The State of DeFi Perp Exchanges

A not so comprehensive look at the current state of the decentralised perp exchanges sector.

Quick TL;DR

·         Layer 2’s have become the primary platform for DeFi perp protocols.

·         Evolution and innovation in the sector has continued to increase as a new winner emerged in 2022.

·         Although centralised exchanges still dominate perp trading activity, decentralised exchanges have started to cover ground as users gradually migrate over.

Derivatives make up a majority of the market value in the traditional finance world. Estimates suggest a gross market value of around $12.4 trillion for derivatives in 2021. It seemed only natural that the crypto markets would experience a similar sort of rise for derivatives. The expectations came to fruition when BitMex pioneered perpetual futures (perps) in 2016. They very quickly grew to become the most actively traded instrument in crypto.

Centralised exchanges (CEXs) currently dominate this realm with an average of $100B - $200B in daily traded volume across all exchanges over the last year. Comparing that to decentralised exchanges (DEXs), they’ve done an estimated average of $200M - $400M in daily trading volume over the last year.

Although DEX perp exchanges are still lagging CEX’s by a considerable amount, regulatory pressures causing stricter restrictions on CEX’s have led users to migrate to decentralised perp protocols.

·         GMX launched on Arbitrum in September 2021 followed by a subsequent launch on Avalanche in April 2022

·         Perpetual Protocol V2 launched in November 2021

·         Gains Network underwent a revamp and launched on Polygon in November 2021

·         dYdX V4 announced in June 2022 with launch expected in Q4 2022.

A quick refresher

Perpetual protocol pioneered on-chain perps in 2020 with the creation of the Virtual AMM (vAMM). They saw the constant product formula of x*y=k popularized by Uniswap and decided to tweak it for it to be applicable for perps. The virtual part comes from the fact that there is no real asset pool in this AMM, the real assets are stored in a separate smart contract vault while the vAMM holds the virtual assets which makes it purely a price discovery mechanism rather than an asset swapping mechanism.

Although it garnered some initial attention, it was still fairly niche. Within the first 3 months, daily trading volume fluctuated between $2m - $20m. The real explosion for DeFi perps came with the introduction of dYdX.

dYdX introduced the Central Limit Order Book (CLOB) to DeFi perps in early 2021. It essentially gave users that familiar CEX-like trading experience but on-chain. As a result, dYdX took off. In the first 6 months of 2021, they averaged a monthly volume of $1.27B with that number only increasing with the subsequent launch of dYdX v2. October 2021 marked their best ever month with $105B in total trading volume for the month.

The growth that dYdX brought to the DeFi perps space spurred on a new wave of innovation amongst new and existing protocols. This newfound growth got the space closer to reaching that ever elusive target of taking over CEXs. Below, I will go over the main players in the DeFi perp landscape through 2022.

Current landscape

Although the current landscape for perps can still broadly be divided into AMMs vs CLOBs, the real highlight over the last year has been the migration to layer 2s. Faster and cheaper transactions help with a better execution environment making them a necessity for perps.

Optimism, Arbitrum, and Polygon (although it is technically not an L2) are all equally responsible for pushing the DeFi perp space forward. The cheap execution environment made a fully on-chain experience feasible, which meant that AMM perps had their resurgence and could finally challenge the dominant CLOB that is dYdX.

CLOBs –

dYdX –

dYdX still maintains its position as the leading CLOB perp exchange in DeFi (with Mango Markets on Solana being the closes competitor) despite having a relatively weak year in terms of volume. The move to Starkware with dYdX v3 sparked a major uptick in volume and user activity. August is when activity really picked up with the total user count jumping from 20 users to 51k users and volume jumping from $14.2B to $105B within 2 months.

Since its peak in October 2021, volume has been trending downward all the way into October 2022 in line with the rest of the crypto market. The only thing other than total users that has continued to trend upwards through 2022 is the claimed rewards for the DYDX token. The Liquidity staking pool which allows users to passively stake their USDC for DYDX rewards saw a continued increase in claims taking the total amount claimed to just over 127M tokens.

A major upgrade is expected in Q4 2022 with dYdX v4. This is arguably the biggest upgrade to the protocol since its inception. V4 is part of their broader mission to completely decentralize the protocol. Now, dYdX will operate as it’s very own blockchain built using the cosmos-SDK.  

