LibraryMarkets2017Design paperCorpus record
Bancor Protocol: Continuous Liquidity and Asynchronous Price Discovery for Tokens through their Smart Contracts
Bancor. Eyal Hertzog, Guy Benartzi, Galia Benartzi.
The 2017 Bancor paper on smart tokens that hold reserves of other tokens and quote a continuous price from a reserve ratio. It is an early automated-liquidity design, distinct from Uniswap's later constant-product pools.
Bancor's paper puts a smart token in the middle of every conversion. The token holds reserves of other assets, and a formula quotes a price from the reserve ratio so a market can exist without a matching seller.
The five-minute read
The token is the market
Instead of an order book, a smart token custodies reserves. Buying the token adds to a reserve and mints supply. Selling does the reverse.
The reserve ratio sets the curve
A connector weight, the fraction of the token's value that should sit in a given reserve, determines how fast the price moves. It is the ancestor of later weighted pools.
Continuous liquidity is the claim
There is always a price, because the formula does not need a counterparty who posted a quote. The depth at that price still depends on how large the reserve is.
Relay tokens pair two reserves
To convert A to B, the paper routes through a smart token that holds both. The user need not hold the relay token afterwards. The hop still passes through its prices.
This is the automated-market-maker ancestor
Later constant-product pools are a special case of this family. Bancor's contribution is the reserve-ratio formula and the smart-token framing, including ambitions the later pools dropped.
One action, walked through
- A smart token is created with a reserve of some other asset and a stated connector weight.
- A buyer sends reserve tokens in. The contract mints smart tokens according to the formula and the price rises.
- A seller returns smart tokens. The contract burns them and releases reserve assets. The price falls.
- A conversion between two outside assets can hop through a relay: sell one side, buy the other, in amounts the two formulas dictate.
- The reserve stays in the contract. There is no order to cancel and no maker to go offline. There is also no one who promised a tight spread beyond what the reserve size supports.
The argument, unpacked
Always-on is not the same as deep
A formula can quote a price for a tiny reserve. The last unit is infinitely expensive in the usual curves, and the first units of a small pool are still easy to move. Marketing that says liquidity is guaranteed should be translated as 'a formula will answer', which is a weaker claim.
The smart token's own price is circular
Price is computed from reserves and supply. If the token is supposed to track a basket, someone must arbitrage it against that basket. The contract does not look at the outside world. A depeg between formula price and market price is corrected by traders, or not at all.
Impermanent loss lives here already
A reserve that rebalances against outside price moves leaves the liquidity inside holding a path-dependent mix. Later AMMs named this more clearly. The arithmetic is in this design.
What has to be true
- Reserve ratios are set and, if adjustable, adjusted by a rule depositors have read.
- Arbitrage against external markets exists where the smart token is supposed to reflect a basket.
- The conversion hop through a relay is still cheaper than the alternatives. Two formulas mean two spreads.
- Reserve tokens behave. A rebasing reserve corrupts the formula's view of inventory.
What happened after the paper
Bancor shipped relay tokens and later insurance-like mechanisms for liquidity providers that are not in the original paper. Uniswap and Balancer took the automated-market idea further into production. Read this paper for the reserve-ratio formula and the claim of counterparty-free quotes. Read later versions separately when they promise protection against divergence loss.
What to check before you use the idea
- How large is the reserve relative to the trade being quoted?
- What is the connector weight, and can it change?
- Is there an external market an arbitrageur would use to correct the formula price?
- Does a later Bancor feature, such as impermanent-loss protection, appear in this paper? It does not.
Terms
- Smart token
- A token that holds reserves and mints or burns itself against them according to a formula.
- Connector weight
- The reserve ratio that sets how sensitive the price is to a trade.
- Relay
- A smart token used as a bridge so one reserve asset can be converted into another.
- Continuous liquidity
- A quote that does not require a resting order. Depth still depends on reserve size.
The problem the paper names
A small token may have no natural counterparty at the moment someone wants to trade. Bancor's proposal is a contract that always offers a price against its reserve, so discovery does not wait for two humans.
What the design proposes
- A smart token names a reserve and a reserve ratio.
- Buying the smart token deposits reserve and mints supply. Selling does the reverse.
- The price is a function of supply and reserve, not of an order book.
How the mechanism is specified
- The reserve ratio determines how violently the price moves as supply changes.
- Connectors between reserves are how the paper imagines paths across tokens. Each hop depends on its own reserve.
- Continuous liquidity is not liquidity at a good price. The formula will quote even when the reserve is nearly exhausted.
What this page does not treat as proven
- Later Bancor products, insurance mechanisms and governance are not the 2017 formula.
- A reserve ratio chosen for marketing rather than for inventory risk will be discovered by the inventory.
- The paper does not make a thinly traded token into a deep market. It makes a formula.
Why a venture studio still reads it
Read Bancor alongside Uniswap v2 when a team says 'automated market maker' without writing an invariant. The two papers are not the same curve, and the difference is the business.
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.