[BIP-928] Orderly Winddown of Balancer and Distribution of the Treasury

[Marcus]

TL;DR

• I propose an orderly winddown of Balancer: no new business development, a phased sunset of the protocol with a defined exit window, and the DAO closed to the extent legally and practically possible.
• The treasury goes to BAL holders in kind and pro rata, after setting aside what the winddown still needs to spend, in two rounds and a final sweep. Round one opens at the end of May 2027: holders burn their BAL and receive their share of the treasury. Round two is an airdrop to the addresses that redeemed in round one, in proportion to what they redeemed, within two months of the close: the unspent budget, what arrived since, and the share not redeemed. A final sweep six months later distributes anything that arrives after that, to the same addresses.
• The treasury is at least $9M at current token prices, the managed treasury reported by kpk; the full inventory of DAO addresses and positions is published in this thread. The figure that counts is measured and audited at the block round one opens. BAL, and anything that resolves into BAL, is excluded, other than the BAL that goes to tetuBAL holders. The BIP-919 buyback is cancelled.
• Contributor notice runs to 31 October 2026. Pools move to withdrawals only, where the contracts allow it, on 30 October 2026. v3 pools that partners ask to keep running stay live until 30 November 2026.
• This vote approves the direction, the timeline, what happens to what the DAO owns beyond the treasury, the distribution of the treasury to BAL holders including every allocation rule below, a winddown budget of $150k from 1 November 2026 to May 2027, $30k from then to the final sweep, and a reserve of $220k drawn only if needed, replacing the remaining BIP-918 allocation from that date. It also cancels the BIP-919 buyback and supersedes BIP-687.
• The Snapshot has two Yes options, which differ only in the tetuBAL fraction: 50% or 100%. Both count as Yes; the one with more votes sets the fraction.
• Spending to 31 October, including the contributor notice, runs under BIP-918 and is not part of this vote. What is left of that allocation is not carried over as a new budget: it is spent down against the winddown budget below, and whatever remains returns to the treasury.
• The mechanism that implements the distribution is published as an implementation specification, for comment, not for a vote. Transferring anything the DAO owns beyond the treasury, such as code, licences or deployments, gets its own Snapshot vote. Any change to an allocation rule comes back to a vote.

Motivation

Balancer tried. In April, token holders approved a plan to take the protocol to profitability on a restructured base: costs cut, emissions ended, the token model simplified, protocol revenue routed to the DAO, growth expected from v3. They backed it, and the team executed it. v3 is live and operating as designed. Boosted Pools remain a category Balancer defined. AutoRange Pools shipped, the product expected to get us there. New integrations and partnerships were pursued throughout. Some generated interest. None converted into sustained revenue growth. Key people left or stepped back over that period, and the plan ran with less than it was designed with.

The revenue picture is why this comes now. Most of the protocol’s revenue still comes from v2. v3 revenue has not grown to replace it. BIP-918 set a review for exactly this case. This proposal is that review, brought early. Waiting for the calendar would change the numbers, not the conclusion, and every month of waiting is spent from the treasury.

The November 2025 exploit hit legacy v2 pools. v3 is a different architecture, but the event followed the name into every conversation since and made traction harder to build. It is part of why the plan was harder than it looked in April; it is not what this rests on. With a working product suite, a lean structure and full effort from everyone who stayed, the traction the plan required did not come. I do not see a funded path that changes this picture.

Continuing on the current path spends the treasury to arrive at the same place later. That treasury belongs to BAL holders. The question is whether what remains reaches holders while it is still substantial, or is spent first on a path that has already been tried.

Specification

This vote approves the following package. Communication changes from this post; winddown actions wait for the vote.

1. The protocol sunsets in phases. The contracts are non custodial: withdrawing does not depend on Balancer or on anyone continuing to operate. Every step is announced before it happens.

   • Exit window, from this post. Documentation for withdrawing directly from the contracts, and through third party tools that already support it, is published in this window.
   • Withdrawals only, on 30 October 2026. This proposal is the notice. On that date, pools that can be paused are paused and move to withdrawals only; pools that cannot be paused keep working, with the protocol fee set to zero where the contracts allow it. The exception is v3 pools that a partner asks to keep running so it can migrate: requests come in by 16 October, the list is published before 30 October, and those pools stay live until 30 November, when the v3 Vault is paused. Where a pool’s contracts require it, recovery mode is enabled when it is paused, so withdrawals stay open. Pools differ; the treatment of each is published before 30 October. After these dates the protocol does not depend on Balancer operating it, and the frontend, routing and communication treat it as discontinued.
   • Minimal infrastructure. From 1 November, infrastructure steps down to what the v3 pools kept running need until 30 November, and then to a simplified withdrawal interface, the subgraph coverage it needs and public documentation, kept available, including the veBAL unlock path, until the veBAL locks expire and through both distribution rounds.
   • Permission cleanup. Admin permissions, multisig roles and remaining privileged access, including the mainnet DAO multisig and the Emergency subDAO, are inventoried, then revoked or transferred as each becomes safe to retire. Permissions relevant to veBAL unlocks are retired only after the end of May 2027; control of the treasury and the distribution contracts only after the distribution process ends. The end state is a protocol that needs no one from Balancer: the DAO gives up its own privileged roles. Roles that cannot be removed are inventoried and published.

