Bitcoin-backed mortgages are no longer experimental — Milo has originated $100M+, and Better and Coinbase launched a Fannie Mae-backed conforming product in March 2026. Roughly one in six US adults report having used digital assets at some point (Pew Research, 2024). But the product trades one set of risks for another. This deep dive builds the category from primitives, runs the math, and gives you an interactive calculator to find out where you actually stand.
Most writing on this product starts with the marketing label ("Bitcoin mortgage") and works outward. That is backwards. The label encodes assumptions you should derive, not import. Start from primitives.
Bitcoin is a digital bearer instrument with five essential properties: scarce (21M cap), bearer (key control = asset control), liquid (24/7 global markets), volatile (50–80% annualized realized vol vs. 12–18% for broad equities), and non-sovereign (no issuer, no central counterparty).
Two of these dominate the mortgage product: bearer makes custody the central design question, and volatile makes the loan structure exotic.
A mortgage is a long-duration secured loan with four essential properties: long-duration (15–30 years), secured by specific collateral, amortizing on a schedule, and recorded with a public lien registry. Its native collateral — real estate — is illiquid, slow-moving, has recorded title, and is use-coupled (the borrower lives in it).
Combining the primitives: a Bitcoin-backed mortgage is a long-duration amortizing loan with security interests in two collateral pools with maximally different liquidity and volatility profiles. The lender takes a recorded lien on the property (slow-moving, illiquid) and a security interest in pledged Bitcoin under a custody arrangement (fast-moving, liquid). The borrower retains beneficial ownership of both; the lender's claim is conditional on default events defined in the loan agreement. No traditional secured loan structure pairs collateral with such different time constants. That dual-collateral structure is the thing to understand — everything else is downstream.
Three products get blurred under the "Bitcoin mortgage" label. They are different from first principles.
| Property | Crypto-Assisted | BTC-Collateralized Loan | BTC-Backed Mortgage |
|---|---|---|---|
| Taxable event at origination | No | Sale required → Yes | No |
| BTC retained by borrower | No | Yes (pledged) | Yes (pledged) |
| Secured by real estate | Yes | No | Yes |
| Secured by BTC | No | Yes | Yes (both) |
| Typical term | 15–30 yr | 1–3 yr | 30 yr |
| Typical LTV vs. collateral | 80–95% | 30–50% | up to 100% vs. BTC |
| Margin call risk | No | Yes | Yes |
Only the third column — Bitcoin-Backed Mortgage — is the subject of this deep dive. The other two are useful reference points for the trade-offs.
Not every Bitcoin-backed mortgage uses the same risk-transfer mechanism. As of 2026 there are three distinct structural models in market, distinguished by what triggers a liquidation event — price moves, rate adjustments, or only payment delinquency. The choice between them is one of the most important decisions a borrower makes.
The mortgage rate is fixed. The lender monitors the BTC collateral's value. If Bitcoin's price drops far enough from origination (Milo's published threshold: a 65% drawdown), the lender issues a margin call. The borrower can add collateral, pay down principal, or accept forced liquidation of pledged BTC.
Milo originated $100M+ on this model. Notable operational detail: Milo has publicly stated it has issued zero margin calls to date, which suggests real-world behavior involves communication and borrower-managed paydowns more than mechanical liquidation at threshold. The structural risk is still real, but the worst case is rarer than the threshold framing alone would suggest.
The mortgage rate is variable, tied to the collateral ratio and Bitcoin's price. If Bitcoin appreciates, the rate falls. If Bitcoin drops, the rate rises — compensating the lender for the added risk. There is no price-triggered margin call. Liquidation occurs only if the borrower defaults on payments.
Peoples Reserve's Bitcoin Powered Mortgage (BPM) is the leading public example. Per company materials and media coverage, rates start around prime + 1 with discounts available for higher collateralization (reportedly reaching low single-digit APRs at substantial over-collateralization); custody uses multisignature wallets with no rehypothecation. Specific rate figures are based on company-quoted terms — verify directly with Peoples Reserve before relying on them.
Peoples Reserve also markets adjacent products (HEBLOC, a Bitcoin-powered Treasury bond) — those are separate structures and out of scope for this deep dive.
Launched March 2026. The mortgage rate is fixed, the loan is a Fannie Mae-backed conforming mortgage, and — structurally distinct from Models A and B — there are no price-based margin calls and no top-up requirements regardless of how far BTC drops. Per Better's published product description, liquidation risk applies only in the event of a 60-day payment delinquency, the same standard that applies to any conforming mortgage.
