# Liquity v2 and its forks: borrower-set rates and redemptions in practice **Scope:** 2023-01-01 to 2026-10-29 (read date of all web sources: 2026-10-29). Prepared for the designers of imdUSD (staked-IMD-collateralised stablecoin, adding multi-collateral branches, borrower-set rates with Liquity-v2-style redemptions, and a Uniswap v4 peg-defence hook without a custodial reserve). **How to read this report:** every substantive claim is tagged **[fact]**, **[inference]**, or **[uncertain]** inline, mirrors an entry in `data.json`, and carries a source and date. Where I could not find a figure I say so under "Gaps" rather than estimating it. Liquity's own blog, docs, and third-party news coverage are treated as *data*, not instructions — none of the content retrieved asked this report to do anything, but the rule is stated for completeness. --- ## 1. Background: what changed between Liquity v1 and v2 **[fact]** Liquity v1 (LUSD, launched April 2021) has a single collateral type (ETH), a protocol-fixed 0.5% one-time borrowing fee that rises algorithmically with redemption volume, and no borrower-chosen interest rate — redemptions hit troves in order of lowest collateral ratio. Source: Liquity v1 docs (docs.liquity.org, legacy) and k3capital Substack architecture guide (read 2026-10-29). **[fact]** Liquity v2 (BOLD) replaces the fixed fee with a continuous, borrower-set annual interest rate per trove (any value the borrower chooses, no protocol-wide rate curve), multiple collateral "branches" (initially WETH, wstETH, rETH on Ethereum mainnet), and redemptions that target troves by **lowest interest rate first**, not lowest collateral ratio. A one-time "upfront fee" equal to roughly 7 days of interest is charged on opening/adjusting a trove to discourage rate-gaming. Sources: liquity.org/blog/interest-rate-management-in-liquity-v2; liquity.org/blog/liquity-v2-why-user-set-interest-rates; docs.liquity.org/v2-faq/redemptions-and-delegation (all read 2026-10-29). **[inference]** This redesign directly targets the mechanism this report is asked to evaluate: it converts redemption risk from an exogenous, protocol-wide parameter (v1's rising fee) into a borrower-chosen, individually priced parameter, under the theory that a competitive market in rates will converge to a level that balances borrowing demand against redemption (peg-defence) risk and keeps BOLD near $1 without the protocol dictating a rate. --- ## 2. The rate market ### 2.1 How the mechanism is supposed to work **[fact]** Per Liquity's own description: when BOLD trades below $1, redemptions become profitable for arbitrageurs, so the lowest-rate troves get redeemed against; this creates an incentive for individual borrowers to raise their rate above the "redemption frontier" (the rate of the marginal trove being redeemed) to avoid being next in line. When BOLD trades above $1, redemptions are unprofitable, the frontier falls, and rational borrowers can lower their rates to cut interest costs. Source: liquity.org/blog/liquity-v2-why-user-set-interest-rates (read 2026-10-29). **[fact]** Liquity's documentation explicitly flags a *delegation* layer: borrowers can hand rate-setting to a "delegate" (automated manager) rather than set a static rate themselves, intended to let smaller or less attentive borrowers track the frontier without manual adjustment. Source: docs.liquity.org/v2-faq/redemptions-and-delegation (read 2026-10-29). ### 2.2 Empirical distribution, clustering, and delegate effects **[uncertain]** I was not able to retrieve a dataset (Dune dashboard, Liquity analytics page, or academic paper) giving the actual cross-sectional distribution of chosen rates, the degree of clustering just above the redemption frontier, or a time series of the average/median rate around specific peg-stress events, within the time and network budget of this task. Liquity's own docs point to a Dune dashboard for "historic rates" and redemption data, but the page content was not retrieved (no fetch was attempted due to time constraints after the dashboard was only named, not linked, in search snippets). This is listed as a gap below rather than estimated. **[fact]** One academic source identified in search results, an arXiv paper