2. Contributor notice and winddown budget. Contributors were given two months of notice on 27 August, running to 31 October, a management decision within the approved budget. Spending to 31 October, including the notice, runs under BIP-918 and is not part of this vote. From 1 November, this proposal’s winddown budget applies: $150k to May 2027, $30k from then to the final sweep, and a reserve of $220k drawn only if needed. It replaces the remaining BIP-918 allocation from that date and is a cap, not a transfer: what is already funded at the OpCo is spent first, and only the balance is drawn. A small transition team on an hourly basis, sized to what the winddown needs, does the remaining work, coordinated by the OpCo. Whatever is not spent returns to the treasury and goes into the distribution.

3. What the DAO owns beyond the treasury. Code, licences, deployments and the like are disposed of by governance, each transfer through its own Snapshot vote. Nothing is handed to anyone by default. Value realized from any of it flows to the treasury. What belongs to the DAO and what sits with the legal entities is established before any such vote. Existing obligations to counterparties are honored through the transition or wound down explicitly with each of them, not dropped. Whatever finds no operator stays open source, with documentation and archives published.

4. The BIP-919 buyback and the BIP-687 earmark. BIP-919 committed the DAO to a voluntary BAL buyback at NAV, capped at 35% of the treasury at that Snapshot, opening twelve months after it. This proposal cancels that buyback and replaces it with the distribution in item 7. The veBAL compensation in the April package was paid in May under BIP-920 and is not affected. BIP-687 ring fences $1M USDC for the bug bounty program. This proposal supersedes it: the earmark is released when coverage ends and the reports open at that date are resolved.

5. Bug bounty. Coverage runs through the exit window and ends on 30 October 2026, the withdrawals only date.

6. Consolidation before round one. Funds in other DAO wallets and sources, including operational safes and fee collection wallets, are consolidated into the Treasury Safe. Open receivables and unclaimed positions in the DAO’s name are recovered. The base reflects everything the DAO holds and is owed, net of what belongs to others, not only what sits in one safe today.

7. Treasury distribution. This vote decides it. In kind, in the tokens actually held, pro rata, in two rounds.

   • Base: measured at the opening snapshot, the block at which round one opens, announced at least two weeks ahead, and audited. The treasury is at least $9M, estimated at current token prices. That is the managed treasury reported by kpk, the DAO’s treasury manager. The DAO holds further assets across other addresses, wallets and positions on several chains. The inventory, with every address and the method, is published in this thread and re-measured at the opening snapshot. The figure moves with prices, costs until then, funds confirmed as belonging to third parties, and recoveries. Positions are moved into yield bearing forms where possible until round one, so the treasury keeps earning, and holders receive what is held at the snapshot. Before round one the smaller and less liquid positions are consolidated so the claim is around six tokens rather than a dozen; round two and the final sweep are consolidated where that reduces the cost of distributing them. kpk proposes the changes under its existing mandate; the Treasury Council approves and signs. BAL, and anything that resolves into BAL, is not a distributable asset.
   • Shares: pro rata over redeemable supply: total supply, minus all BAL held by the DAO and its entities, minus the BAL behind the Tetu permalock, plus the BAL paid from the treasury to tetuBAL holders. BAL lent under BIP-664 is excluded while it is owed to the treasury. Both eligibility and the denominator are fixed at the opening snapshot, and BAL that leaves an excluded address after it is not redeemable.
   • Round one opens at the end of May 2027. By then every veBAL lock that exists today has expired, whether held directly or through auraBAL and sdBAL. tetuBAL is handled below. A lock extended after this post is the holder’s choice; it redeems when it unlocks, within the window or not at all. The claim window runs six months, to the end of November 2027. Redeeming burns the BAL; the claim contract records each address and the amount it redeemed, and that record is the basis for round two. Round one distributes the bulk of the treasury. What is retained is the unspent winddown budget and whatever is not redeemed within the window, which stays in the claim contract and is distributed in round two.
   • veBAL unlocks into 80/20 BAL/WETH BPT. The pool stays exitable throughout; holders exit to BAL and redeem.
   • Wrapper protocols run their own claim calendars. A holder who has not unwound to BAL by the close of round one does not redeem.
   • tetuBAL is an immutable permalock; it never becomes BAL. Both the holder set and the amount are fixed at the block of this post: holders of record at that block, and the BAL behind each tetuBAL measured at that block, so neither can be moved afterwards. Positions held inside contracts, pools and vaults are looked through to the underlying holders at the same block. When round one opens, tetuBAL holders receive, on Ethereum, BAL from the treasury equal to 50% or 100% of the measured amount, as decided in this vote (see Governance path), and can redeem it in the same window as any other holder, counted in the denominator like any other circulating BAL. This is the one case where treasury BAL is distributed rather than excluded. 50% reflects that the lock can never be unwound into BAL, so this adds new redeemable BAL, paid out of what every other holder redeems; 100% treats it as if it could be.
   • Round two is an airdrop to the addresses that redeemed in round one, in proportion to the BAL each redeemed, within two months of the close: the unspent budget, what has arrived since (fees earned before the pause and collected after it, revenue from pools that keep running, recovered receivables, returned funds), and the share not redeemed in round one. No claim is needed and nothing is transferable: an address that did not redeem in round one has no share in round two. Round two assets may be consolidated before distribution if that materially reduces the cost.
   • A final sweep follows, six months after round two, on a date fixed in advance. Whatever has arrived by then is distributed to the same addresses in the same proportion, consolidated first if that reduces the cost. The process ends there.
   • Nothing is paid out before round one. The distribution starts once the winddown budget is funded and set aside; whatever is not spent goes back into the distribution.
   • The mechanism that implements this, the claim contract, the snapshots and the distributions, is published on the forum as an implementation specification by February 2027, for comment. The claim contract is audited before round one. Anything that would change an allocation comes back to a vote.

