Loans methodology

How Pledge scores Bitcoin-backed loans.

This page explains the eight-factor scoring model used to rank Bitcoin-backed lenders. It covers what each factor measures, how a score computes, what a 10/10 loan would look like, the landscape in one view, and what the methodology cannot capture.

Composite scores range from 0 to 10. 8 weighted factors sum to 100%. Scores last verified August 23, 2026. Past performance does not guarantee future results.

Compare Bitcoin Loans Open the loans glossary

Scoring factors

Each factor is weighted and scored 0–10 before blending into a composite.

Total weight

100%

Factor weights sum to 100%. No hidden adjustments.

Lenders scored

Live scores run 3.39–9.02 across the tracked landscape.

Limitations

Documented

Category-relative caveats, conflicting flags, and data snapshots are listed explicitly.

What Pledge scores

Every Bitcoin loan is a custody arrangement wearing a rate sheet.

Loans are the flagship scored vertical on Pledge \u2014 the deepest compare desk \u2014 and this page is the dedicated methodology behind them. It answers one question first: what can the lender do to your bitcoin, and what happens to you if the loan goes sideways?

APR and LTV are features, not safety. The cheapest loan wrapped in the wrong custody structure is the most expensive loan there is. So Pledge scores loans on borrower protection \u2014 the safety of the arrangement \u2014 and leaves rate math to the comparison pages.

A loan score measures borrower protection for that category. It is not a safety rating across product types: a 7.5 loan and a 7.5 treasury are not the same thing.

Scoring factors

Eight weighted factors determine each lender\u2019s score.

Every factor is scored 0\u201310, then weighted. Weights are fixed and published; editors do not override individual scores to elevate or demote a lender. Rehypothecation is deliberately the heaviest factor at 20%.

What 10/10 means

A 10/10 Bitcoin loan would be nearly boring.

The borrower keeps meaningful control of collateral, collateral cannot be reused, reserves or collateral handling are independently verifiable, legal status is clear, liquidation rules are conservative and explicit, and the lender has survived multiple market cycles with clean public evidence.

20% weight≥ 20%

Rehypothecation Policy

Can the lender re-lend, re-pledge, or reuse your collateral? This is the red line — the single feature that turned lender stress into customer losses in 2022.

10/10 standard

Collateral cannot be re-lent, re-pledged, sold, transferred, or otherwise used by the lender or funding partners, and that restriction is explicit in borrower-facing terms.

Evidence checked

Rehypothecation policy (none / partial / full / undisclosed)

Borrower-facing terms language on collateral use

Funding-partner or re-posting disclosures

15% weight10–15%

Custody Model

Who controls the collateral keys — custodial, multisig, MPC, or protocol. The trust axis of every loan.

10/10 standard

Borrower collateral uses a control model where the lender cannot unilaterally move BTC outside documented liquidation or recovery procedures.

Evidence checked

Custody model (custodial / multisig / non-custodial)

Lender type (CeFi / DeFi)

Key-control and recovery documentation

15% weight10–15%

Regulatory Compliance

Legal standing, licensing, jurisdiction, and KYC — whether the product is offered through a clearly regulated structure.

10/10 standard

The product is offered through a clearly licensed or regulated structure, with visible jurisdiction rules, KYC requirements, and no buried access constraints.

Evidence checked

Regulatory status (licensed / registered / offshore / undisclosed)

Jurisdictions served and US-state exclusions

KYC requirements

12% weight10–15%

Reserves & Transparency

Whether collateral or solvency evidence is current, independent, and verifiable.

10/10 standard

Collateral or solvency evidence is current, independently reviewed, easy to find, and specific enough for borrowers to understand what is actually protected.

Evidence checked

Proof-of-reserves published status

Last reserves publication date

Reserves auditor

10% weight10–15%

Track Record

Survival through Bitcoin drawdowns with clean public evidence and meaningful funded volume.

10/10 standard

The lender has a long operating history through multiple Bitcoin drawdowns, meaningful funded volume, and no unresolved borrower-protection red flags.

Evidence checked

Year founded

Total volume funded

Last verification date

10% weight10–15%

Liquidation Practices

Margin-call, cure, partial, and final liquidation rules — whether they are conservative and explicit.

10/10 standard

Margin-call, cure, partial-liquidation, and final-liquidation rules are conservative, specific, borrower-visible, and not dependent on vague discretionary language.