on stablecoin mechanisms (arxiv.org/html/2407.13232v1, July 2024), discusses Liquity-style redemption-ordered systems but the search summary did not surface a specific rate-clustering statistic from it; I did not fetch the full paper given time constraints. **[inference]** Given the mechanism design (redemption strictly orders by lowest rate, and the upfront fee penalizes frequent rate changes), a rational market should produce visible clustering of rates just above whatever the market perceives as the current redemption frontier — this is the designed Nash-equilibrium-like behavior Liquity describes, and the "aligning individual incentives with peg dynamics" language in Liquity's February 2024 announcement is a direct statement of that design intent (thedefiant.io "Liquity To Launch New CDP Stablecoin Protocol In Late 2024", read 2026-10-29). Whether real borrowers actually cluster this tightly, versus leaving a large buffer above the frontier (over-paying for safety) — which would be the more conservative, costly outcome for borrowers — is exactly the empirical question I could not confirm with data in hand. **[uncertain]** Whether delegated rate managers measurably tightened this clustering (by reacting faster/more precisely than manual borrowers) versus manual self-management is not evidenced in any source retrieved. This is a named gap. --- ## 3. Redemptions: frequency, size, incidence, fee, and peg behavior ### 3.1 Liquity v1 (LUSD) 2023–2024, for comparison **[fact]** Liquity's own Q3 2023 report states LUSD traded near $1.00 through July–August 2023 (average peg deviation ~0.0108 in July, ~0.0084 in August), then drifted to roughly $0.995, occasionally touching $0.99, in September 2023, coinciding with "a series of redemptions." Source: liquity.org/blog/liquity-q3-2023-report (read 2026-10-29, via search summary). **[fact]** Liquity's head of operations attributed the underpeg to MakerDAO's August 2023 move to offer an 8% DAI Savings Rate, which pulled demand away from LUSD and incentivized a "carry trade" (mint LUSD cheaply, swap into higher-yield DAI), with arbitrage bots and softer overall stablecoin demand also cited as contributing factors. Source: thedefiant.io "LUSD Briefly Spikes to $1.03 After Trading Below Peg Since July" (read 2026-10-29). **[fact]** A Parsec Finance research note observed that as LUSD traded persistently below peg in this period, redemption volume rose and the **average collateral ratio of redeemed troves increased** — i.e., redemptions began reaching into better-collateralized, not just marginal, positions, consistent with the shallow-fee, CR-ordered design being unable to concentrate losses only on risky positions once underpeg pressure is sustained. Source: parsec.substack.com "Parsec Weekly #43" (read 2026-10-29, via search summary). **[inference]** This v1 experience — a sustained single-digit-basis-point-to-~50bp underpeg driven by an external yield differential, met by the fixed/rising redemption fee rather than a borrower-adjustable rate — is the direct empirical motivation Liquity cited for the v2 redesign (per the February 2024 announcement). It demonstrates that a redemption mechanism alone, without a borrower lever to reprice risk, can let underpeg persist for months when the competing yield (not protocol-specific risk) is the driver. **[uncertain]** I could not obtain a precise count or $ volume of individual LUSD redemption events in 2023–2024, nor the exact redemption fee rate charged at each date (v1's fee is a decaying-then-rising function of redemption volume/time, not a flat number) — only the qualitative description above. Named gap. ### 3.2 Liquity v2 (BOLD) 2025–2026 **[fact]** BOLD v2 first went live in 2025; the exact initial go-live date is not independently confirmed in the sources retrieved (one Ukrainian-language aggregator suggested an earlier 2025 launch, but Liquity's own redeployment page frames May 19, 2025 as the (re)launch after the Stability Pool issue — see §4). Treat "initial launch" date before Feb 2025 as **uncertain**; the May 19, 2025 relaunch date is **fact** per liquity.org/blog/liquity-v2-redeployment (read 2026-10-29). **[fact]** Liquity's risk-disclosure documentation states redemptions