8. Funds recovered from the attacks on the protocol. They belong to the LPs of the affected pools. They are outside the treasury and outside this distribution, and this proposal does not change how those funds are allocated, which stays as governance approved it. Some of these funds sit in DAO controlled addresses; they are identified and excluded when the full list is published.

9. Recovery from the attacks continues. The investigations continue with private investigators and law enforcement. Anything recovered goes to the affected LPs, not to the treasury. The Balancer Foundation carries this work, overseen by the Treasury Council, until the entities close. Who carries it after that, and on what terms, is decided by its own Snapshot vote before the Foundation closes.

This proposal covers the technical detail at the level a decision needs. Pool by pool, chain by chain, there is more of it, and some of it will only be settled as the inventory is finished. The same applies to the legal entities and the order of their closure. Ask in this thread and I will answer, and the team stays available for questions and support through the exit window and beyond it.

What this means for each group

LPs. Your funds stay withdrawable; withdrawing does not depend on Balancer operating. Plan your exit during the window. On 30 October 2026, pools that can be paused are paused and move to withdrawals only; the rest keep working, with the protocol fee set to zero where the contracts allow it. v3 pools that partners ask to keep running stay live until 30 November 2026, without bug bounty coverage after 30 October; the list is published before 30 October. Withdrawal paths through the interface, third party tools and the contracts directly are documented before 30 October.

BAL holders. You decide this. If it passes, the BIP-919 buyback does not run. From the end of May 2027 you have six months to redeem: you burn your BAL and receive your pro rata share of the treasury in kind. Round two goes to the address you redeemed from, in proportion to what you redeemed, as an airdrop within two months of the close. If you do not redeem in round one, you have no share in round two. What you receive is not BAL and does not resolve into BAL.

veBAL holders. You vote with the BAL underlying your lock, at face value. Round one opens at the end of May 2027, after every lock that exists today has expired. Your lock unlocks into 80/20 BAL/WETH BPT; the pool stays exitable, you exit to BAL and redeem. A lock extended after this post redeems when it unlocks, within the window or not at all.

BAL held through tetuBAL, sdBAL and auraBAL. sdBAL and auraBAL unwind before round one on their protocols’ own calendars; you exit to BAL and redeem directly, like any other holder. A position not unwound to BAL by the close of round one does not redeem. tetuBAL never becomes BAL. Your balance at the block of this post is what counts, and so is the BAL behind each tetuBAL at that block. When round one opens you receive BAL from the treasury, on Ethereum, equal to 50% or 100% of that amount as decided in this vote, and redeem it exactly like any other holder. Positions inside contracts, pools and vaults are looked through to you.

Partners with live pools or integrations. You hear from us directly. Migrate liquidity during the exit window; the withdrawals only date is 30 October 2026, and this proposal is the notice. If you need more time for a v3 pool, ask us by 16 October and it can keep running until 30 November. Obligations are honored or wound down explicitly with you. Requests to carry a deployment or product forward go through their own Snapshot vote.

Bug bounty researchers. The program runs through the exit window and ends on 30 October 2026. After that date the program no longer exists.

LPs affected by the attacks. Recovered funds are yours and stay outside this distribution. A separate BIP will propose how what remains unclaimed after the claim windows, and anything recovered later, is returned to you.

The team. Two months of notice, 27 August to 31 October, then a small hourly transition team through the distribution, coordinated by the OpCo. Everyone leaves with references and introductions from me, and public credit for what they built.

Costs

Baseline. Monthly burn today, all in: about $150k a month. Monthly protocol revenue: about $30k in August, down from $97k in June, most of it from v2. The treasury earns about $25k a month from how kpk manages it, and keeps earning through the winddown.

To 31 October. Spending runs under BIP-918 as approved. The notice period, contracts running normally, is about $300k (September + October), shown in the transparency reports so the full cost is visible. It is not part of this vote.

Winddown budget, from 1 November. $150k to May 2027, $30k from then to the final sweep, and a reserve of $220k drawn only if needed. Total $400k. It replaces the remaining BIP-918 allocation from 1 November.

It buys the minimal exit stack, documentation, permission cleanup, unlock support, both distribution rounds, the final sweep and the closure of the entities. Whatever is unspent goes into the distribution at the next round or the final sweep.

Offsets. Protocol revenue flows to the DAO until the pause; what arrives after it goes to round two. Cancelling the BIP-919 buyback releases whatever was earmarked for it into the distribution; superseding BIP-687 releases what remains of the earmark once coverage ends and open reports are resolved. Proceeds from anything the DAO owns beyond the treasury are not built in; anything realized is added.

The comparison. The winddown costs $400k and ends. Continuation costs the run rate above for as long as it runs. Whatever is not spent from the winddown budget goes back to BAL holders; money spent continuing does not come back.

Runway. The treasury, at least $9M estimated at current token prices, covers this budget with a margin. The argument is not that Balancer is running out, but that what remains belongs to holders and should reach them.

Governance path

The winddown is decided by token holders: forum post, discussion, Snapshot vote on the standard cadence. Voting power is raw BAL held or delegated on every chain where BAL is deployed, plus, on Ethereum, the BAL underlying 80/20 BAL/WETH BPT held directly or locked in veBAL, counted at face value (BIP-921). Positions excluded under BIP-924 carry no voting power. Quorum is 5M BAL (BIP-924).

One vote, plus a Snapshot vote for each transfer of what the DAO owns beyond the treasury. This proposal decides the package above, including every allocation rule of the distribution; the mechanism that implements it follows item 7. Anything that would change an allocation comes back to a vote.

