Never sell your bags
to raise cash.

A holder of tokenized equity who needs liquidity has exactly one option today: sell. Selling is the expensive choice in three separate ways, and all three are avoidable.

• It ends the position. Upside after the sale belongs to someone else. For a holder with conviction this is the entire cost.
• It realizes a taxable event. Borrowing against an asset is not a disposal. Selling to raise cash converts a paper gain into a tax bill.
• It is one-directional. Re-entering costs spread and slippage twice, and the price rarely waits.

Traditional finance solved this a century ago with margin loans and securities-based lending. On-chain, the same primitive exists for ETH and stablecoins but not for tokenized stocks — the asset class is too new to have been listed anywhere.

Lending is the largest and most durable category in DeFi. It survived every cycle since 2020 because the demand is structural: leverage and yield are wanted in all market conditions.

• Collateral drives borrow volume. Across major money markets, borrows typically run at a meaningful fraction of supplied collateral. Every new collateral type is additive to that base.
• Tokenized equity is the growth asset class. Real-world assets moved from thesis to shipped product, and Robinhood — a retail broker with tens of millions of funded accounts — is the distribution.
• The chain is new. Robinhood Chain has no incumbent money market. Lending is typically the first protocol to matter on a new chain, because every other application needs it.
• First-mover advantage compounds here. Liquidity attracts borrowers, borrowers deepen liquidity, and the resulting rate curve is hard for a later entrant to beat.

Four actions, one pool. Suppliers provide collateral, borrowers draw USDG against it, interest accrues per second, and liquidators keep the book solvent.

Higher-volatility equities carry a lower LTV and a larger liquidation bonus. Parameters are per-asset, never a single blended constant.

Solvency reduces to one number. The health factor is the ratio of risk-weighted collateral to debt, and it is computed per-asset then blended — a mixed portfolio is never scored against a single average.

Worked example — a real position

A borrower supplies 100 AMZN at $200 and 5 TSLA at $90, then draws $9,000 USDG.

Oracle design

• Chainlink feeds, normalized to 8-decimal USD through a single router.
• Staleness and round guards reject a feed that has stopped updating rather than pricing against a frozen quote.
• A market whose feed is unconfigured ships disabled, never mispriced. Feed addresses are never guessed.
• ERC-8056 uiMultiplier() is deliberately not re-applied — Chainlink equity feeds already include it, and double-multiplying would lie dormant until a stock split.

A two-slope jump-rate curve. Rates track utilization, so the pool prices its own scarcity: cheap when liquidity is abundant, punitive when withdrawals are at risk.

Suppliers receive the borrow interest net of the reserve factor, pro-rated across the whole pool:

When a position falls below 1.00, anyone may repay part of its debt and seize collateral at a discount. The bonus is the incentive that makes third parties do the protocol's risk management for free.

• Close factor 50%. A single liquidation may retire at most half the debt, so a small breach does not wipe out the whole position.
• Bonus 7.5–10% by asset — larger where the collateral is more volatile, because the liquidator carries more price risk.
• The protocol takes 10% of the bonus, in collateral. Not from the pool's cash, which would fund the cut out of supplier liquidity.
• Liquidation cannot worsen a position. Health factor is re-read after state writes and the call reverts on regression.

Two streams, both mechanical. Neither depends on token emissions, inflation, or new deposits to pay old ones.

Why the fee is on interest, not deposits

A deposit fee taxes the side of the market that is hardest to attract. Taxing interest means the protocol earns only when it has actually provided credit — the incentive is aligned with the thing being sold.

Fixed supply of 100,000,000, capped in the contract and fully minted at deploy — no inflation is possible. The token does two things, both live on chain.

• 1 · Borrow-rate discount. Staking cuts the interest paid on every open borrow. A borrower compares the token's cost against a cash saving, which is a valuation anchor rather than a story.
• 2 · Fee share. Staked tokens earn USDG from protocol revenue through a Synthetix-style vault. No lock-up.

The model has one input that matters: total value supplied. Everything else follows from the curve. Three scenarios at a constant 60% utilization, where the borrow rate is 8.0%.

This is a deployed system, not a design document. Every claim below was verified against chain 46630.

• Core loop works end to end — supply, withdraw, borrow, repay, liquidate, all exercised on-chain.
• Guards fire correctly. Over-borrowing reverts ExceedsBorrowPower; liquidating a solvent position reverts NotLiquidatable.
• Staking discount confirmed at 4000 bps with a 39.7% measured interest reduction.
• Oracles fresh across all seven assets, staleness guards passing.
• Frontend in production — five routes, live health-factor preview, mobile responsive.

Ordered by dependency, not by date. Each phase gates the next; nothing here is parallelizable past the audit.

The mainnet script is written and compiles, but deliberately refuses to run until real feed addresses are populated. A market whose feed is unset ships disabled rather than mispriced — a guard, not an oversight.