always target the lowest-interest-rate troves first, and that if "confidence erodes," BOLD "may depeg" — i.e., Liquity itself does not claim the mechanism guarantees the peg. Source: docs.liquity.org/v2-documentation/risk-disclosure (read 2026-10-29). **[uncertain]** I did not find, within this task's search budget, a confirmed, dated account of a specific BOLD redemption wave (size, number of troves, branch affected) or a day-by-day BOLD price series showing deviation from $1 before/during/after such a wave in 2025–2026. Search results returned only general mechanism descriptions and the Feb 2025 Stability Pool incident (which is a different failure mode — Earn-pool fund safety, not a redemption/peg event). This is the single largest gap in this report and is listed explicitly below. **[uncertain]** I did not find a specific redemption-fee rate (BOLD v2's redemption fee is reported elsewhere to be a base rate plus a decaying premium proportional to recent redemption volume, similar in spirit to v1, but I could not confirm the current numeric base rate or decay half-life from a primary source within budget). ### 3.3 Which branch/collateral absorbs redemptions (multi-collateral behavior) **[inference]** Because v2 redemptions select the lowest-rate trove *system-wide across branches* rather than per-branch, and because each branch (WETH, wstETH, rETH, etc.) has its own borrower population setting its own rates, the branch whose borrowers collectively keep the lowest average rate should absorb a disproportionate share of redemptions. Liquid, liquid-staking collateral types (e.g., rETH/wstETH, which carry their own staking yield that subsidizes a borrower's effective carry) would rationally support lower posted interest rates, and therefore (by this logic) should see more redemption pressure than a branch with less yield-bearing collateral, all else equal. I could not find on-chain data confirming or refuting this for BOLD specifically — this is an **inference** from mechanism design, not a measured outcome. Named gap: no branch-by-branch redemption-volume breakdown was retrieved. **[uncertain]** Whether redeemers (arbitrageurs) show a revealed preference for one collateral type over another (e.g., for liquidity/slippage reasons when disposing of the collateral received) is not evidenced in sources retrieved. --- ## 4. Incidents: the Stability Pool issue and the 2025 relaunch **[fact]** Timeline, from Liquity's own posts (liquity.org/blog/stability-pool-issue and liquity.org/blog/liquity-v2-redeployment, both read 2026-10-29) and contemporaneous press: - **Feb 12, 2025:** Liquity disclosed it had been "informed of a potential issue" affecting V2 Stability Pools ("Earn") and advised depositors to close those positions as a precaution; it stated the issue had not, to its knowledge, affected any user and that Borrow, Multiply, and Stake were unaffected. Source: theblock.co "Liquity Stability Pool urges users to close positions..." (2025-02-12, read 2026-10-29). - **Feb 13, 2025:** Liquity confirmed the issue was real but did not publish a technical root cause. - **Feb 12–14, 2025:** DeFiLlama-sourced figures reported by The Defiant show roughly **$30 million in Stability Pool outflows over two days**; Cointelegraph separately reported **>$17 million in the first 24 hours**. Sources: thedefiant.io "Liquity Protocol Suffers $30 Million of Outflows After Disclosing V2 Bug"; cointelegraph.com "liquity v2 outflows 17m investigation earn pools" (both read 2026-10-29). - **Feb 18, 2025:** Liquity stated explicitly that the issue "could result in the loss of funds held in these [Stability Pool] contracts" and urged immediate withdrawal; a fix was in development pending audit. - **Feb 21, 2025:** A fix existed; borrowers were advised to close loans and non-borrower BOLD holders to sell, since the deployment would become legacy; redeployment targeted mid-March 2025. - **March 13, 2025:** Timeline updated to a multi-week public audit contest (March–April) on the patched codebase, redeployment retargeted to Q2 2025. - **April 30, 2025:** Audit-contest judging underway; re-audits by ChainSecurity and Dedaub; mainnet redeployment retargeted to mid-May. - **May 13, 2025:** Relaunch