The Snapshot uses four options instead of the standard three, so holders decide the tetuBAL fraction in the same vote: Yes, with tetuBAL at 50%; Yes, with tetuBAL at 100%; No; Abstain. The two Yes options are added together: the proposal passes if their sum is greater than No, with quorum counted over all votes cast. Between the two Yes options, the one with more votes sets the tetuBAL fraction; if they tie, it is 50%.

Communication changes from this post. Winddown actions, including the pause, fee changes, permission revocations and any winddown related movement of funds, wait for the vote.

The Treasury Council controls the Treasury Safe and, under BIP-882, oversees distributions and liquidations proposed by the Foundation Directors and objects to what is not in the ecosystem’s interest. That mandate is oversight and signing, not a standing authority to distribute the treasury to holders. This vote authorizes the distribution. kpk manages the assets under the Treasury Council’s control; the Council supervises the opening snapshot and signs; the Foundation executes as agent of the DAO. The Foundation is therefore the last entity to close, after the distribution process ends. The Council remains the signer set until then; any change to its composition goes through governance, as it does today.

The DAO vote governs the protocol and the treasury. The legal entities execute their own closures under the law that governs them, consistent with the approved plan. Where that law diverges from the plan, the law wins and the divergence is reported to holders.

If this doesn’t pass

A “no” vote leaves the current framework in place: the BIP-918 mandate and its budget, the BIP-919 buyback on its schedule, the BIP-687 bounty, the contracts as they are.

To be clear about what the notice was: contributor contracts end on 31 October, and I gave that notice in August so people had time to look for other opportunities rather than find out with the vote. If holders reject this, contributors who want to carry on under the existing mandate can, and it is funded.

Contributors are working on a separate proposal to keep the infrastructure alive under a new name. That work is theirs, and I told them it has my support as a fork. It is posted on its own and decided on its own.

I will not front a continuation plan, and nothing here funds one. Any other direction comes through the same forum process, priced and staffed by whoever brings it, decided by the same holders. Where it needs a handover, the OpCo supports it in order: infrastructure, documentation, permissions and open obligations transferred, not dropped. This vote decides this proposal.

Disclosures

Contributors hold BAL from vesting granted before the April restructuring. BIP-918 contemplated new vesting for continuing contributors; it was never implemented. What has been distributed since relates to the earlier agreements. No new vesting or BAL grant to contributors is proposed here, and I redeem on the same terms as any other holder. I am paid through my notice period like everyone else.

I sit on the Treasury Council and on the multisigs that execute what governance approves, alongside other contributors. I was added to the Council under BIP-918. The Council signs what governance approves; it does not decide this.

The transition team is staffed within the winddown budget; whoever carries the recovery work after the entities close is named, with terms, in its own Snapshot vote. I am not claiming a place in either, and if I end up doing paid work in either it goes through the same approval as anyone else’s, published in the budget or in that vote. I stay available to the transition team either way. I remain on the Treasury Council as a signer until the process ends.

Before this post I told the Treasury Council and the largest holders what I was going to propose, so they could push back before it was public, and asked them not to trade on it. I am disclosing that this happened, not what was said. The proposal has to win in the open.

I author and publish this proposal. Where the DAO votes to transfer something it owns beyond the treasury, any counterparty relationship I have is disclosed in that vote.

Timeline

When	What
27 Aug 2026	Contributor notice starts
14 Sep 2026	This proposal posted
25–29 Sep 2026	Snapshot vote
16 Oct 2026	Deadline for partners to ask to keep a v3 pool running until 30 November
30 Oct 2026	Withdrawals only date: pausable pools paused, recovery mode where the contracts require it, protocol fee to zero on the rest where allowed. v3 pools requested by partners stay live. Bug bounty coverage ends
31 Oct 2026	Notice ends. From 1 November: transition team, minimal exit stack, winddown budget
30 Nov 2026	v3 Vault paused: remaining v3 pools move to withdrawals only
Nov–Dec 2026	Low risk permission revocations
End Feb 2027	Implementation specification published for comment
Before round one	Claim contract audited. Opening snapshot block announced at least two weeks ahead
End May 2027	Round one opens; opening snapshot taken: BAL burned, bulk of the treasury distributed in kind. Unlock related permissions retired after that
End Nov 2027	Round one closes. The record of who redeemed and how much is final
End Jan 2028	Round two airdrop to the addresses that redeemed: unspent budget, later arrivals, unredeemed share
End Jul 2028	Final sweep: whatever has arrived is distributed to the same addresses. The process ends
After that	Treasury and distribution control retired. Entity closures complete

Edited 20 September 2026: the TL;DR and item 7 updated with the denominator change and the consolidation for cost, both explained in the update in this thread.

Edited 24 September 2026: v3 pools that partners ask to keep running stay live until 30 November; the tetuBAL fraction goes to this vote as 50% or 100%, with look through; Snapshot options and counting rule added under Governance path; the reference to a parallel bug bounty proposal removed, as none was posted.

[Marcus]

Context on how we got here, in my own words

[Wise_Enthusiast]

I think winding down would be a huge mistake.

I wrote an alternate proposal to partner with WiseSoft, immediately solving revenue issues:

[BIP-XXX] KEEP BALANCER ALIVE! Partner with Wise, put treasury stables to work

Author: Wise Enthusiast (@Wise_Enthusiast on X), founder of Wise / WiseSoft LLC Conflict of interest: I founded the company on the other side of this deal. I am not asking the DAO for a salary, a BAL grant, or a consulting invoice. WiseSoft makes money if this ships because we keep the spread on additional TVL deposited. Read this as a partnership proposal. Related: Marcus’s orderly winddown proposal. This does not cancel redemption of the rest of the treasury, does not touch exploit recoverie…

[crypto1nvest]

Hi Winding down at the bottom of the market in difficult times is not the best idea.

