Programable Liquidity

What is the maximal capital efficiency a decentralised orderbook can achieve?

Empirical Observations on Spreads and Liquidity Depth

Empirical studies across various asset classes have quantified the contributions of inventory risk and adverse selection to the overall spread. For example, in U.S. Treasury markets, the inside spread is often measured in basis points, with studies reporting that only about 8%--13% of the spread can be attributed to adverse selection, with the remainder compensating for inventory risk and transaction costs. These observations validate the notion that reducing transaction costs and improving capital efficiency can lead to tighter spreads and a more liquid market.

In addition, market microstructure research indicates that the competitive dynamics among multiple liquidity providers tend to compress spreads. In modern electronic trading, where multiple market makers operate with high-speed algorithms, the bid--ask spread can be driven down to near the minimum tick size. This competition also leads to substantial depth at the best bid and ask levels, as market makers are incentivized to provide liquidity aggressively while managing their inventory risk through rapid rebalancing.

Implementation of Programmable Liquidity on Fermi

A key differentiator for Fermi is how it transforms capital efficiency relative to competitors. In AMMs (e.g. Uniswap, Curve), liquidity providers suffer from capital sitting in unused price ranges and from impermanent loss, leading to low utilization of funds. Concentrated liquidity (as in Uniswap v3) improved this, but LPs still cannot use that capital elsewhere concurrently. Order book DEXs before Fermi (e.g. dYdX v3, Openbook, etc.) required users to deposit funds into the exchange contract, meaning those funds were isolated from other uses, much like on a centralized exchange.

Fermi allows the same capital to serve multiple purposes, a feature not present in any competitor at the time of writing. For example, on dYdX (StarkEx version), if a user deposits 1000 USDC to trade, that 1000 USDC sits solely for trading. On Fermi, a trader with 1000 USDC could keep it in a yield farm earning interest, yet still use it to back a limit order — the funds only move if the order actually executes, and even then only for the duration of settlement. This means Fermi effectively has zero opportunity cost for liquidity, whereas all competitors have a positive opportunity cost (funds committed to the exchange or pool cannot earn elsewhere). The result is that Fermi can attract liquidity from more sources and in greater volume.

From an ROI perspective, a market maker on Fermi can earn trading fees and spreads on top of any yield their capital was already earning, potentially achieving much higher returns on capital than a market maker on another platform. This model may draw in capital that would otherwise stay on the sidelines or only in lending, because it can be utilized without sacrificing existing yields. No centralized exchange or other DEX currently offers such composable liquidity; even so-called "hybrid" DEXs or aggregators do not enable one pot of money to simultaneously engage in trading and other activities. This is a fundamental innovation — effectively, Fermi increases the liquidity pie rather than just redistributing it. By doing so, it can attain deeper liquidity than competitors not just by efficiency, but by unlocking additional sources of liquidity (e.g. allowing investors to place passive limit orders with idle assets that are in cold storage or vaults, knowing they won't lose yield unless a trade happens). This advantage in capital efficiency is expected to be the hardest for competitors to replicate, as it requires a ground-up rearchitecture of how orders and settlements interact with user funds.

Committed Liquidity vs. Locked Liquidity

This mechanism of committing liquidity to a trade, instead of actually locking it, allows traders to, first, unlock the time value of money - the same funds can be utilized for other purposes, such as earning yield, or filling other trades, while this quote lies unfilled. Second, it allows traders MMs with limited capital to publish the same liquidity multiply; as long as the liquidity isn't called upon at the same time, this enhances OB depth even given no additional capital.

Multiple claims on the liquidity at the same time do not pose any problem at a protocol level, as we have trade liquidation mechanism to losslessly handle any unfulfilled commitments; they also should not pose a problem to the MMs doing the committing if they follow effective portfolio margining strategies.