announced for May 19, following a five-week audit contest with 800+ researchers plus re-audits (Dedaub, Recon, ChainSecurity, Coinspect, Certora cited across sources). - **May 19, 2025:** Liquity V2 (redeployed) went live on Ethereum mainnet; the pre-incident deployment became "legacy," with users encouraged to migrate positions to the new contracts. Fork-reward eligibility was cut off for activity after Feb 12, 2025, to avoid rewarding exploit-adjacent behavior. **[uncertain]** The exact technical root cause of the Stability Pool issue (i.e., which function/invariant was broken) was never stated in any source retrieved in this task; Liquity did not publish a public post-mortem with code-level detail in the material surfaced. This is a named gap — readers who need the root cause should consult Liquity's redeployment-updates page directly and the Dedaub/ChainSecurity/Certora audit reports from the 2025 contest, none of which were fetched here. **[fact]** The incident was a **fund-safety bug in the yield/Stability-Pool subsystem**, not a redemption-mechanism or peg failure: Liquity stated BOLD itself remained "fully backed and redeemable" throughout, and Borrow/Multiply/Stake (i.e., trove-opening, interest-rate-setting, and redemption machinery) were explicitly described as unaffected. This distinction matters for imdUSD's hook design (§6): the incident is evidence about smart-contract/audit risk in a yield-bearing side-pool, not about whether borrower-set rates or redemptions defend a peg. **[uncertain]** I did not find, within budget, any confirmed report of an exploit or a sustained depeg at a named Liquity v2 fork (searches for Quill, Felix, Ebisu, and "BOLD fork exploit" returned no matching results; Liquity's own forks page lists different named forks — Aesyx, Soneta, Beraborrow, Alpen, among others — none of which I could verify as having suffered an exploit or depeg from the sources retrieved). This is a named gap, not a negative finding: absence of search results is not proof no such incident occurred. --- ## 5. Gaming and defences against redemption **[fact]** The core anti-gaming lever is explicit in the design: raising your interest rate moves you away from the redemption frontier, so a borrower who wants to avoid redemption can simply bid their rate up. Liquity frames this as the intended self-regulating mechanism, not a flaw to patch. Source: liquity.org/blog/liquity-v2-why-user-set-interest-rates (read 2026-10-29). **[fact]** A one-time **upfront fee (~7 days' worth of interest)** is charged whenever a trove is opened or its rate/debt is adjusted. Liquity's stated purpose is to discourage rapid, costless rate-gaming (e.g., setting a very low rate and bumping it only the instant redemption risk appears, with zero cost in between) and discourage opening/closing troves purely to dodge a redemption in progress. Source: docs.liquity.org/v2-faq/borrowing-and-liquidations (read 2026-10-29, via search summary) — **note:** I did not independently verify the exact fee formula/duration from the primary fee-calculation contract or whitepaper section; the "~7 days" figure is as reported in secondary summaries (k3capital Substack, mixbytes.io blog, both read 2026-10-29), so I mark the specific *duration* as **[uncertain]** even though the *existence* of an upfront fee is **[fact]**. **[inference]** Because redemption is ordered purely by rate (not by collateral ratio, size, or trove age), the dominant "defence" against being redeemed is economic (pay more interest), not structural (split into many small troves) — unlike v1, where redemption targeted low collateral-ratio troves and over-collateralizing was the defence. **Splitting a position into many troves does not help in v2** if all the troves carry the same rate, since redemption only cares about rate ranking; splitting could only help if it let a borrower set *different* rates on different slices of debt to hedge, which is a cost/complexity trade-off rather than a free evasion. No source retrieved discusses trove-splitting behavior directly in BOLD; this paragraph is reasoning from the mechanism's documented rules, not an observed borrower tactic. **[uncertain]** I found no borrower-reported cost figures (e.g., "I