I always see Balancer as a tool for Index pools, actually there is a huge market on Stock Index . Unfortunatelly Balancer is not live on Robinhood chain.

I would recommand to launch stock pools on Robinhood chain (e.g. Nas/ NVDIA / GOOGLE / APPLE )

[Fernando]

It’s a bittersweet end for Balancer but I fully agree with Marcus’ proposal.

It’s sad on the one hand because we all wished we had more traction and PMF, but it feels fulfilling and rewarding to see how many people supported us and said they used Balancer a lot in their DeFi lives.

I think bringing Balancer to an end is the most serene and practical choice as the team tried all they could to pull off a turnaround after the hack of 2025: but they just could not overcome the burden it caused.

[maxyz.xyz]

(Pitch)fork Incoming

Authors: @zekraken @gosuto

GM Ballers! We are still working out the details, but wanted to let everyone know we are gauging interest for an alternative direction.

Some of our initial comments:

• The urgency in Marcus’ wind-down proposal is not justified. Instead stick to the original timeline of the restructure (BIP-918): gradually pause and wind-down end date Q2 2027

• Explore an official fork/reincarnation. Too many opportunities have been missed in the last months by inaction (dev team fell apart almost immediately; plus there was no actual vision/roadmap) and bad timing (institutional stonks bull market just warming up). Multiple parties have indicated interest already in some kind of continuation

• Let MAXYZ lead this initiative; currently there is no plan for management after 31 October (we count at least 6 TBD votes in Marcus’ proposal that are still undefined). No significant deviation from Marcus’ budget would be required

• Generally aligned on the following:

  • Balancer Foundation/entities/brand/dapp winds down completely

  • BAL token will become redeemable against treasury

Join Us

If we haven’t spoken to you yet, but you want to be a part of this or want to share your thoughts: reach out!

A definitive proposal will be posted before next week.

[0xMaha]

Generally supportive so long as the number or timeline of buyback does not deviate too far from BIP-919. Seems like this proposal generally follows the same timelines. Previously, the NAV was listed at $0.16, so from Aura’s perspective and other veBAL holders, it would be sensible if a buyback number could be listed here too (accounting for current treasury, the positioned inventoried outside of this, and any expenses) even if based on current prices.

Some other notes:

• Voting on this proposal is complicated because most locks are mid way through unlocking. Aura specifically no longer has governance infrastructure or vlAURA to operate with, so I don’t believe it can vote
• Also generally supportive of people wanting to continue running a fork so long as there is a clean break to avoid any delayed liability to the DAO
• Seems like there are a few loose ends created by this proposal that could maybe be cleaned up like what happens to IP etc

[crypto1nvest]

Just launch this pool Index on Base and Robinhood and send BAL emission to the index pools and you will have liquidity / TVL

Apple, Google, Nvidia, Meta , Amazon, Sandisk …

[refri]

Bittersweet is how it tastes to me too. But do you really think time can’t change? Ofc it’s hard to restore trust after the bug in v2, and no one knows if v3 might have other shortfalls, but as longer it stays as more it will gain trust. For me, not a developer but just a user, a couple of years have to go by to gain full trust, no matter if there was a bug before or not. And IMHO Balancer did the communication around the bug very well, so I still see opportunity to recover. But ofc only if the ppl behind this are willing to stay. If the team is in dissolution mode anyway (which I can understand after these hard times) the shutdown might be the best solution.

However this will turn out, thx for building!

[mieszko_thx.network]

@zekraken @gosuto my 2 cents; a fork only makes sense if it’s approached like a startup instead of a ‘blockchain project’ meaning;

1. target a very narrow use case (at first)
2. solve real customer problems, validating demand instead of chasing theoretical innovations
3. launch with initial traction and with committed partners
4. ideally no fully anon team

What is Aave, Rocket Pool planning to do? If one of these could be involved this endeavour would become much more credible. Even a smaller project like Telcoin would do.

And for the love of all that’s holy; don’t launch with complicated tokenomics that will eat up all the teams time. Product market fit first, tokenomics later. If you must a simple gov token distributed to early community and launching partners with no other utility would do fine.

[Mr_Po]

Hey everyone, sharing my two cents as a complete outsider to Balancer.

The Business Case

When I read Marcus’s post, what I see is not a dead business but a lack of optimism.

Balancer is in an extremely healthy state from the runway standpoint.

The idea that somehow the DAO must be cashflow positive less than a year after a hack while sailing through the depths of the DeFi bear market sounds a bit extreme to me.

Even if that were the target, I would argue that Balancer can become cashflow neutral through external factors alone.
I would imagine with crypto getting a bid just the V2 revenue + the treasury yield would already more or less cover the costs, which itself can be cut too.

I also don’t see the security reputation as an unsolvable problem.

There is a way to prove that your contracts are secure: formal verification.
AI tooling is advancing rapidly, and I imagine if you can’t prove that your contracts are safe today, you will be able to do that within a year or two.
That should address what seems to be the main concern.

Another barrier to growth is incentives - and I would argue that this one is as relevant as security posture and the overall market conditions.
BAL is in a great state, however, to reinvent its tokenomics: it’s fully backed - it doesn’t get better than that for a reset.

So yeah, I don’t find a strong reason for Balancer to disappear.

If the old leadership has had enough of the shiteating, that’s fine - find new leadership that hasn’t.

The Winddown Structure

It is on the stewards of the DAO to ensure that governance proposals are fair to all affected parties.

This proposal is not fair to passive BAL holders who do not keep track of the governance - and takes advantage of them.

