TL;DR
YieldBasis holds a Curve Cryptoswap LP (liquidity provider) Depositing two tokens into a trading pool to earn its swap fees. If their prices drift apart, you end up holding more of the one that fell. position under a constant 2x crvUSD loan, so that arbitrage rebalancing can cancel impermanent loss. The Leverage Borrowing to hold a position larger than your capital. At 3x you hold $3 for every $1 of your own, and gains, losses and liquidation risk grow with it. is hard-coded: Factory.LEVERAGE() still reads exactly 2e18. The contracts are non-upgradeable Vyper, and after about 14 months there is still no exploit on record.
This month gave the engine a real test. Between 2 Oct and 7 Nov, BTC fell 12.5% and ETH 19%. All four v3 markets still read between 1.99x and 2.02x leverage. Share prices in the BTC markets moved by less than 0.3% in coin terms.
The verdict stays caution, and the governance concern is no longer only potential. Last month we showed that four wallets funded from the Team vesting contract held about 29% of voting power, just under the 30% quorum. Since then, two of those wallets voted Yes on every DAO proposal that executed. Two of the three proposals would not have reached quorum without them. TVL Total value locked: the dollar value of everything deposited in a protocol or pool. It shows size, not safety. is down 33% to about $93.5M, and YB set an all-time low of $0.0499 on 7 Nov.
Checklist
Audits & contracts. The docs Audit A security firm's review of a protocol's code. It lowers the odds of a bug but does not rule one out, so check what was in scope. page is unchanged since our last report:
- Core: Statemind, ChainSecurity, Quantstamp, Electisec and Pashov, all in 2025.
- DAO contracts: MixBytes, 2025-08-11.
- Security contest: Sherlock, 1 Aug to 30 Sep 2025, scoped to the full protocol.
- Hybrid Vaults: MixBytes and ChainSecurity, 2026-03-26.
- FeeDistributor: a Firepan AI review. It should not be counted as a human audit.
- Newest: Audit #10, ChainSecurity's delta review of AMM and LT hardening, YBLendingOracle and YBPriceProxy, dated 2026-07-24.
Three contracts that gained live roles in October are not on that page:
- YBOraclizedPolicy (0xAEa95d55), the new fee and price policy on the WETH pool.
- LTMigrator (0xeacb6032).
- LTMigratorGated (0x1c8ee3e6).
All three are verified on Sourcify, but this run found no audit that covers them. The audits page does not mention a bug bounty. It gives only a security email address. About 14 months after launch, there is still no public bounty.
Admin control. Re-read on-chain at block 26,365,827 (2026-11-07, about 22:00 UTC). The structure is unchanged:
Factory.admin()is HybridFactoryOwner 0xb8BA33CD, and itsADMIN()is the Aragon DAO 0x42F2A41A.- The voting plugin reads minDuration 604800, supportThreshold 550000, minParticipation 300000 and
votingMode()1. Mode 1 is early execution, and no Timelock A built-in delay between an admin approving a change and the change taking effect, so users can withdraw first if they disagree. follows the vote. Factory.emergency_admin()is still Curve's Emergency DAO 0x467947EE, a Safe that needs 5 of its 9 owners to sign.
The DAO executed three proposals after 2 Oct. Each passed with zero No votes and zero abstentions:
- Proposal 0x982e3d61. Voting ran from 16 to 23 Oct with 61.5M Yes votes, and it executed on 23 Oct at 12:46 UTC. It made LTMigrator 0xeacb6032 a "limit setter" and revoked the older migrator 0xd5b450fd. It also reset crvUSD allocations to 99.999M (WBTC), 100.0M (cbBTC), 35.74M (tBTC) and 62.12M (WETH).
- Proposal 0xd1577b1e. Voting ran from 22 to 29 Oct with 81.7M Yes votes. It executed on 29 Oct at 11:39 UTC, about half an hour before its window closed. It ran a CallComparator check that the WETH pool's policy is the new YBOraclizedPolicy. It also set the WETH market's AMM fee to 1.5%, against 1.2% on the BTC markets. The WETH borrow rate now reads about 2.6% a year, against 1.8% on the BTC markets.
- Proposal 0xe4b257f7. Voting ran from 23 to 30 Oct with 65.1M Yes votes, and it executed on 30 Oct at 11:42 UTC. It made LTMigratorGated 0x1c8ee3e6 a limit setter.
