👁️‍🗨️Vision

Our focus can be summed up in three words: ruthless capital efficiency

Our Vision

We believe that the core challenge with on-chain DeFi today can be boiled down to inefficient utilisation of capital. This is often driven by a variety of factors, such as latency and complexity limitations for on-chain programs. Indeed, the first generation of DEXes, such as Uniswap and Curve, followed a unified, Automated Market Maker (AMM) model: which traded off increased trade precision of orderbooks, for reduced code complexity to be compatible with EVMs. Solana, as a high throughput blockchain with accounts abstractions, enabled the second generation of DEXes, namely orderbook based DEXes, such as Serum, Openbook-dex, and now Phoenix. These significantly advanced the state of the art in on-chain trade settlement; however, many of them are still some way off from their centralised counterparts in terms of capital utilisation and resultant ROI for LPs.

With Fermi DEX, we sought to build a DEX that allows users to place orders without locking liquidity. In current implementations like Serum, liquidity is locked in advance at the stage of placing an order, and remains locked regardless of if and when the order is actually filled. This liquidity could have been earning yield for its owner each second that it has been locked.

This is the idea behind "Just in Time" or JIT Liquidity. With JIT, financial applications are designed with the principal focus of delaying the provision of liquidity to the last possible moment, allowing for highly efficient capital utilisation. A type of JIT has been implemented by Drift v2 on perpetuals trades - they allow for a brief, JIT-style auction for every trade, before the trade is filled by an AMM. As far as we know, no CLOBs based purely on JIT style liquidity have been implemented on-chain so far.

Efficient Market hypothesis

A transaction between two parties consists of a desirable exchange of goods or services, thereby resulting in the increased well being of both participants. An efficient market should therefore assist the discovery and execution of such trades to the maximal degree possible. A corollary is that the marginal cost of expressing a credible intention to make a trade should be reduced as much as possible.

The goal of efficient markets is to optimise for an increased number of exchanges, which can arise when the most number of people are allowed to express their full range of intentions/interest to trade - with these intents being recorded on chain. Any restriction (via imposed cost/opportunity cost) on this ability to express intention necessarily means a reduction in the number of exchanges that are completed - since the set of possible exchanges (as expressed by bids and asks) is lower. As above, this is a market inefficiency that hinders the goal of maximised well being for all participants. Cost vs. Credibility : A tradeoff or independent variables? Most existing on-chain exchanges and markets do not explore the credibility-cost trade-off at all, and simply pick a default binary value of 100% cost upfront, maximizing credibility without regard for the costliness of placing limit orders (intended trades) that may never get filled. Firstly, it is possible to explore many other points along this cost-credibility continuum - for example, being required to lock only 10% of a trades value may free up a ton of capital to express diverse views across a number of assets, instead of being constrained to picking one order to lock all the collateral in - and may come with only a marginal reduction in credibility.

However, Secondly, in a multi turn game theoretic scenario, it is possible to separate credibility from instantaneous cost of trading almost entirely, making them orthogonal axes instead of a trade-off continuum. In such a setup, credibility can be enforced via an independent system of rewards, penalties, and reputation - each of which do not restrict genuine credible traders from expressing their desire to make a trade, while eliminating most of the spam, dishonest, or otherwise non-credible traders. With these considerations in mind, we set out to design Fermi DEX. Design On Fermi, intentions to trade are a first class citizen - the orderbook is composed of credible trade intentions registered by market participants, instead of locked liquidity under unfilled orders. Not being limited by opportunity cost, this allows market participants to express their opinion on a diverse and varied number of scenarios and assets. Relatively unlikely scenarios (or long tail assets) that makers previously would not have bid/asked for, based on a marginal cost : value analysis, now become feasible options. When orders are matched, Fermi works to settle them on a just-in-time basis, focusing on seamless integration and minimal complexity. From a takers perspective, trading on Fermi should feel very similar to trading on another sophisticated CLOB - with the major difference of a potentially deeper orderbook available on Fermi. Fermi abstracts away the complexities of just-in-time settlement to a large degree - leaving the user with only one additional step - calling “finalize” once orders have been filled (this too can be automated in many scenarios).

The end of TVL

Locking value in protocols represents an inefficient utilisation of capital. We aim to minimise TVL to the maximum degree possible, while maintaing deeper orderbooks than the competition - by locking the minimum amount of capital neccesary to ensure that orders placed are credible - but not requiring full capital to be locked upfront as that represents a wasteful allocation of capital.