What makes it worse is that for active holders, voting against this proposal means voting against getting a premium on their shares.

I would imagine if there was a council at Balancer that oversees the governance integrity, it would have a strong case to block this proposal: it exploits the tokenholders who by definition cannot look after their interests.

And doing it this way is absolutely not necessary.
The hard claim is not needed at all to begin with: the same can be achieved by just putting the funds into a concentrated liquidity position.

If existing BAL holders believe they should have a premium on the treasury that comes from dead supply, even that can be achievable through an LP - but with much milder terms for the holders who miss the news.

And this is a guaranteed outcome - there will be people who find out in a year and a half that their BAL bags are worthless and that their share was appropriated based on the liveliness check.

It does not have to happen this way - there are multiple ways to wind down gracefully if that is the choice.
Getting a premium on your share at the expense of rugging your fellow holders is not one of them.

[0xDanko]

Disclosure: I am currently Head of Operations at Balancer OpCo and former Director of the Balancer Foundation, and I’m currently under the contributor notice Marcus described. Nothing here is a Foundation position.

I support the proposal itself. [BIP-918] set a review for exactly this case, and the numbers Marcus posted are the review: ~$150k/month of burn against ~$30k of protocol revenue in August (down from ~$97k in June). Waiting for the Q2 2027 date, as suggested above, would spend roughly $90–100k a month of the treasury. I understand the objection that the timing is bad and that the market may turn, but I don’t think a wind-down should gamble on market conditions.

One change I’d consider before Snapshot is to distribute in stables, not in-kind. [BIP-919] priced the buyback at NAV in stablecoins. This proposal replaces it with a distribution “in kind, in the tokens actually held”, with positions moved into yield-bearing forms until round one. In practice that means a holder’s redemption value in May 2027 depends on where ETH, AAVE, COW and the other non-stable positions trade between now and the opening snapshot, and then on what it costs to sell them. Holders who are not following governance closely carry that risk without having chosen it.

Converting the liquid non-stable positions to stables over a window, under kpk’s mandate and the Treasury Council’s oversight, would:

• give a NAV-per-BAL figure that can be quoted now and stays meaningful, which is what @0xMaha is referring to;
• remove market risk from passive holders between the vote and round one;
• reduce the claim contract to one token, which makes the audit, the round-two consolidation and the final sweep simpler and cheaper;
• keep the treasury earning yield in stables.

Ofc, the cost of this is giving up upside if the market rallies before May, but I think that’s the right trade: the treasury shouldn’t gamble on its own portfolio during a wind-down.

On the fork (tbd) by the @maxyz.xyz, I have to say I still believe in the product and would be excited to keep building; but I also don’t believe the brand can carry it. In fact, internally, I was the first one to float the rebranding idea around after the hack. It turned out winding down the Balancer entities and the brand was/is the right call, and the open-source codes can continue under a new name is a separate path. A clean break matters for the liability the DAO and the entities would otherwise carry forward.

Lastly, on the legal entities closures, the Foundation closes last. The costs presented here (~$700k) were quoted and the proposal makes sense. The DAO should see a good amount of it flowing back to redemption after the corporate liquidation.

Happy to answer questions on the operational side.

[Marcus]

Update: the numbers, and one change to the denominator

Thanks for the comments so far. Here is what was asked for.

The measurement is done. Each BAL redeems against $0.1579 of treasury at today’s prices, above the $0.1487 that BIP-919 priced the buyback at in April. @0xMaha, this is the figure you asked for.

Everything below is read on-chain on 18 September, priced at DefiLlama at 14:00 UTC. It is my own measurement, not an audited figure: the method and every address are here, so anyone can check it, and the opening snapshot is audited before round one.

Distributable base, non-BAL	$9,959,416
Redeemable BAL	63,068,821
Per BAL	$0.1579

This is already net of the winddown budget, which lives in the OpCo Safe and sits outside the base. Today’s value, not a promise: every line is re-measured at the opening snapshot at the end of May 2027.

Where it sits

Treasury Safe (kpk)	$9,127,957
v3 buffers, owned by the Treasury Safe	$397,709
Innovation Fund	$155,692
MerkleRedeem LDO (BIP-922)	$137,709
BizDev	$74,196
Onchain Ltd	$66,152

The MerkleRedeem LDO and BizDev figures are under review until 30 September and may move.

Two things sit outside that base and can add to it. The OpCo Safe holds $763,010 and is the operating account: it carries September and October under BIP-918 and then the winddown budget, with the residual returning to holders. And the treasury keeps earning until round one.

One change: the denominator

The proposal says the denominator is BIP-919’s: total supply minus the treasury and the two BLabs safes. Measured properly, that leaves DAO-held BAL in other safes redeeming against the DAO’s own treasury.

So it becomes total supply minus all BAL held by the DAO and its entities, minus the BAL behind the Tetu permalock, which can never be withdrawn, plus the BAL paid from the treasury to tetuBAL holders, which does redeem. About 9.3M BAL does not redeem, and the effect is to raise the value per BAL. I am flagging it as a change rather than folding it into the arithmetic.

One conditional line: the 600,000 BAL on loan to Amber under BIP-664 is excluded because it returns before round one. If it has not returned by the snapshot, it is circulating and redeems.

On converting to stables

@0xDanko is right that a claim across a dozen tokens is expensive for the holder and to audit. My inclination is to keep the distribution as it is written, in kind. Approving a conversion does not fix when it happens, and somebody would still have to choose the moments to sell over months, with no mandate on price. Distributing what is there keeps that decision out of the winddown.