A limit setter can call lt_allocate_stablecoins with a non-zero limit, which raises a market's crvUSD credit line without a vote. The migrator source raises the allocation during a migration and then restores the previous figure. Even so, two contracts that are not on the audit list now hold that power.
Oracle. Factory.agg() is still Curve's AggregatorStablePrice 0x18672b1b. It returned crvUSD at 0.99818 this run, slightly below the 1.00004 we read on 2 Oct. LP valuation is unchanged: Curve pool math multiplied by that aggregator price.
The new element is on the WETH pool. Its policy prices swaps against signed price reports from Pyth Pro, which the docs also call Pyth Lazer. Anyone can deliver a report, and arbitrageurs do so because a fresh report can lower their fee. The docs list three limits:
- The pool clamps the policy's price_scale target to within 20% of its own EMA price.
- Each move is capped in size.
- A stale report stops steering the target.
The policy's owner() is the YB DAO, and its EMERGENCY_ADMIN() is Curve's Emergency DAO. Its fee settings in the docs' deployment configuration:
- base fee 0.50%
- fallback fee 1.75%
- capture share 67.5%
The WBTC, cbBTC and tBTC pools have their own policy contracts, and this run did not identify what type they are. The Pyth feed is a new external price dependency for one market's fees and rebalancing target. It does not feed LP valuation.
Liquidity & exit. All four v3 markets return is_killed() = false on both the LT and the AMM. In each, stablecoin_allocated equals stablecoin_allocation: 99.98M (WBTC), 100.0M (cbBTC), 35.74M (tBTC) and 60.31M (WETH). Actual AMM debt is much lower: 22.0M, 30.2M, 17.6M and 14.6M, for 84.4M drawn against 296M allocated.
The WETH allocation reads 60.31M against the 62.12M set by vote. This run did not trace the transaction that lowered it.
The deprecated v2 markets return disabled_lts = true on HybridFactoryOwner. Each AMM's value_oracle() gives the equity still in them:
- v2 cbBTC: $8.91M
- v2 WBTC: $0.53M
- v2 tBTC: $0.30M
- v2 WETH: $0.08M
That is $9.8M, about 10% of TVL. Disabled does not mean killed, and LPs in these markets can still withdraw. v2 cbBTC stakers still receive YB Emissions and points Emissions are new tokens a protocol issues to pay rewards; they count in the reward yield. Points may turn into tokens later and are never counted in a rate., 3.7% APR Yearly rate without compounding. Funding here is shown as an APR: the hourly rate × 8,760. on DefiLlama.
The exit mechanics have not changed:
emergency_withdrawstill works on a killed market. It has no minimum-output guard and can pull crvUSD from the caller's wallet.- On the WETH pool, the docs say a yb-LP deposit or withdrawal pays half the pool's fee rate on the imbalanced part. While the policy is active, that rate is the 1.75% fallback fee, so the charge is 0.875% on that part.
Before you withdraw, run preview_withdraw and set min_assets.
Yield: real vs emission. Unstaked LPs earn organic fees. pricePerShare now against 2 Oct:
| Market | Now | 2 Oct | Change since 2 Oct | Since 2026-05-25 deploy | Per year |
|---|---|---|---|---|---|
| WBTC | 1.005559 | 1.007478 | -0.19% | +0.56% | ~1.2% |
| cbBTC | 1.007740 | 1.006588 | +0.11% | +0.77% | ~1.7% |
| tBTC | 1.000849 | 1.003425 | -0.26% | +0.08% | ~0.2% |
| WETH | 1.043605 | 1.006034 | +3.7% | +4.4% |
The WETH jump is the largest move we have recorded in any YieldBasis market. It came while most WETH deposits were leaving. The protocol's X account credited the WETH pool's new policy with high volume around 29 Oct, which fits a fee-driven gain. This run did not break the gain down, so its cause stays not verified in this review.
DefiLlama has revised its daily-fee series since our last read. September now reads -$132.6k instead of -$103.6k, and August -$48.4k instead of +$146.6k. On the current series:
- October was +$240.8k, with 18 of 31 days negative.
- 1 to 7 November is -$127.5k, including -$167.7k on 5 Nov.
- The trailing 30 days are +$145.8k.
Staked LPs give up those fees and earn YB emissions instead. DefiLlama shows those emissions at about 2.5% to 4.6% APR on the v3 markets. YB trades at about $0.051 (CoinGecko, 2026-11-07 22:02 UTC), 42% below its $0.0879 price on 2 Oct. It set a new all-time low of $0.0499 on 7 Nov.