Evidence checked

Margin-call threshold

Liquidation threshold

Liquidation clause summary

Margin-call grace period

8% weight≤ 8%

Loss Protection

Downside protection beyond basic margin-call mechanics, without hidden costs or eligibility restrictions.

10/10 standard

The product adds meaningful borrower downside protection beyond basic margin-call mechanics without hiding material costs or eligibility restrictions.

Evidence checked

Lender notes and highlights

Product notes and other fees

10% weight10–15%

Transparency & Governance

Whether terms, ownership, and governance verify without sales copy or a private quote flow.

10/10 standard

Borrowers can verify terms, ownership, source evidence, governance, public profiles, and correction paths without relying on sales copy or a private quote flow.

Evidence checked

Proof-of-reserves publication and auditor

Public social / ownership profiles

Composite formula

The overall score is a simple weighted sum.

No machine learning, no secret sauce, no editor overrides. Each factor score (0\u201310) is multiplied by its weight, and the results are added together.

Overall = \u03a3 (factor_score \u00d7 factor_weight), rounded to 2 decimals

Rehypothecation Policy: score \u00d7 20%

Custody Model: score \u00d7 15%

Regulatory Compliance: score \u00d7 15%

Reserves & Transparency: score \u00d7 12%

Track Record: score \u00d7 10%

Liquidation Practices: score \u00d7 10%

Loss Protection: score \u00d7 8%

Transparency & Governance: score \u00d7 10%

Example (Unchained, 9.02): every published score is reproducible to within 0.01 by passing a lender\u2019s factor sub-scores through this formula. The methodology output IS the rank.

The 10/10 standard

A 10/10 Bitcoin loan would be nearly boring.

It means a loan where the borrower keeps meaningful control of the collateral; the collateral cannot be reused; reserves or collateral handling are independently verifiable; the legal status is clear; liquidation rules are conservative and explicit; and the lender has survived multiple market cycles with clean public evidence.

The honest caveat

Most real products should not reach 10/10 because every borrowing path adds a tradeoff: custodial exposure, smart-contract risk, jurisdiction limits, quote-dependent pricing, liquidation mechanics, or thinner public proof. That does not make a 7 or 8 weak; it means the product is strong relative to a deliberately strict ideal.

Closest today

Unchained 9.02 — collaborative multisig (borrower controls 1 of 3 keys, with a key-agent co-sign); no rehypothecation; closest to the “keep control of your collateral” ideal.

Xapo Bank 8.83 — regulated bank, human margining, most conservative 40% starting LTV, no rehypothecation.

Aave 8.33 — non-custodial DeFi, algorithmic liquidation.

Why the 10 stays rare

The top products each miss a leg \u2014 Unchained\u2019s live pricing carries a conflicting flag, and Ledn re-posts collateral to funding partners. The perfect-10 is buildable, but no live lender clears every leg today. Live scores currently run 3.39 to 9.02 \u2014 Lava at 3.39 carries a data-quality flag (its custody model is unconfirmable), which is a verification problem, not necessarily a bad product.

The landscape in one view

Fifteen lenders, scored 0\u201310.

Data as of August 23, 2026. Structural facts (custody, rehypothecation, LTV caps, model) are stable; live APRs re-verify. Scores and figures below are read from the loans data layer.

Lender	Structure	Starting LTV	Custody	Rehyp	Score
Unchained	Fixed 12mo commercial	50%	Multisig	None	9.02
Xapo Bank	Fixed 30/90/180/365d	40%	Custodial	None	8.83
Aave	DeFi (WBTC)	70%	Non-custodial	None	8.33
Maker (Sky)	DeFi (WBTC vault)	67%	Non-custodial	None	8.15
Ledn	Fixed 12mo	50%	Custodial	Partial	7.41
Figure	Fixed 12mo	50%	Custodial	None	7.10
Arch Lending	Revolving LOC	60%	Custodial	None	7.03
Strike	Fixed 12mo	50%	Custodial	Undisclosed	6.35
Coinbase / Morpho	Revolving USDC	75%	Non-custodial	None	6.32
Debifi	Marketplace	70%	Multisig	None	6.08
SALT Lending	Fixed 12/36/60mo	50%	Custodial	Partial	5.53
Nexo	Revolving credit line	50%	Custodial	Partial	5.25
Moon Mortgage	Fixed mortgage	30%	Custodial	Undisclosed	4.91
YouHodler	30-day daily-fee	90%	Custodial	Partial	4.02
Lava	Revolving LOC	50%	Custodial	None	3.39