What the proposal will say either way is consolidate for cost. Before round one the smaller and less liquid positions are consolidated so the claim is around six tokens; round two and the final sweep are consolidated where that reduces the cost of distributing them.

This one is worth hearing from holders before the vote. @0xMaha, @Fernando, it affects you more than most.

On holders who do not follow governance

@Mr_Po is right that this matters. From this post to the close of round one is fourteen months, six of them with the claim open, announced on the forum, on X, in the interface and directly to partners, when it opens and again before it closes.

There is a window at all because an open ended claim means a live contract and a live multisig, indefinitely, which is the opposite of a protocol that depends on no one. And the treasury share that goes unredeemed is not kept by the DAO or set aside for anyone. It is distributed in round two to whoever redeemed in round one, on the same pro rata basis, me and any contributor who holds BAL included.

Three things I have not covered here and will answer properly when asked: the work is front loaded so almost nothing is left after October, with the OpCo coordinating from November; the votes the proposal defers are not open items but the rule that nothing is handed to anyone by default; and everything the DAO owns beyond the treasury, IP included, falls under item 3.

I am editing item 7 in the proposal with the denominator change and the consolidation, and noting the edit.

[risky-pete]

Disclosure first: I’m part of the team behind Sentralis, a crypto portfolio risk and scenario analysis solution. Nobody at Balancer asked or paid for this, and none of it is advice.

@Marcus, on the in-kind versus stables point you asked holders about in #13 (raised by @0xDanko in #12), the useful number is how wide the band around $0.1579 is by the opening snapshot. I ran the distributable base from your gist (every position except HOPR and dust; stkAAVE priced as AAVE, sGHO as USDC) over the 252 days to 31 May 2027, at 21 September prices, which put it at $0.1584 on the same 63.07M denominator. Everything below is a model estimate under those assumptions, not a fact or a prediction.

Held in kind, the value per BAL at the snapshot lands between $0.115 and $0.285 in 90% of paths under a zero-drift model calibrated on the last year’s volatility, median $0.162. A bootstrap that resamples the last twelve months as they happened (AAVE −54%, ETH −41%) gives $0.103 to $0.206, median $0.132. The chance of opening below the $0.1487 that BIP-919 priced is 37% in the first model and 71% in the second. A repeat of October 2025 to July 2026 on this book takes it to $0.108, close to BAL’s own market price. Converting to stables now fixes the number near $0.158 plus what the stables earn, and gives up the upper half of that band. Neither choice is wrong. The vote is choosing between a fixed $0.158 and a band, and the width of the band is the thing to know.

Most of the band is one position. stkAAVE is 27% of the base and 58% of everything that is not a stablecoin. The ETH wrappers (weETH, rETH, wstETH, the WETH buffers) are another 15%, and the two blocks move together (AAVE to ETH correlation 0.75 over the last year, the wrappers at 1.0 with each other). A 50% move in AAVE alone moves the redemption value 13.5%, to $0.137. Shifting correlations to a crisis regime changes almost nothing (−0.2% on daily VaR), because the only diversification in the book is the 53% in stables.

Converting is cheap where it matters. The Safety Module cooldown reads two days on-chain today, and 19,549 stkAAVE sells in about ten days at 5% of daily volume in a stressed tape, for roughly $5k of slippage, or in under two days at 10% of normal volume. The slow parts are the small ones. The 1.23M COW needs about 19 days at the same stressed rate with slippage near 6%, and SAFE, LDO and POL together are under 3% of the base. The position that carries the risk is the cheap one to convert; the consolidation before round one is where the cost sits.

The basis is the gist’s positions and denominator, prices of 21 September, 25,000 paths per model with a fixed seed, a 365-day covariance window, and liquidity from order-book volume only (the ether.fi and Rocket Pool withdrawal queues are not modelled, so the weETH and rETH exits are not quoted). If any line of the data basis is wrong, say so and I will correct it. Happy to re-run with different assumptions at no cost.

Analysis produced with Sentralis, a crypto portfolio risk and scenario analysis solution for everyone. contact@sentralis.io

[Fernando]

I’m quite strongly in favor of distributing in kind as originally planned for 2 main reasons:

1. We are probably at the beginning of what could be the biggest bull market ever. Things could explode up quickly. I think we’d miss out on a huge upside by selling now.

2. Putting myself in the shoes of the partner projects whose tokens we hold (mainly Aave but not only) we are creating downward pressure unnecessarily, given the BAL holders that will redeem might like that exposure. If they want, they can sell the tokens once they redeemed, but the DAO should not make that decision on their behalf.

Also, not having to sell is simpler: things can go wrong when selling big amounts like this and we want to mitigate any liability to the admins of this winddown process.

[humpydumpy007]

On the tetuBAL allocation — this needs to change before Snapshot

Marcus, the 50% allocation to tetuBAL holders is wrong, unjustified, and needs to be fixed to 100% before this goes to a vote. I’m not raising this as a minor tweak. As written, I don’t think this line item should pass.

The permalock was Balancer’s condition, not Tetu’s choice.

Go back to the 2022 whitelist thread. Tetu’s original submission used upgradeable proxies. Governance rejected it — nventuro flagged that the contracts’ behaviour could be arbitrarily changed, and LongJuanSilverado called whitelisting an upgradeable contract a “no go” because it would let Tetu circumvent the whitelist entirely. Tetu’s response was to give up the flexibility: belbix built an immutable dedicated contract, BalLocker, with voting and delegation control handed to a dedicated multisig. That address, 0x9cC56Fa7734DA21aC88F6a816aF10C5b898596Ce, is the exact one solarcurve took to Snapshot, and the one this DAO voted to approve.