Holder concentration. Measured on-chain against a supply of 758.9M YB. Six contracts that Sourcify verifies as VestingEscrow or InflationaryVest hold 390.1M YB, 51.4% of supply. Last month's comparable figure was 54.8%.
| Contract | YB |
|---|---|
| Team vesting 0x93Eb25E3 | 145.76M |
| VestingEscrow 0x11988547 | 72.79M |
| VestingEscrow 0x52544360 | 63.96M |
| InflationaryVest 0x36e36d5d | 60.87M |
| VestingEscrow 0x7ac59227 | 35.50M |
| VestingEscrow 0x1cb8f667 | 11.21M |
Total voting power reads 145.0M veYB. The four Team-claim wallets from last month hold:
- 0xb75bb762: 28.12M veYB
- 0xd10782f4: 10.24M veYB
- 0xff642740: 1.99M veYB
- 0x858a2fef: 1.28M veYB
Together that is 41.6M, or 28.7% of voting power. They also hold 31.7M liquid YB, and locking it would take them to about 41.5%. Only 0x858a2fef added to its lock, topping it up after Team claims. All four wallets are minimal-proxy clones of the same implementation contract. That fits one wallet system, but it does not prove that one person or team controls all four.
What changed is that the votes are now on record. 0xb75bb762 and 0xd10782f4 each voted Yes on all three executed proposals. Each wallet has its own operator address, and in every vote the two transactions landed within one to three blocks of each other. This run found no vote transactions from the other two wallets. The combined 38.4M veYB of the two voting wallets decided two of the three proposals:
| Proposal | Quorum | Yes votes | Yes without the two wallets |
|---|---|---|---|
| 0x982e3d61 | 42.95M | 61.5M | 23.2M, short of quorum |
| 0xe4b257f7 | 43.02M | 65.1M | 26.8M, short of quorum |
| 0xd1577b1e | 42.95M | 81.7M | 43.3M, clears quorum by about 0.4M |
The two wallets cannot pass a vote alone, because 38.4M is below the roughly 43M quorum. In practice they are the swing bloc. At about $0.051 per YB, a 43M quorum is roughly $2.2M of YB, against about $93.5M of user deposits. Red.
Track record. The Factory has been live since 2025-09-15, about 14 months. The v3 markets were deployed on 2026-05-25, about 5.5 months ago. DefiLlama's hacks list has no YieldBasis entry. An X search from 2 Oct to 8 Nov found no exploit, loss, Peg and depeg A stablecoin's peg is the price it targets, usually $1. A depeg is a move away from it, most often down. The stables page checks prices every 5 minutes., stuck withdrawal, market kill or governance dispute. The official posts covered the WETH pool policy and calls to vote.
TVL fell from $139.6M on 2 Oct (DefiLlama live read that day; the daily series point is $137.7M) to $93.5M on 7 Nov. The per-market figures below come from DefiLlama's yield pools, staked plus unstaked:
| Market | 2 Oct | 7 Nov | Change in USD | Change in coins | Read |
|---|---|---|---|---|---|
| WBTC v3 | $40.6M | $22.5M | -45% | about -37% | net withdrawals; $8.4M left after 31 Oct |
| WETH v3 | $27.8M | $14.6M | -48% | about -35% | net withdrawals; most left by 17 Oct |
| tBTC v3 | $21.7M | $18.3M | -16% | about -4% | mostly price |
| cbBTC v3 | $30.2M | $28.4M | -6% | about +8% | net deposits |
| v2 markets | $20.1M | $9.4M | -53% | about -46% | net withdrawals |
If every market had kept its 2 Oct coin balance, these pools would hold about $121M now. So roughly $19M of the decline, about 40%, is price. Roughly $25M to $28M, about 60%, is net withdrawals. These are approximations from DefiLlama pool data, which sums slightly above the protocol-level series.
Between 29 and 30 Oct, BTC fell 3.5% and ETH 3.8%. DefiLlama shows positive fees on both days (+$304k and +$143k). This month's per-share changes show no sign of a material leverage loss. This run could not read per-market state for 29 to 30 Oct directly, because the RPC node it used keeps no archive.