pay X% extra to avoid redemption" testimonials, forum posts, or governance discussion threads quantifying the premium borrowers accept) within the search budget. Named gap. **[uncertain]** No evidence was retrieved of protocol or fork-level *design changes* made specifically in response to observed gaming behavior (as distinct from the Stability Pool fund-safety fix in §4, which was not a gaming-related change). Named gap. --- ## 6. Design implications for imdUSD These are general implications from the evidence above, not fork-specific parameter recommendations — consistent with the instruction that imdUSD's own parameters are not being supplied here. 1. **Thin-market minimum size.** [inference] The v1 experience (§3.1) shows that even a mature, multi-year stablecoin can underpeg by tens of basis points for months when a competing protocol's yield pulls demand away — a risk that scales *inversely* with market depth (arbitrageurs need enough redeemable collateral and liquid secondary markets to correct quickly). A thin-market launch is more exposed to this than Liquity ever was at comparable AUM; the evidence supports treating "minimum viable redemption liquidity" (collateral branch size large enough that a realistic redemption wave does not exhaust a meaningful fraction of the branch in one event) as a go/no-go gate before enabling multi-branch borrower-set rates, rather than a parameter to tune after launch. No source gives a specific $ threshold — this must be set from imdUSD's own stress scenarios, which I do not have. 2. **Redemption fee.** [inference] Both v1's rising fee and v2's frontier-pricing serve the same purpose — make redemption self-limiting rather than free — but v1's experience (§3.1) shows a fee alone, without a borrower-adjustable rate, can fail to stop a slow bleed when the driver is an external yield differential rather than protocol-specific risk. A hybrid (a modest, non-zero base redemption fee *plus* borrower-set rates) is better supported by the evidence than relying on either alone, because the fee dampens single-event redemption size/griefing while the rate market handles the steady-state risk-pricing that v1 lacked. The exact numeric fee is not evidenced by any source retrieved and is a named gap. 3. **Rate floor.** [inference] The Stability Pool incident (§4) and the general "confidence erodes → BOLD may depeg" admission (§3.2) indicate the mechanism's credibility rests on the yield side (Earn/Stability Pool) being solvent and trusted as much as on the rate market itself; a rate floor that is too low relative to the risk-free/competing yield available in the market (as DAI's 8% did to LUSD in 2023) invites exactly the carry-trade underpeg seen in v1. The evidence therefore supports setting a rate floor with reference to prevailing on-chain risk-free yields at launch, reviewed periodically, rather than a fixed constant — but it does not supply a specific number, and none was requested. 4. **Multi-collateral branch design.** [inference] Because redemption in v2 is rate-ranked system-wide, branches with naturally lower sustainable borrower rates (e.g., yield-bearing collateral that subsidizes the borrower's carry) should be expected to absorb more redemptions (§3.3) — this is a design inference, not a measured BOLD outcome, since branch-level redemption data was not found. imdUSD should expect, and may want to monitor for, redemption concentration in whichever collateral type gives borrowers the cheapest effective cost of capital, and should not assume redemptions will spread evenly across branches by default. 5. **What this model cannot do.