Borrowing against Bitcoin,
the risk page.

LTV: the number that decides everything

Loan-to-value (LTV) is the loan amount divided by the current market value of your collateral.^[1] Borrow $60,000 against $120,000 of Bitcoin and your LTV is 50%. It is the single number every Bitcoin-backed lender watches, and because it has the live BTC price in the denominator, it moves every second the market is open, which is always.

There are three LTV levels that matter, and you need all three before you sign anything:

Illustrative thresholds matching River's published example^[1] (start 50%, margin call 70%, liquidation 80%). Every lender sets its own numbers and can change them, often mid-volatility. Read the actual loan agreement, not the marketing page.

The trap is that initial LTV feels conservative and the other two feel far away. At 50% LTV your collateral can be cut in half before you're liquidated, which sounds like an enormous cushion. Bitcoin erases that cushion routinely. A 50% drawdown is not a tail event for this asset; it has happened in every multi-year cycle.

A single wick can liquidate you, even one that reverses

Liquidation engines look at price, not at your conviction or at where price closes an hour later. If BTC prints a low that crosses your liquidation LTV, even for seconds on thin weekend order books, the engine can sell. You can be liquidated at the worst price of the month and watch the candle fully recover minutes later, holding nothing but the loan proceeds and a tax bill.

This is not hypothetical. On October 10–11, 2025, Bitcoin fell from roughly $120,000 to near $102,000 and over 1.6 million leveraged positions were force-liquidated in 24 hours, the largest liquidation event in crypto history^[5] verify×DON'T TRUST, VERIFYClaim: On Oct 10–11, 2025, BTC fell from ~$120K to ~$102K and over 1.6 million positions were liquidated in 24 hours, the largest such event on record.Verify at: market data aggregator ↗Exact crash low and liquidation count vary by source and exchange. Confirm against historical price data before relying on the figures.. The trigger hit late on a Friday, when liquidity is thinnest, order books are shallow, and a moderate sell order moves price further than it would midweek. Below the market sat a near-empty book down to $100K, so price fell straight through the void and rebounded just as fast once it overshot.

Some lenders soften the trigger with a grace window. Strike, for example, calls a margin at 70% LTV, gives borrowers 72 hours (extended from 24) to bring it back to 65%, and stresses that a margin call is not yet a liquidation.^[2] That helps against a slow grind. It does not help against a 90-second wick that blows through the liquidation line before you've seen the notification, and it depends entirely on the lender honoring its own stated window under stress.

Rehypothecation: your collateral becomes someone else's bet

Rehypothecation is when the lender takes the Bitcoin you posted and re-lends or re-pledges it to fund its own borrowing, trading, or yield.^[3] The same coins now back multiple obligations at once. You no longer have a clean, bankruptcy-remote claim to your specific Bitcoin; you have a claim on a balance sheet that may have lent your collateral down a chain you can't see.

Why it matters: when a rehypothecating firm fails, customers whose collateral was reused become unsecured creditors standing behind secured creditors.^[3] If your Bitcoin sits on the custodian's balance sheet, it is an asset of the estate, available to satisfy creditor claims that have nothing to do with you. Most crypto lenders that blew up in 2022 operated without the asset-segregation and reserve rules that constrain regulated broker-dealers.^[3]

The 2022–2023 wave is the case study, and the numbers are not small:

In each case the base-layer coins still existed on-chain. The customer claims on them did not survive intact. That is the entire lesson: when you post Bitcoin as collateral, you have converted a bearer asset you controlled into an IOU from a company you don't.

Why "I'll never get liquidated" is overconfidence

The plan is always the same: borrow at a low LTV, watch it closely, and post more collateral if BTC drops. It fails in predictable ways, and they tend to arrive together:

• Volatility is the base case, not the exception. A 30–50% Bitcoin drawdown has happened in every cycle. Your "cushion" is sized against an asset that deletes cushions for a living.
• Crashes are correlated with your inability to respond. The worst wicks hit on weekends and overnight. To "just add collateral" you need spare BTC or cash you can move in minutes, on the exact night the price is collapsing and exchanges are congested. Most people don't.
• Adding collateral is forced selling in disguise. Topping up means committing more of your stack to the same falling knife. You are doubling down to avoid a loss, the textbook way to turn a drawdown into a wipeout.
• Thin liquidity makes the wick deeper than the "real" move. The price that liquidates you may be a number BTC only touched because the order book was empty for ninety seconds. The engine doesn't care that it wasn't a fair price.

"I'll watch it" assumes you will be awake, liquid, and faster than an automated engine during the single most chaotic hour of the year. Liquidation systems are built precisely for the people who were sure they'd never need them.

Counterparty risk: not your keys, by definition

River states the custody risk plainly: "If the company is hacked, mismanages funds, or goes bankrupt, you may never get your bitcoin back, even if you make all your loan payments on time."^[1] Paying perfectly does not protect you from the counterparty itself failing. This is the same "not your keys, not your coins" principle covered in Bitcoin Security, applied to the moment you voluntarily give up the keys.

Some lenders advertise segregated, non-rehypothecated custody, sometimes with a qualified custodian or a multi-key arrangement. That is genuinely better than the Celsius model. It is still a promise on paper, enforced only if the company actually does what it says and survives the stress that breaks the promise. Verify the custody structure in writing; do not take the marketing line on faith.

When, if ever, it's defensible

This site's default is don't. Selling spot and paying the tax is usually the honest, lower-risk move, and a long-term holder who waits 12 months pays a much lower rate (see Bitcoin Taxes). But there is a narrow set of conditions under which a Bitcoin-backed loan is at least defensible, and even then only cautiously:

• Very low LTV. Borrowers who stayed under ~30% LTV avoided liquidation through the 2025 volatility entirely^[4] verify×DON'T TRUST, VERIFYClaim: Borrowers under ~30% LTV avoided liquidation through the 2025 volatility; those above ~70% faced margin calls.Verify at: AInvest ↗A generalization across lenders; your specific lender's thresholds and the exact drawdown determine your outcome.. The lower the LTV, the more of a crash you can absorb before a call. Low single-digit-to-twenties LTV, not 50%.
• Short, defined term and a specific need. A bridge for a closing date or a known tax-timing problem, where you can repay and reclaim your keys quickly, is a very different risk than an open-ended lifestyle loan.
• A genuine reason not to sell. A concrete tax or liquidity reason, e.g. avoiding a large realized gain you'll offset or defer, not just "I don't feel like selling."
• A reputable, over-collateralized venue with verified custody. Published terms, segregated collateral, no rehypothecation in writing, a real liquidation policy you've read. Treat every yield-bearing or opaque lender as a Celsius until proven otherwise.
• Headroom you can fund in minutes. Cash or BTC on standby to meet a call during a weekend crash, before you ever need it.

Related reading

See the glossary for plain-English definitions of LTV, liquidation, rehypothecation, and every other term used here.