Worst case
Governance is now the sharper risk. One proposal can repoint the price aggregator, raise borrow rates as high as the 100% APR ceiling, reconfigure the WETH pool's price policy, or appoint new limit setters. It needs 30% participation and 55% support, can execute early, and faces no timelock. Two Team-linked contract wallets, voting in lockstep, already decided two of the last three proposals. With their liquid YB, the four Team-claim wallets could reach about 41% of the vote. A vote that is captured, coordinated or simply wrong would take effect before most LPs noticed.
The market risk has not changed in kind. A fast BTC or ETH drop combined with a crvUSD slip lets effective leverage drift away from 2x. Rebalancing then relies on flash Liquidity How much you can sell or withdraw right now without moving the price or waiting. Thin liquidity means a slow or costly exit. and arbitrage at the moment they are hardest to get, and on the WETH pool it also relies on fresh Pyth Pro reports. The only stop is a kill switch held by Curve's Emergency DAO. It leads to an exit with no minimum-output guard that may ask you for crvUSD first.
Curve sits under every layer: crvUSD as the debt, Curve pools as the venue, Curve's aggregator as the Oracle The price feed a protocol trusts to value collateral. A wrong or manipulated price can cause unfair liquidations or let bad loans through. and Curve's Emergency DAO as the kill switch. No insurance fund exists, so position sizing is the only mitigation.
Bottom line
Caution, as last month. The contract design keeps proving itself. Leverage held near 2x through a 12.5% BTC and 19% ETH drawdown, nothing can be upgraded, and the forced-exit key cannot redirect funds.
The verdict turns on who governs those contracts:
- A Team-linked bloc is now the swing vote in a DAO that can change the oracle and the credit lines with no delay.
- New privileged contracts went live in October without a listed audit.
- There is still no bug bounty.
- Fees turned negative again in November.
If you hold it:
- Treat it as an experimental position. Prefer the unstaked position, because staked returns are paid in a token at its all-time low.
- Keep some crvUSD on hand for the emergency path, and set
min_assetson every withdrawal. On the WETH pool, expect a fee of up to 0.875% on the imbalanced part of an exit. - If your position is still in a v2 market, especially v2 cbBTC, migrate or withdraw. Those markets are disabled.
- Watch the DAO's open proposals, and check who votes on them.
Data appendix
- TVL:
- Protocol level: $93.52M on DefiLlama (
yield-basis, no parent, Ethereum only, Leveraged Farming), last point 2026-11-07 20:49 UTC. - Series: $139.58M live read on 2 Oct (daily point $137.67M), $129.68M (16 Oct), $110.28M (30 Oct), $93.99M (6 Nov). ATH $247.30M on 2026-01-15, now down 62%.
- DefiLlama's token breakdown carried a separate CRVUSD line ($33.7M on 2 Oct). It shrank to $13.4M on 5 Nov and has read negative since 6 Nov (-$2.2M now). What it measures is not verified in this review.
- Per-market TVL is from DefiLlama yield pools, staked plus unstaked, as tabled in Track record. v2 on 2 Oct: cbBTC $11.9M, tBTC $4.5M, WBTC $3.6M, WETH $0.15M.
- On-chain equity from
value_oracle(): WBTC v3 $22.19M, cbBTC v3 $29.67M, tBTC v3 $17.86M, WETH v3 $14.62M; v2 cbBTC $8.91M, v2 WBTC $0.53M, v2 tBTC $0.30M, v2 WETH $0.08M; $94.2M in total. - Prices (DefiLlama coins): BTC $86,390 on 2 Oct and about $75,650 now; ETH $2,745 and about $2,230.
- Protocol level: $93.52M on DefiLlama (
- Markets:
market_count()= 11. The v3 LTs are WBTC 0x651D4b81, cbBTC 0x722FC364, tBTC 0x771F7290 and WETH 0x2B9c9f3B, none of them killed.- Leverage (collateral value over equity): 1.99x, 2.02x, 1.99x and 2.00x.
- AMM
fee(): 1.2%, 1.2%, 1.2%, 1.5%.rate(): 1.8%, 1.8%, 1.8%, 2.6% a year. - v2 LTs at market indexes 3 to 6 have
disabled_lts= true and are not killed. - Pool policies: WETH pool 0x656341ef runs YBOraclizedPolicy 0xAEa95d55 (owner YB DAO, EMERGENCY_ADMIN Curve Emergency DAO). The WBTC, cbBTC and tBTC pools have policies 0x0436c064, 0x038f549d and 0xe99ee179, types not identified.