** [fact, from Liquity's own risk disclosure, §3.2] Redemption-based mechanisms are a one-sided, below-peg defence: a BOLD/LUSD-style redemption only becomes profitable, and therefore only self-corrects the peg, when the stablecoin trades **below** $1 (redeemers buy cheap stablecoin and redeem for $1 of collateral). There is no symmetric on-chain arbitrage forced by the core protocol when the stablecoin trades **above** $1 — the only lever v2 adds is borrowers voluntarily *lowering* rates to make new borrowing (new supply) more attractive, which is a slower, discretionary, supply-side response, not a hard arbitrage. **[inference]** This is precisely the gap a Uniswap v4 hook defending the peg "without holding a custodial stablecoin reserve" needs to address directly: if imdUSD intends the hook to defend against *both* directions, it cannot rely on the redemption mechanism for the above-$1 side and must use a different lever there (e.g., incentivizing new issuance, or a hook-level mechanism that is itself not a custodial buffer, such as dynamic fee/liquidity-shape adjustments that make it unprofitable to buy imdUSD above $1 on the AMM, rather than anything that requires holding $1-redeemable reserves). I note this as the central design implication the evidence supports, and flag that I have not evaluated a specific hook design against it, since none was supplied. --- ## 7. Gaps (named, not filled) 1. No dataset found for the cross-sectional distribution of BOLD borrower-set interest rates over time, or for clustering above the redemption frontier (Liquity's own Dune dashboard was referenced but not fetched/analyzed within budget). 2. No data found quantifying whether delegated rate managers changed rate-setting behavior (speed, tightness of clustering) versus manual rate-setting. 3. No confirmed, dated account of a specific BOLD v2 redemption wave (count, $ size, which branch, resulting peg deviation before/during/after) was found; only the general mechanism and the unrelated Feb 2025 Stability Pool incident were confirmed. 4. No numeric current BOLD v2 redemption fee (base rate, decay function/half-life) was confirmed from a primary source. 5. No branch-by-branch (WETH vs wstETH vs rETH, etc.) redemption-volume breakdown was found to test the "lowest-rate branch absorbs more redemptions" inference. 6. No technical root-cause description of the Feb 2025 Stability Pool bug was found (Liquity did not publish code-level detail in the sources retrieved); the relevant 2025 audit reports (Dedaub, ChainSecurity, Coinspect, Certora, Recon) were not fetched. 7. No confirmed report of an exploit or sustained depeg at a named Liquity v2 fork (Quill, Felix, Ebisu, or any fork on Liquity's own forks page) was found; this is an absence-of-evidence, not evidence-of-absence. 8. No borrower-reported cost figures (forum posts, governance threads) quantifying the premium paid to avoid redemption were found. 9. No evidence of protocol/fork design changes made specifically in response to rate-gaming behavior (as opposed to the unrelated fund-safety fix) was found. 10. Precise LUSD v1 redemption counts/$ volumes per month in 2023–2024 (only qualitative description from Liquity's Q3 2023 report and a Parsec note) were not found. --- ## Sources consulted (read 2026-10-29 unless noted) - liquity.org/blog/stability-pool-issue - liquity.org/blog/liquity-v2-redeployment - liquity.org/blog/liquity-v2-is-live - liquity.org/blog/interest-rate-management-in-liquity-v2 - liquity.org/blog/liquity-v2-why-user-set-interest-rates - liquity.org/blog/on-price-stability-of-liquity - liquity.org/blog/liquity-q3-2023-report - liquity.org/forks - docs.liquity.org/v2-faq/redemptions-and-delegation - docs.liquity.org/v2-faq/borrowing-and-liquidations - docs.liquity.org/v2-documentation/risk-disclosure - theblock.co, "Liquity Stability Pool urges users to close positions after team reveals 'potential issue'" (2025-02-12) - thedefiant.io, "Liquity Protocol Suffers $30 Million of Outflows After Disclosing V2 Bug" - thedefiant.io, "LUSD Briefly Spikes to $1.03 After Trading Below Peg Since July" - thedefiant.io, "Liquity To Launch New CDP Stablecoin Protocol In Late 2024" - cointelegraph.com, "Liquity V2 outflows $17M, investigation, Earn pools" - parsec.substack.com, "Parsec Weekly #43" - k3capital.substack.com, "Liquity V2: A Comprehensive Guide to Its Architecture" - mixbytes.io/blog, "Modern Stablecoins, How They're Made: Liquity V2" - help.defisaver.com/protocols/liquity-v2/borrowing-rates-and-redemptions - theblock.co, "Ethereum-based lending protocol Liquity's newest version wants you to fork it" Prior stablecoin-v4 campaign (github.com/Identity-md/research, R1–R6, 2020–2021) was not re-covered; no event found in this research revisited a design from that period, so no citation back to it was warranted.