01 / THE PREMISE
Don't sell. Borrow β but know what you're really risking.
If you believe Bitcoin trends higher over a long horizon, selling to raise cash is the one move that permanently ends your position and hands the tax office a bill. Borrowing against the same coins does the opposite: you keep the stack, you stay exposed to the upside, and β critically β drawing a loan is not a disposal, so there's no capital-gains event at the moment you take the money.
That's the appealing half of the story. The other half is the part that actually ruined people. In 2022, the holders who lost Bitcoin mostly didn't lose it to the price. They lost it to the platform. Celsius, BlockFi and Voyager didn't just fall in a falling market β they froze withdrawals and went bankrupt while still holding customer collateral, and depositors became unsecured creditors in a queue. The coins were gone before the liquidation engine ever ran.
So the first risk to price isn't market risk. It's: will the place holding my Bitcoin still be solvent β and still have my Bitcoin β when I want it back?
That reframes the whole exercise. A headline interest rate tells you almost nothing about whether you'll get your coins back. The thing that does is the custody model β who controls the keys, and whether your collateral can be lent out behind your back. That's the axis we sort by.
02 / THE LANDSCAPE
Sorted by custody, not by rate
Lenders cluster into three custody models. The further down this list you go, the more control you keep over the actual coins β and the more the risk shifts from "will they go bust" to "will the code or the market get me."
They hold the keys β but ring-fence your coins
A company custodies your Bitcoin, but holds it 1:1 in segregated storage and contractually does not lend it out. You're trusting their custody and solvency, but not betting your collateral is funding someone else's trade. This is the closest a custodial product gets to safe.
Ledn (the Custodied product) Β· Vield (AU) Β· Block Earner (AU)
You co-sign β nobody can move the coins alone
Collateral sits in a 2-of-3 multisig where you hold a key. The lender can't unilaterally move, lend, or rehypothecate the Bitcoin; a liquidation requires keys to agree. The trade-off is that this model is currently aimed at larger, business-grade borrowers.
Unchained β business loans only, US$150k minimum
Code holds the collateral, not a company
Your collateral is locked by a smart contract or a discreet-log/multisig escrow rather than a counterparty's balance sheet. No company can go bankrupt with your coins β but you inherit smart-contract and oracle risk instead. On Ethereum, "BTC" means wrapped WBTC, which adds bridge/custodian risk on top.
Lava Β· Hodl Hodl (Lend) Β· Aave (BTC only as wrapped WBTC)
The rehypothecation tell β the Celsius failure mode
One heuristic catches most of the dangerous platforms: if a platform pays you yield on deposited Bitcoin, your collateral is almost certainly being lent out. Yield has to come from somewhere, and "somewhere" is your coins financing third-party leverage. That's rehypothecation, and it's exactly what turned a price drawdown into permanent loss in 2022. A pure borrowing product against segregated, non-rehypothecated collateral should pay you nothing on the BTC you post β and that's the point.
03 / RISK CATEGORIES
Five ways a Bitcoin loan goes wrong
Every platform trades one of these risks for another. Read a product by asking which of the five it has reduced β and which it has quietly increased.
(a)
Counterparty & custody
The lender becomes insolvent, freezes withdrawals, or has rehypothecated your collateral. The 2022 failure mode. Mitigated by segregation, no rehypothecation, and ideally a key in your own hands.
(b)
Liquidation & market
Price falls to your liquidation level and the platform sells your collateral β often on its own oracle, often on a brief wick. The risk you can actually size, and the focus of the calculator below.
(c)
Smart-contract (DeFi)
A bug, exploit, oracle manipulation, or governance attack drains the protocol β or a wrapped-BTC bridge fails. No counterparty to sue; the code is the counterparty.
(d)
Regulatory & tax (AU)
Rules change, a provider exits your jurisdiction, or a transfer/liquidation triggers an unexpected CGT event. Covered in detail below β and the most overlooked.
(e)
Rate, refinance & term
Variable rates climb, the loan matures into a bad market, or you can't refinance and are forced to sell. Interest that accrues to the loan quietly raises your liquidation price over time.
04 / THE AUSTRALIAN TAX ANGLE
A loan isn't a sale β until it's a forced one
General information only, current as of June 2026. Not tax advice. The ATO's treatment turns on specifics β confirm your situation with a registered tax agent.
Drawing the loan is not a CGT disposal
Receiving loan proceeds against your Bitcoin is borrowing, not selling. You haven't disposed of a CGT asset, so there's no capital-gains event when the cash hits your account. This is the core tax advantage over selling.
A forced liquidation is a disposal
If the market falls and the platform sells your collateral to cover the loan, that sale is a disposal β it crystallises capital gains, and it does so at a cycle low, the worst possible time to realise a gain. You can owe tax on a position you no longer hold, paid out of coins sold near the bottom.