Collateralization is set up-front: pledge $2.50 of Bitcoin (or $1.25 of USDC) for every $1 of down payment financed. The crypto sits in Coinbase Prime custody for the life of the down-payment loan. BTC price moves do not affect the borrower's monthly obligation, do not trigger margin calls, and do not put the collateral at price-driven liquidation risk.
Structurally this is the closest product to a traditional conforming mortgage. The crypto serves as up-front collateral for the down-payment loan, not as a continuously monitored LTV variable. If you pay your monthly mortgage on time, the crypto stays where it is, regardless of BTC's price path.
All three address the same problem (BTC holder wants real estate without selling BTC) but transfer risk through different mechanisms:
| Property | Model A — Price margin call | Model B — Dynamic rate | Model C — Delinquency-only |
|---|---|---|---|
| Mortgage rate | Fixed | Variable (tied to BTC + collateral) | Fixed (conforming) |
| Price-based margin call | Yes | No | No |
| Liquidation trigger | BTC price drop past threshold | Payment default only | 60-day payment delinquency only |
| Forced sale of pledged BTC mid-term | Possible | Default only | Default only |
| Monthly payment certainty | High (fixed rate) | Low (depends on BTC path) | High (conforming rate) |
| Initial collateralization | Up to 100% of loan vs. BTC value | Variable; higher collateral = lower rate | $2.50 BTC (or $1.25 USDC) per $1 of down-payment financed |
| Custody model | Lender-held or qualified custodian | Multisig, no rehypothecation | Coinbase Prime custody |
| Conforming/agency-backed? | No (non-QM portfolio) | No (non-QM portfolio) | Yes (Fannie Mae) |
| Where the risk lives | In the BTC collateral (forced sale) | In the cash flow (variable payment) | In the borrower's ability to make payments |
| Example lender | Milo | Peoples Reserve | Better + Coinbase |
Model A's risk lives in the asset (forced sale on price). Model B's risk lives in the cash flow (variable payment). Model C's risk lives in the borrower's payment discipline — structurally the closest thing to a traditional mortgage with extra collateral up front.
None is universally better. A borrower with stable income and a strong view that BTC will compound steadily may prefer Model C — the conforming structure with no active price-monitoring threshold. A borrower with stable income and a willingness to absorb price-path-dependent payment variability may prefer Model B. A borrower with fixed income, lower BTC conviction, and tolerance for the tail-risk of a forced sale event may accept Model A (often at the lowest headline rate).
The remainder of this deep dive — including the formal mechanics in Part III and the interactive calculator in Part V — primarily models Model A, because its price-triggered dynamics produce the cleanest closed-form math and are the most informative for risk-decomposition purposes. Model C borrowers face no price-based liquidation surface, so the LTV-evolution math in Part III does not apply to them in the same way. Model B requires a stochastic interest-rate model dependent on BTC's path and is best evaluated case-by-case with the originating lender.
The product can be described by a small set of state variables. Stating them formally clarifies what is and is not at risk.
P(t) — price of Bitcoin at time t (USD per BTC)Q — quantity of BTC pledged (fixed at origination)L(t) — outstanding loan balance at time t (decreases via amortization)LTV(t) = L(t) / (P(t) × Q) — current loan-to-valueM — the lender's margin-call threshold, expressed as either an LTV ratio or a percentage drop in P(t) from originationImportant caveat about the threshold: The mapping between "BTC price drop from origination" and "LTV ratio at which margin call triggers" depends on the initial collateralization. Milo publishes a 65% price-drop threshold for a product structured with ~1:1 BTC-value to loan-principal at origination, which translates to LTV ≈ 2.86 at trigger. Products structured with different initial collateralization ratios (e.g., Better/Coinbase's 2.5:1 BTC-to-loan ratio for the down-payment loan) map a 65% price drop to a very different effective LTV — and Better/Coinbase doesn't use a price-based trigger at all (Model C above). The threshold logic in this section formalizes Model A (price-based margin call) at the initial-collateralization point Milo uses; it is not a universal rule.
For a borrower in a Model A product: good standing as long as LTV(t) < M. When LTV(t) ≥ M, the lender issues a margin call. The borrower can add to Q, reduce L(t), or do both — or the lender liquidates pledged BTC to restore the LTV.