1. LTV conventions

CeFi starting LTV clusters at 40\u201360% (Xapo 40% most conservative; Ledn/Unchained/Figure/Strike/Nexo/Lava 50%; Arch 60%; SALT tiers to 70%; YouHodler up to 90% \u2014 highest and thinnest). DeFi runs materially higher (Aave 70%, Maker 67%, Coinbase/Morpho 75%) because liquidation is automated \u2014 the protocol can instantly seize collateral, so it tolerates less cushion. The buffer between starting LTV and the liquidation threshold is the real risk metric.

2. The custody spectrum (the trust axis)

Custodial (Ledn, Figure, Arch, Nexo, YouHodler, Xapo, Strike, SALT, Lava, Moon) \u2192 Multisig/collaborative (Unchained, Debifi) \u2192 Non-custodial/smart-contract (Aave, Maker, Coinbase/Morpho). Rehypothecation is the quiet differentiator: stated none = Unchained, Figure, Arch, Xapo, Aave, Maker, Morpho; partial/contract-dependent = Nexo, YouHodler, SALT (marketing says none but risk disclosure permits re-pledging \u2014 Pledge flags the conflict), and Ledn (marked Partial \u2014 may re-post collateral to institutional USD funding partners while stating it does not lend client BTC for yield; Pledge flags the same tension); undisclosed = Strike, Moon.

3. Liquidation \u2014 human vs algorithmic

CeFi is human/managed with cure windows (Xapo human-reviewed warning 50% / formal 65%; Strike 72h cure to \u226465%; Unchained 24h grace; Arch partial at 80% ~2% fee; Ledn 70% notify / 75% alert / 80% forced sale; SALT staged 75/83.33/88 \u2192 90.91). DeFi is algorithmic, instant, no grace (Aave liquidation bots ~78% + 5% penalty; Maker auction; Coinbase/Morpho 86% LLTV + 4.38% penalty). The human-vs-algorithmic trade is central to how Pledge frames liquidation risk.

Data sources

Every data point is traced back to a named source.

Primary sources are official lender pages, regulatory records, and proof-of-reserves reports. Protocol parameters and third-party aggregators are used as cross-checks, not primary sources.

Data source

Official lender product pages

Public loan terms, rate tables, LTV caps, custody descriptions, and liquidation clauses from each lender’s own site. Primary source for product structure.

Data source

Regulatory and licensing records

Licensing databases (e.g. NMLS), bank and lending registrations, and jurisdiction disclosures for verifying regulatory standing.

Data source

Proof-of-reserves and audit reports

Published reserve attestations, auditor names, and on-chain verification where a lender discloses collateral handling.

Data source

Protocol documentation

For DeFi lenders (Aave, Maker/Sky, Morpho), live governance parameters such as liquidation thresholds and loan-to-value limits.

What moves a score

Scores change when the underlying data changes or when the methodology is corrected.

These are the conditions that trigger a score review.

Review trigger

Rehypothecation policy changes

A lender adds or removes collateral re-use, re-posting, or funding-partner language in its terms.

Review trigger

Custody model or custodian changes

A move between custodial, multisig, MPC, or non-custodial structures, or a change in the custodian itself.

Review trigger

Regulatory status changes

A new license, a revoked registration, or a change in jurisdiction or US-state availability.

Review trigger

Proof-of-reserves published or withdrawn

A new auditor attestation, publication cadence change, or removal of reserve transparency moves the reserves and transparency scores.

Review trigger

Liquidation terms change

Margin-call thresholds, cure windows, partial-liquidation rules, or final-liquidation terms are revised.

Review trigger

The methodology is corrected

A weight or scoring rule adjustment is noted publicly on this page with an explanation of why.

Limitations

What this methodology does not capture.

A scoring model is only as good as its inputs. These are the known gaps \u2014 documented here rather than averaged away.

Scores are category-relative. A 7.5 loan and a 7.5 treasury are not the same safety claim — read a loan score against other lenders, not across categories.