- Fees:
- DefiLlama daily fees, revised series: June +$624.7k, July +$187.6k, August -$48.4k, September -$132.6k, October +$240.8k, 1 to 7 Nov -$127.5k.
- Trailing 30 days +$145.8k; 7 days -$113.1k.
- DefiLlama's 30-day revenue figure reportedly changed from zero last month, which may be a methodology change (not verified in this review).
- Token:
- YB about $0.051 (CoinGecko, 22:02 UTC). Market cap $14.8M, FDV $38.7M, circulating 289.3M.
- Supply 758,912,352 on-chain; max 1B.
- ATH $0.8185 (2025-10-15). ATL $0.0499 (2026-11-07). 30 days -34.0%, 60 days -51.8%.
- Audits:
- Docs audits page: Audits #1 to #10 and the Sherlock contest, unchanged.
- No audit found for YBOraclizedPolicy, LTMigrator or LTMigratorGated (not verified in this review).
- Bug bounty: none listed. The page gives a security email only.
- Whether Audit #10 matches the deployed v3 bytecode: not verified in this review.
- Admin/governance:
Factory.admin()= 0xb8BA33CD1Ccb091a8468572950bD3669723FA5C6. ItsADMIN()= 0x42F2A41A0D0e65A440813190880c8a65124895Fa.- Plugin 0x2be6670DE1cCEC715bDBBa2e3A6C1A05E496ec78: 604800 / 550000 / 300000 / mode 1.
emergency_admin()= 0x467947EE34aF926cF1DCac093870f613C96B1E0c (5 of 9).LEVERAGE()= 2e18.min_admin_fee()= 10%.- Proposals executed since 2 Oct, decoded with
getProposal:- 0x982e3d61 (16 to 23 Oct; 61.55M Yes, 0 No, 0 abstain; minVP 42.95M; executed 23 Oct 12:46 UTC, tx 0x8a8071bc)
- 0xd1577b1e (22 to 29 Oct; 81.72M Yes; minVP 42.95M; executed 29 Oct 11:39 UTC, tx 0x47f46ece)
- 0xe4b257f7 (23 to 30 Oct; 65.15M Yes; minVP 43.02M; executed 30 Oct 11:42 UTC, tx 0x0aba790f)
- Allocations in 0x982e3d61 calldata: 99.9991M, 100.0003M, 35.74M, 62.1237M.
- Limit setters now: 0x1c8ee3e6 and 0xeacb6032 are true, 0xd5b450fd is false.
- Oracle:
agg()= 0x18672b1b0c623a30089A280Ed9256379fb0E4E62,price()= 0.998183. DefiLlama coins crvUSD $0.9979. WETH pool policy oracle: Pyth Pro (Lazer) signed reports, with the bounds described in the docs' pool-policy reference. - Holder concentration:
- The six vesting and reserve contracts as tabled: 390.1M (51.4%). How they map to the docs' tokenomics labels (Investors, Development Reserve, Curve Licensing, Ecosystem) is not verified in this review.
- The veYB contract 0x8235c179 holds about 151.7M YB (Ethplorer).
totalVotes()= 145,032,259. - Team-claim wallets' votes: 28,123,643 / 10,244,514 / 1,992,991 / 1,282,379 = 41.64M (28.7%). Liquid YB: 6,667,466 / 24,546,541 / 499,006 / ~0 = 31.71M. All four are EIP-1167 clones of implementation 0x60043a54.
- Team vesting claims since 2 Oct (Etherscan): 410,635 YB to 0xca09e2e9 (5 Oct), plus four claims to 0x858a2fef. The latest was 56,675 YB on 2026-11-05.
- Yes votes from 0xb75bb762 (operator 0x62a3e8cb) and 0xd10782f4 (operator 0xf846b76f) on all three proposals: Etherscan normal transactions, method 0x9dca59aa, vote option 2. No vote transactions from 0xff642740 or 0x858a2fef since block 26,200,000.
- Whether the four wallets share an owner: not verified in this review.
- Recent-news and incident scan:
- DefiLlama hacks list re-read 2026-11-07 (1,341 entries): no YieldBasis entry.
- X search (Grok x_search), 2 Oct to 8 Nov: no exploit, loss, depeg, kill or governance dispute. Official posts covered the WETH pool policy (22 and 29 Oct) and calls to vote (25 Oct).
Maintained monthly. Methodology: DeFi Research Instruction v2.