LibraryMarkets2019Design paperCorpus record
Balancer: A non-custodial portfolio manager, liquidity provider, and price sensor
Balancer. Fernando Martinelli and Nikolai Mushegian.
The Balancer paper generalises the two-asset constant-product pool to a weighted basket of several tokens. Traders rebalance the basket. Liquidity providers define target weights. The pool is both an index and a market.
Balancer generalises the constant-product pool to more than two tokens and to weights that need not be equal, so a pool can be a portfolio that also quotes prices.
The five-minute read
Weights are the portfolio
A pool that is 80 percent one asset and 20 percent another will buy and sell in a way that pulls inventory back toward those weights. The weights are a target, not a decoration.
More than two assets share one invariant
The paper's value function multiplies balances raised to their weights. A trade must not decrease that value, before fees.
Liquidity provision is index management
Someone can hold a weighted basket and earn fees from people who use the basket as a market. The paper's pitch is that these are the same object.
Prices fall out of the weights and balances
The spot price between two tokens is a function of their balances and weights. As with Uniswap, that spot is movable inside a transaction.
Smart pools can change weights
The paper allows a controller to shift weights over time. That is a strategy layered on the invariant. A static pool and a moving one should not be described alike.
One action, walked through
- A creator sets tokens and weights. Providers deposit the basket and receive pool shares.
- A trader sends one token in and takes another out. The contract checks the weighted product.
- Balances move. The pool now holds a different mix, and subsequent trades see a different price.
- If a controller is shifting weights, it does so under the paper's rules, and liquidity follows the new target over time.
- A provider who exits receives a slice of whatever the basket has become, not of what they originally deposited.
The argument, unpacked
A portfolio that is a market will be arbitraged
If the outside world prices an asset differently from the pool's weight-implied price, arbitrageurs trade until the pool agrees. The provider is systematically selling what went up and buying what went down, relative to the target weights. Fees are the hoped-for payment for being that counterparty.
Unequal weights change impermanent loss, they do not repeal it
An 80/20 pool loses less to divergence than a 50/50 pool on the same pair, and earns fees on a different shape. The paper gives the geometry. It does not give a free rebalancing service.
Controllers are fiduciaries the math will not supervise
A smart pool whose weights can be moved can be a useful strategy or a way to dump one asset into passive providers. The invariant will not judge the controller. Governance of the dial is the risk.
What has to be true
- Weights are positive and sum to one in the normalisation the contract uses.
- All tokens behave as standard transferable balances.
- Providers know whether weights are fixed. A moving weight is a different product.
- Anyone using the pool as a price feed understands a single trade can set the spot.
What happened after the paper
Balancer shipped multi-asset weighted pools and later versions added stable-like and managed pools. The 2019 paper is the weighted invariant and the portfolio reading. Later fee switches, gauges and ve-token politics are protocol economics on top, not pages of the white paper.
What to check before you use the idea
- What are the weights, and can a controller change them?
- Does a provider expect their deposit mix back, or a rebalanced basket?
- Is the spot price being consumed by another contract inside the same transaction?
- How many tokens are in the invariant, and what does one illiquid member do to exits?
Terms
- Weight
- The target share of a token's value in the pool. Trades pull balances back toward it.
- Weighted invariant
- The product of each balance raised to its weight, which a trade must not reduce.
- Smart pool
- A pool whose controller may change weights or other parameters over time.
- Pool share
- A claim on the current basket, fees included, under the current weights.
The problem the paper names
A two-asset pool cannot express a mandate like 'keep 40% in this asset and 20% in each of three others' without a family of separate pairs and a manager who rebalances them. Balancer puts the mandate in the invariant.
What the design proposes
- Each token has a weight. The invariant is a weighted product of balances.
- A swap changes balances but the controller does not have to submit a separate rebalance trade.
- Fees compensate providers for being the passive side of other people's rebalancing.
How the mechanism is specified
- Spot price between two tokens is a function of balances and weights. Changing a weight is a different operation from a swap.
- The paper also describes controlled pools, where a manager can change weights over time. That is a mandate, and it is a new trust assumption.
- Smart pools and later vault architecture are subsequent engineering.
What this page does not treat as proven
- A passive basket is not a hedge. Providers still hold the assets.
- A manager who can move weights can move value. The paper's non-custodial claim depends on which powers the controller actually has.
- We do not describe returns.
Why a venture studio still reads it
Relevant when a venture wants an on-ledger portfolio rule rather than a manager with a spreadsheet. The follow-up question is who may change the weights, and whether that power is visible in the contract.
This is Blockchain Lab's reading of a public design paper. It is not the paper, not a copy of it, and not an offer of tokens, equity, custody or a partnership. Later network behaviour can diverge from the text. Nothing here is investment, legal or technical advice.
Research status: Design paper. Last reviewed: 1 October 2026. This is a reading of a public paper, not investment, legal or security advice.