Three observations from the dynamics:
The 65% margin threshold sounds far away. Bitcoin's history says it has happened every cycle.
| Cycle | Peak | Trough | Drawdown | Recovery to peak |
|---|---|---|---|---|
| 2011 | ~$32 | ~$2 | −94% | ~22 mo |
| 2013–15 | ~$1,150 | ~$170 | −85% | ~36 mo |
| 2017–18 | ~$19,800 | ~$3,200 | −84% | ~36 mo |
| 2021–22 | ~$69,000 | ~$15,500 | −77% | ~24 mo |
| Median | — | — | −85% | ~30 mo |
Every major cycle has produced a drawdown larger than the 65% threshold. The orange line shows where Milo's margin call would trigger.
Past recovery windows have spanned 22–36 months. The structural lesson: historically, major BTC drawdowns have often reversed over multi-year periods, which means an early forced liquidation can convert a temporary impairment into a permanent loss. Whether any specific borrower experiences a margin event depends on entry timing, initial collateralization, principal amortization, prepayment behavior, and the lender's specific liquidation logic — so "plan around the assumption that one will occur" is one defensible framing, but is not a mechanical rule. A borrower originating mid-cycle with conservative collateralization and steady prepayments faces materially different risk than one originating at peak with maximum leverage.
The strongest defense is mechanical: front-loaded principal paydown. Extra payments in years 1–5 are functionally identical to buying margin-call insurance.
Adjust the inputs to your stack, the property, and the lender's terms. The model projects 30 years of LTV evolution, net worth under each scenario, and tax efficiency — and tells you the break-even BTC return.
What this calculator models — and what it doesn't. This projects Model A (price-based margin call, Milo-style — fixed rate, margin call at BTC drawdown threshold). It uses smooth annual compounding for BTC price paths, which is a path-independent simplification. Real BTC price paths are highly path-dependent — a 60% drawdown in year 2 followed by a recovery to the same end-of-term price produces very different margin-call outcomes than a smooth path with the same CAGR. The calculator's "margin call" output should be read as "did the path ever cross the threshold," not as "did the smooth CAGR end below the threshold." For Model B (dynamic-rate, Peoples Reserve-style) and Model C (delinquency-only, Better+Coinbase), this projection does not apply — see the Models section above and run the comparison with the originating lender.
Adjust the inputs to see the decision logic for your situation.
| Year | BTC Price | BTC Value | Loan Balance | LTV | Pledge NW | Sell NW |
|---|
Risk on this product is not one-dimensional. Decomposing it makes the evaluation tractable.
The risks are correlated. A BTC drawdown that triggers a margin call typically arrives in the same macro environment as job-loss risk and lender-stress risk. Diversifying across them is harder than it sounds.
Market risk is publicly observable. You can model it, hedge it, watch it. Counterparty risk hides until it ruptures, and then ruptures fully. The 2022 cohort (Celsius, Voyager, BlockFi, Genesis) proved this — all failed for the same four structural reasons: maturity mismatch, rehypothecation, counterparty concentration, and pooled non-segregated custody. Ledn and Milo survived because they had term-matched books, no or limited rehypothecation, counterparty diversification, and bankruptcy-remote custody.
The lesson for lender evaluation today: structural questions, not promotional claims. Marketing about "safety" is noise. What matters is the legal structure of custody, the disclosure of rehypothecation, and the published bankruptcy-remoteness arrangements.
The product fits a specific borrower profile cleanly. Outside that profile it ranges from suboptimal to catastrophic.
The strongest version of the trade is for a holder who is already past the wealth-building phase with respect to Bitcoin, has multiple stacks across different storage arrangements, has stable income from non-Bitcoin sources, and is using the mortgage to deploy a small slice of their stack into housing without disturbing the rest.
The markdown companion goes deeper on lender evaluation, the 2022 failure autopsy, and the open questions worth watching as the category matures.
Read the full piece → All Deep DivesThis calculator and accompanying piece are educational. They do not constitute legal, tax, or financial advice. Specific decisions require a CPA, attorney, or financial professional with Bitcoin experience.
Projections are scenario analyses, not predictions. Bitcoin's future return distribution is unknown. The calculator uses simplified compounding (smooth CAGR) and does not model intra-year volatility or stochastic margin call events. Real lender terms — rates, margin thresholds, custody structure — vary; verify with the originating lender before relying on any figure.