The risk that thread was explicitly guarding against was an owner declining to renew the lock, waiting out the year, and selling. Balancer required Tetu to give up that option as the price of integration. That option is exactly what auraBAL and sdBAL holders get in this proposal, in full. tetuBAL is being discounted by half for lacking an exit that this DAO itself removed. That is not a neutral technical fact about tetuBAL’s structure — it is a consequence of a decision Balancer governance made, and Balancer governance should not now profit from it.

No other wrapper takes a haircut, and no basis is given for this one.

sdBAL and auraBAL unwind on their own calendars and redeem in full. Direct veBAL unlocks and redeems in full. tetuBAL alone is marked down, and the number appears in the spec with zero derivation — no valuation, no cost basis, no legal rationale. This proposal asks holders to approve every allocation rule in one vote. An unexplained 50% cut to one specifically named integrator is not something this DAO should wave through by bundling it with everything else.

It contradicts commitments made elsewhere in this same document.

You write that existing obligations to counterparties are honored or wound down explicitly with them, not dropped. You count BAL underlying a veBAL lock at face value for voting under BIP-921 — this DAO has already decided being locked doesn’t reduce what your BAL is worth. The distribution itself is defined as pro rata over circulating supply. The BAL behind tetuBAL is real circulating BAL that was locked on terms this DAO set. Cutting it in half is inconsistent with all three of your own stated principles, not just with fairness in the abstract.

This is a legal problem, not just a governance one.

Tetu accepted an irreversible lock as a direct condition of a whitelist vote this DAO ran and approved. A unilateral 50% impairment of that position, applied to no other integrator, at the exact moment the Foundation is trying to close its affairs cleanly, is precisely the kind of decision that invites a claim from Tetu or tetuBAL holders against the entities behind the Balancer team and Foundation. You are proposing to wind down the DAO and close the legal entities — that is the worst possible time to leave an unresolved, foreseeable dispute with a counterparty who relied on governance’s own terms. Fixing this now costs the treasury a small, known amount. Leaving it as-is risks a claim that outlasts the entities you’re trying to close.

The fix: tetuBAL holders receive BAL equivalent to the full measured amount at the block of this post, redeemable in the same window, counted in the denominator on the same terms as any other holder. Nothing else in the package needs to change.

I’d like a response from you or the Treasury Council on why 50% was chosen before this goes to Snapshot, not after.

[Marcus]

@humpydumpy007, thanks for going through it in detail.

The proposal said tetuBAL never becomes BAL, but it didn’t say why that leads to half. Fair point, and I’ve added it.

On the 2022 thread: governance asked for a contract that couldn’t be upgraded, so the whitelist couldn’t be circumvented by changing its logic later. The permanence was tetuBAL’s own design from the first post, “veBAL locked forever”, with selling tetuBAL in its pool as the way out, and it’s what the Snapshot that approved it described. That history is part of why this proposal brings tetuBAL in at all, where BIP-919 didn’t.

tetuBAL trades around $0.12 on its main pool, on Polygon. At 50%, each tetuBAL receives about 2.54 BAL, around $0.40 at the value per BAL in the update above. That is more than three times what the market pays for it.

It is also the only case where the proposal adds redeemable BAL. auraBAL, sdBAL and direct veBAL holders redeem their own BAL once their locks expire. The BAL behind tetuBAL can never leave its contract, so the proposal gives tetuBAL holders BAL from the DAO’s own, which otherwise would not redeem. Once it does, it redeems against the same treasury as everyone else: about $220k at 50%, about $431k at 100%.

But since this affects what every holder redeems, it’s fair for holders to decide between 50% and 100%. I’ve adjusted the proposal so the Snapshot has both: yes at 50%, yes at 100%, no, and abstain. Both yes options count toward passing, and the one with more votes sets the fraction.

[Marcus]

Timeline update, following partner feedback, including @valdorff’s on the fork thread.

Pools still move to withdrawals only on 30 October. The exception is v3: if you’re a partner and need more time to migrate a v3 pool, ask us by 16 October and it can keep running until 30 November, when the v3 Vault is paused. The list of pools kept running is published before 30 October. For partners like Rocket Pool, most of the liquidity on Balancer is on v3, so that’s where the extra time helps.

Bug bounty coverage still ends on 30 October, for every pool.

If you’re an LP in one of those v3 pools and want out while it’s covered, exit by 30 October. Withdrawals stay open throughout.

Val, it isn’t the Q2 you asked for, but an extra month is what we can keep running with the minimal team after October. The proposal is edited to reflect this.

[Fernando]

I got a report from a token holder that wanted to vote that even though they held BAL for the last many months their voting power is zero.

Can anyone technical verify what’s wrong? Passing their message on:

---

Address:

0x54c3c925b9d715af541b77f9817544bdc663345e

Message:

I want to vote on Balancer’s snapshot but somehow my voting power shows 0 even though I held over 400k BAL on the snapshot day. Something is wrong and this vote has to be redone with this problem corrected.

Signature:

0xb9b2bf7484f0d415bf1a46f8ea419a98067415f004b3f16a3081db7a5c1fb20157482e07cee6752754247aeccf4d7f4c38a2014beff3feb2acd4981252c79e381c

[maxyz.xyz]

Thanks for the report @Fernando. When upgrading this Snapshot space to the 1-BAL-1-vote voting strategy, the delegate override was not taken into account properly. We have fixed this (if A delegates to B and both vote, A’s choice supersedes B) and because of this had to reset the vote. Please see https://snapshot.org/#/s:balancer.eth/proposal/0xeae4f8bab6f2fbfe22cfeae51ec336ef238c82e6e0017ee511ee67995235d53d