Data is a dated snapshot. Structural facts (custody, rehypothecation, LTV caps, model) are stable, but live APRs are flagged as reported, not asserted — re-verify rates before relying on them.

Conflicting flags are surfaced, not hidden. Named lenders with conflicting public sources (Unchained pricing, Ledn rate table and rehypothecation, Xapo max cap, SALT rehypothecation marketing-vs-disclosure) are marked rather than averaged away.

Pending flags remain unresolved. Lava, YouHodler, Debifi, and Moon Mortgage carry unresolved custody or terms questions.

Aave and Maker figures are live governance parameters and directional — verify in-app before acting on them.

The two least-quantified factors are among the most safety-critical. Loss Protection (8% weight) is prose-backed only, and Transparency & Governance (10% weight) leans on proof-of-reserves and social links. We score what we can verify and say when the evidence is thinner than we’d like.

The methodology does not capture every smart-contract, oracle, or operational risk in DeFi lending.

Not investment advice. Pledge provides comparison data for education; it does not recommend borrowing or any lender.

Use these pages alongside the methodology.

The methodology connects to the loans comparison, the loans glossary, the main methodology, and the disclosure policy.

Loans

Compare Bitcoin Loans

Side-by-side comparison of Bitcoin-backed lenders with sortable scores, LTV, custody, and product cards.

Glossary

Loans glossary

Plain-English definitions for rehypothecation, LTV ladder, cure window, margin calls, and more.

Methodology

Main methodology

The 8-factor safety methodology behind every Pledge score, across all categories.

Disclosure

Read the disclosure policy

Commercial relationships and how monetization stays separate from rankings.

Not investment advice. Pledge provides comparison data for educational purposes only. All comparison data is a dated snapshot; scores last verified August 23, 2026. Past performance does not guarantee future results.

Affiliate disclosure. Pledge currently does not earn affiliate or referral fees from any product listed on this page. If that changes, monetized links will carry a clear \u2197 AFFILIATE marker.

Data accuracy. Live APRs are flagged as reported, not asserted. Structural facts (custody, rehypothecation, LTV caps, model) are stable and read from the verified loans data layer.

Loans-specific risks. Bitcoin-backed loans carry custody risk, rehypothecation risk, liquidation risk, and (for DeFi) smart-contract and oracle risk. Scores do not capture all relevant risks. Read the full limitations section above.

Compare Other Bitcoin Products

Loans

APR, safety, custody compared

Custodians

Insurance, security, regulation

How Pledge scores Bitcoin-backed loans.

Composite scores range from 0 to 10. 8 weighted factors sum to 100%. Scores last verified August 23, 2026. Past performance does not guarantee future results.

Scoring factors

Each factor is weighted and scored 0–10 before blending into a composite.

Total weight

100%

Factor weights sum to 100%. No hidden adjustments.

Lenders scored

Live scores run 3.39–9.02 across the tracked landscape.

Limitations

Documented

Category-relative caveats, conflicting flags, and data snapshots are listed explicitly.

A 10/10 Bitcoin loan would be nearly boring.

Lender

Structure

Starting LTV

Custody

Rehyp

Score

Unchained

Fixed 12mo commercial

50%

Multisig

None

9.02

Xapo Bank

Fixed 30/90/180/365d

40%

Custodial

None

8.83

Aave

DeFi (WBTC)

70%

Non-custodial

None

8.33

Maker (Sky)

DeFi (WBTC vault)

67%

Non-custodial

None

8.15

Ledn

Fixed 12mo

50%

Custodial

Partial

7.41

Figure

Fixed 12mo

50%

Custodial

None

7.10

Arch Lending

Revolving LOC

60%

Custodial

None

7.03

Strike

Fixed 12mo

50%

Custodial

Undisclosed

6.35

Coinbase / Morpho

Revolving USDC

75%

Non-custodial

None

6.32

Debifi

Marketplace

70%

Multisig

None

6.08

SALT Lending

Fixed 12/36/60mo

50%

Custodial

Partial

5.53

Nexo

Revolving credit line

50%

Custodial

Partial

5.25

Moon Mortgage

Fixed mortgage

30%

Custodial

Undisclosed

4.91

YouHodler

30-day daily-fee

90%

Custodial

Partial

4.02

Lava

Revolving LOC

50%

Custodial

None

3.39