tether$13.45bn—reviewed unaudited
2026-06-30
unstatedno — collateral unstated
nexo$0–$1.70bn
mid $850m
UPPER BOUND ONLY, arithmetic midpoint.derived
2026-09-16
all collateralno — whole book, not bitcoin
ledn$785m—self reported
2026-10-01
bitcoinyes
galaxy$668m–$859m
mid $763m
OPERATOR ASSUMPTION, NOT A DISCLOSURE (operator, 2026-09-21: "a high BTC split").derived
2026-06-30
bitcoinyes
cantor$100m–$2.00bn
mid $447m
GEOMETRIC MIDPOINT of a bound spanning a factor of twenty.derived
2026-09-16
bitcoinyes
strike$50m–$2.10bn
mid $324m
GEOMETRIC MIDPOINT of a bound spanning a factor of forty-two.derived
2026-09-16
bitcoinyes
xapo$20m–$500m
mid $100m
ANCHORED, NOT INTERPOLATED.derived
2026-09-16
bitcoinyes
salt$10m–$200m
mid $45m
ANCHORED ON THE LAST FILED FIGURE, WIDENED FOR FIVE YEARS OF SILENCE.derived
2026-09-16
all collateralno — whole book, not bitcoin
blockfi$0—reported no assurance
2026-09-16
bitcoinyes
celsius$0—reported no assurance
2026-09-16
bitcoinyes
two prime———bitcoinno — collateral unstated

⚠ THIS SEGMENT IS PART MEASUREMENT AND PART ESTIMATE, and the two are never merged into one figure. 3 of 7 counted desks publish or file a book; the rest are bounded from stated inputs, each carrying its own method and sources. The bar draws the midpoint; the range is what is actually known. Where a bound spans an order of magnitude the midpoint is the GEOMETRIC mean, because the arithmetic one sits near the top of such a bound and asserts a precision the bound does not have.

Every one of the 13 carded desks now carries a figure, and they are not the same kind of figure. Two are ZERO and exact — BlockFi and Celsius are in confirmed Chapter 11 wind-down and originate nothing. One publishes monthly (Ledn). One filed a book with the SEC, in 2021, and has not filed since (SALT). Two have announced FACILITIES rather than books — Cantor $2bn, Strike $2.1bn — which is capacity, not credit, and is carried under its own metric so no total can reach it. The rest are bounded estimates. ⚠ Coinbase is absent from this segment ON PURPOSE: it originates its retail bitcoin-backed loans into Morpho Blue markets on Base, so its book is already read on-chain every day and already counted in the named-wrapper segment. Adding it here would count it twice.

NOT ONE FIGURE IN THIS SEGMENT IS AUDITED AS A CURRENT BOOK. Ledn's monthly metrics are computed by an independent CPA firm from the desk's own systems, and that firm disclaims assurance in terms: "not an audit, review, or attestation engagement, nor does it express any form of assurance, opinion, or conclusion". What IS attested at Ledn is a separate Proof of Reserves — assets at least client liabilities — as of 2026-03-31. SALT's $63.4m IS audited and is five years stale, which is a different problem wearing the better grade. Everything else is reported, announced or derived.

Leverage — where borrowers sit, and where they mayno time axis · /v1/market/ltv-history (not on the MCP yet) · /v1/market/cefi-rates · get_desk_rates()
0%20%40%60%80%100%loan-to-valuerealised217 mkts · 8 venues54.9%86.8% liquidation32.0pp headroom48.7%market-wide bclv — a different denominator, not a correctionalgorithmic48.7% · 2 venues · $2.48bnauction44.5% · 1 venue · $17kminted20.3% · 1 venue · $2.2mPOSTED MAXIMUM — A CEILING, NOT A POSITIONXapo40% · liquidates on LTVStrike45% · no price liquidation (hedged) · card from 2026-07-21Ledn50% · liquidates on LTVMilo50% · liquidates on LTVNexo50% · liquidates on LTVSALT50% · liquidates on LTVUnchained50% · liquidates on LTVArch60% · liquidates on LTVCoinbase75% · oracle liquidation · card from 2026-07-21Figure75% · liquidates on LTVa card is marked stale after 45 days
What the realised LTV covers, and what it leaves out, by class

On the realised figure. Debt-weighted, over the venue-API markets where both a liquidation threshold and both sides of the book are published. NOT the whole market: Aave and Compound do not publish a per-market threshold this estate ingests.

algorithmic. Covers only the venues read exactly (isolated markets). Pooled venues carry no realised series, so this is true and narrow; venues and debt are on every point. Their bitcoin-only borrowers are served per venue (bitcoin_only_borrowers), and join the class LTV on the headline.

posted-card. No desk in this class discloses a realised LTV series. One desk (Ledn) has published one average LTV at one date; one point is not a series. The desks' POSTED maximum LTVs are served on /v1/venues/compare as ltv.max, a policy and not an observation.

corporate. A corporate instrument has no collateral posted to a lender, so there is no LTV. The coverage ratio against the attachment point is served below as corporate_analogue — an analogue, not an LTV.

A position and a permission: why the two bands share one axis

The two bands are different quantities and the rule between them says so. A realised LTV is a position; a posted maximum is a permission nobody is obliged to use. They share an axis because both are loan-to-value, and nothing else.

What the bitcoin is — collateral wrappers161,581 BTC at 81 venues — contract reads and the aggregator
cbBTC 47% · WBTC 35% · kBTC 5% · BTCB 4% · LBTC 3% · 15 more · get_credit_state
How that mix got herelevels, not shares · the bar is this chart's right edge · get_credit_state
2026-10-12161,581 BTCcbBTC 76,439WBTC 56,612other wrappers 9,102kBTC 7,635BTCB 6,655LBTC 5,137observed at 81 venues, 258 markets · contract reads + aggregator · daily— hover to read any day
0100k200kBTCcontracts at past blocks · weeklyaggregator · weekly2020202120222023202420252026

Levels in BTC, not shares. Every wrapper, and whether each is backed →

The five classes, each with its venues → Every venue and its criteria →

Ask your AI about any of this, with your question attached: the MCP serves the market, its classes, venues and indicators. Panels that name only a REST path are on the API alone for now.Connect your AI ↗