Comparison with Liquity V2 (BOLD)
RAI Dollar started from the Liquity codebase. If you already know Liquity V2 (BOLD), this page is the fastest way to see where RAI Dollar agrees and where it diverges. If you don't know Liquity, skip to How it stays stable. You don't need that background to use RD.
What's the same
- The trove model. Open a trove with collateral, mint stablecoin debt against it, repay later. ICR, TCR, MCR, CCR mean what they mean in BOLD.
- Stability Pools per collateral. Each collateral type has its own Stability Pool. Depositors absorb liquidations on that pool's branch.
- Liquidation paths. Offset against the Stability Pool first; redistribute to remaining troves on the same branch if the Pool can't cover.
- Per-branch trove organization. Ordered for cheap liquidation and redemption walks. (RAI keeps two lists per branch: unshielded and shielded.)
- Gas compensation. Each trove reserves stablecoin gas comp paid to the liquidator. RAI uses 200 RD per trove.
- Hint helpers. Position-changing operations take
upperHint/lowerHintfor cheap inserts. (Front ends compute these for you.)
What's different
| Liquity V2 (BOLD) | RAI Dollar | |
|---|---|---|
| Borrower rate | User-set per trove. Each borrower picks their own interest rate. To avoid being redeemed, borrowers must keep their rate high enough relative to the rest of the market, or pay a delegate service to manage it. | Adaptive, automated. One system rate set by the controller; per-branch utilization offset added on top. No per-trove rate-management decisions. |
| Redemption protection | None. Trove owners can pay a higher rate to reduce their chance of being redeemed against, but there is no guarantee. Redemptions chip away at collateral whenever a trove's rate is among the lowest in the system. | Optional shield. Borrowers pay a small premium to guarantee their trove's collateral is not redeemed against in normal operation. The only path through which a shielded trove can be redeemed against is branch shutdown. |
| Collateral types | WETH, wstETH, rETH | WETH, wstETH, rETH, WBTC. |
| Stability deposit yield | Liquidation gains + interest share, claim manually. | Liquidation gains + auto-compounded RD inflows: borrow interest is routed into SP balances continuously, growing depositors' RD without a claim step. |
| LP rewards | LQTY governance must vote to direct emissions to LP gauges. | Automated. A PI-controlled split sends both FEE emissions and RD fee inflows to LP stakers based on live LP utilization. No vote required. |
| Undercollateralized debt | Branch redistribution shifts bad debt across surviving troves on the same branch, but the protocol has no mechanism to clear undercollateralized debt. It is carried indefinitely. | Automated, layered repayment. When a branch goes underwater the Aggregator funnels three flows into its payDebt: (1) BAD_DEBT_REDEEM_FRAC = 0.5% of every healthy-branch redemption, (2) the redemption fee that would otherwise have been retained by the redeemed trove, and (3) UNDERCOLLATERALIZED_FRAC = 50% of the FEE-staker allocation from GlobalFeeRouter. Solvency is restored without admin intervention. |
| Collateral that can blacklist | No graceful handling. Protocol relies on collateral being non-restrictive. | Per-branch isolation. A blacklist-poisoned branch can be shut down via BorrowerOperations.shutdown(), removing it from cross-branch redemption and freezing its issuance. Other branches keep operating normally. A truly blacklisted collateral cannot wind down (its pools cannot transfer), but isolating it stops the damage from propagating. |
| Governance token | LQTY (V2). Must vote to direct emissions to LP gauges and elsewhere. | FEE. Fee revenue is routed by an automated PI pipeline to maximize FEE-staker yield. No voting required to receive emissions. |
| FEE-equivalent staking | LQTY staking with gauge voting | Three lockup tiers: 7d / 30d / 90d at 1.00× / 1.15× / 1.50× reward multipliers. |
| Emissions curve | BOLD has no protocol-level emissions. For reference, Liquity V1's LQTY issuance was a 1-year-half-life exponential decay, which is extremely front-loaded: roughly 50% of all LQTY in year one, then halving each year forever (asymptotic cap). | Front-loaded, peaks at month 12, terminates at month 47. ~681k FEE/month at launch, climbing to ~1.17M FEE/month at month 12, then declining to the full 31.5M-per-side cap at month 47. Immutable. Full schedule: FEE emissions schedule. |
The biggest day-to-day difference: unredeemable troves
The largest user pain in BOLD is that no trove is safe from redemption. Even when your trove is comfortably above MCR, redemptions chip away at your collateral whenever your rate is among the lowest in the system. The only ways to avoid it are:
- Watching your rate continuously and adjusting it whenever the market median moves.
- Paying a third-party "interest rate manager" to do that for you, typically for a fee.
- Accepting periodic redemptions and rebalancing after each one.
None of these gives you a position that just sits there. Your balance sheet keeps shifting.
In RAI Dollar, the shield removes this entire mechanic. A shielded trove is unredeemable in normal operation — full stop. You pick base or shielded when you open the trove, and the collateral you opened with stays yours. There is no per-trove rate to manage and no third-party service to pay. The only exception is a full branch shutdown, in which case the shielded protection is voided as the branch winds down.
This makes RD usable for "set and forget" collateral positions: long-duration loans, vault-style strategies, or any case where you want predictable collateral exposure rather than active management.
Fee pipeline and emissions
BOLD relies on LQTY governance votes to allocate emissions to gauges. That's flexible but slow: emissions flow only after a quorum, and a passive token holder can sit out and not receive a share.
RAI Dollar's pipeline is fully automated:
- Per-branch
FeeRoutersplits borrow interest between the local Stability Pool and the cross-branchGlobalFeeRouter. Split is PI-controlled by SP utilization (target 25%, bounds 10–60%). GlobalFeeRoutertops up keeper rewards, divertsUNDERCOLLATERALIZED_FRAC = 50%of the staker allocation to clear bad debt if any branch is underwater, then splits the rest between LP staking and FEE staking, again PI-controlled, by LP utilization (target 8%, bounds 10–60%).
Result: FEE stakers receive their share of borrow interest in RD continuously, without any vote or claim window. LP stakers receive both FEE emissions and RD fee inflows. Everyone is paid by the controller's allocation, not by governance fiat. (Redemption fees do not flow to stakers; they are retained by the redeemed trove, or in the bad-debt-sink case routed to that branch's payDebt. See Fees.)
For BOLD-LQTY holders specifically: in BOLD, opting out of governance means opting out of emissions. In RAI Dollar, holding and staking FEE is sufficient. Emissions flow regardless of whether you vote on anything, and there is no on-chain governance vote to participate in.
Stability and the peg
BOLD's peg model relies on redemption arbitrage and on whatever rate the market is willing to pay individual troves. RD adds two more mechanisms:
- An adaptive system rate set by a PI controller on the RD/USD market price (BOLD has no analogous system-wide controller; rates are individual).
- An internal price called par that the controller can adjust within $0.85–$1.20 under sustained stress, widening the redemption-arbitrage spread to help drive market RD back to $1. Par is normally exactly $1.00.
See How it stays stable and How RD stays near $1.