Transferring BTC to the platform may itself be a disposal
For some arrangements the ATO may treat sending your Bitcoin to a lender β where beneficial ownership changes or it's pooled/rehypothecated β as a disposal at the moment of transfer, before any loan even funds. Whether posting collateral is a CGT event depends on the platform's legal structure. Confirm per platform; this is precisely where custody model and tax outcome collide.
05 / THE CORE INSIGHT
Liquidation risk is distance-to-support, not distance-to-spot
Here's the mistake almost everyone makes. They look at the gap between today's price and their liquidation price and feel safe: "BTC would have to fall 30% to liquidate me β and it's up here, so I'm fine." But 30% from spot is the wrong measure. Bitcoin doesn't fall to spot-minus-thirty and stop. It falls toward its support line.
The power law gives that support a shape. Bitcoin's price has spent its entire history inside a rising corridor; the lower edge β the β2Ο band of the long-run trend β is a floor it has only briefly broken, once, for hours, in March 2020. The drawdown that can actually hit you is spot minus support. That's how far price can realistically fall in an ordinary cycle. So the question isn't "how far is my liquidation price below spot." It's:
Does my liquidation price sit below the support line β for the whole loan term?
If your liquidation price is below the power-law support, then to be liquidated Bitcoin has to break through its historical floor β something that essentially hasn't happened. A routine 50β80% cycle drawdown that merely visits the floor and bounces never reaches you. If your liquidation price is above the support, an entirely normal cycle bottom β the kind that happens every few years β wipes you out inside the corridor the model considers business-as-usual.
And because the support line rises with time while a sensibly-sized loan's liquidation price stays roughly flat, your safety margin usually widens as the loan ages. Risk is concentrated at the start. Size the loan so the liquidation price clears the support line on day one, and the model carries you the rest of the way.
06 / THE MATH
Sizing the loan
Two formulas do all the work. Your liquidation price is set entirely by the ratio of the LTV you start at to the LTV that triggers liquidation β the entry price just scales it:
Liquidation price β Entry price Γ ( Initial LTV Γ· Liquidation LTV )
// how far price can fall before that happens
Survivable drawdown = 1 β ( Initial LTV Γ· Liquidation LTV )
The lesson is blunt: borrow less and you survive more. At a 70% liquidation trigger β a typical threshold β halving your starting LTV roughly doubles the drawdown you can absorb:
| Initial LTV | Liq. price (as % of entry) | Liquidated on a drop of | Survives a drawdown of |
|---|---|---|---|
| 50% | 71.4% | β 29% | β 29% only |
| 30% | 42.9% | β 57% | β 57% |
| 20% | 28.6% | β 71% | β 71% |
Now put that next to history. Bitcoin routinely draws down 20β50% within a cycle, and full bear markets have run far deeper β the early cycles fell 80β85% top to bottom. The current 2026 cycle is already ~50% off its October 2025 high (about US$126k down to roughly US$64k). A 50% LTV loan survives a 29% drop β it would already be in trouble this cycle. A 20% LTV loan, surviving ~71%, sits below even an early-cycle-style collapse. The drawdown column is the one to compare against the support line, because that drop is what the floor lets through.
07 / STRESS-TESTER
Liquidation vs the floor
Set up a loan and the tool plots your flat liquidation price against the rising power-law support line over the full term. Green if your liquidation price stays below support the whole way β model-safe. Red if the support line is above your liquidation price at any point in the term, meaning a normal visit to the floor could take you out. The power-law parameters are the exact ones from the powerlawfloor calculator (genesis 2009-01-03, β2Ο support band).
Configure the loan
connectingβ¦Note: the chart draws the liquidation line flat for clarity, but interest that accrues to the loan slowly raises it β the total interest stat shows how much. The support line is the model's β2Ο band; real lenders liquidate on their own index and can trigger on brief wicks below your level. Illustrative only.
08 / THE COMPARISON
Where to borrow
Filter by custody model β the axis that matters most β and sort any column. Every rate is an indicative snapshot, verify with the provider; maximum LTVs are headline figures and prudent borrowing sits well below them (see the maths above).
| Platform | Custody model | Max LTV | Indicative rate | Jurisdiction / notes |
|---|
β Rates & limits are indicative, as of June 2026 β verify with each provider. Custodial platforms still carry counterparty risk even when collateral is segregated.
09 / IN SHORT
The whole framework in four lines
- Pick on custody first. Segregated & no-rehypothecation, collaborative multisig, or true non-custodial β never a yield-paying product holding your collateral.
- Size by support, not spot. Put your liquidation price below the power-law floor for the entire term.
- Borrow less than you think. A 20β30% LTV survives a real bear market; a 50% LTV barely survives a routine dip.
- Remember the tax trap. A forced liquidation is a sale β it crystallises CGT at the bottom. Don't let it happen.
Related: the Loan Risk tool, the power-law calculator, and the floor explainer.