There are two main reasons to why they decided to move away from L2s and build their own sovereign chain. Customisability and scalability. Although L2s do provide high throughput, it is nowhere near sufficient enough. Currently, dYdX can process 10 trades per second and 1,000 order places/cancellations per second. While this is good, it is not good enough in the long run where they intend to attract millions of users. Instead, it made more sense to have an entire network that runs the orderbook off-chain. Each validator of the network will run an in-memory orderbook with placed/cancelled orders propagating through the network like regular transactions. This proves to be at least a 100x more scalable than before according to the team.

Additional benefits from the customizability come in the form of no gas fees. Rather than paying gas fees, traders will simply pay fees for their orders and these fees will accrue back to the validators and stakers. If DYDX is chosen as the fee token then this gives the token an additional mechanism for value accrual. Other functions for the token are still being deliberated upon in the dYdX community forums.

AMMs

Perpetual Protocol –

V1 launched in December 2020 and had a successful first 7 months facilitating over $19B in total trading volume. Despite the good performance, competition in the space meant that Perpetual Protocol had to make improvements. Therefore, V2 launched on Optimism in November 2021 with a whole suite of new upgrades.

Other than the obvious improvement of faster and cheaper transactions on L2, Perpetual protocol leveraged Uniswap V3 as the execution layer for perpetual swaps while also integrating their famous concentrated liquidity model into the protocol. The use of Uni V3 allowed for the creation of a new type of LP position, a leveraged liquidity positions which uses concentrated liquidity.  

Under the hood, this means an LP can deposit USDC with Perpetual protocol and then instruct the clearing house to mint vUSDC with leverage up to 10x. Suppose a user deposits 100 USDC to mint 1000 vUSDC, they have the ability to deploy it as 0.25 vETH and 500 vUSDC (assuming a price of $2000 per ETH). They can then choose the price range to deploy the liquidity in and just like that they’ve created a leveraged liquidity position.

Other features include a cross-margin mode to allow traders to open multiple positions under a common pool of collateral, as well as the permissionless creation of new markets. Since Curie will support Uniswap V3 and Chainlink, any asset that has a price feed with these two protocols can subsequently be listed as a perp with perpetual protocol making it the most diverse perp protocol in terms of asset choice.

Although V2 has only been live during mostly adverse market conditions, it has facilitated a total of $13.7B in volume with a total of 6.1M in total trades executed. However, the standout highlight is the $12.7M in fees collected thanks to their new fee model. A part of the V2 upgrades included taking fees not only from public markets (DAO listed), but also from private markets (user listed) and redistributing them to LPs, Stakers, and the insurance fund. Funds in the insurance fund will also be used in low risk interest protocols to not only grow the fund to protect against extreme conditions but also to redistribute a portion to stakers.

GMX –

As the original teams in the DeFi perps space continued to improve upon their existing products, a new competitor emerged from the shadows in GMX. GMX, which started on BSC as Gambit Financial, quickly rose to popularity after its launch on Arbitrum and Avalanche. It is currently the leading DEX perp protocol after just recently surpassing dYdX in MarketCap ($365M at the time of writing).

Facilitating a cumulative volume of $54.6B since launch across Avalanche and Arbitrum, the rapid success for GMX stems from two major factors. The unique protocol design interwoven with well designed tokenomics.

At the forefront of this design is the shared liquidity model of GLP. GLP is an index of large-cap crypto assets which rebalances every week (similar to the S&P 500). Users can deposit the accepted assets into the pool to mint GLP, and burn GLP to redeem any asset in the pool (doesn’t have to be the same one that was originally deposited). So there is no orderbook, trades can be made at current oracle price using assets in GLP giving the protocol theoretically infinite depth for supported assets. This means large trades can be executed with close to no slippage. This model is similar to Mycelium with their MLP pool but improves upon the model with the unique yet simple token incentive design.

The GLP model creates the simple mechanism of, if traders lose, GLP holders win, and if traders win, GLP holders lose. The main source of this is the borrow fee that is deducted at the start of every hour. The fee amount is based on the utilization ratio of the asset in question and is then paid to the counterparty of the trade, i.e. GLP holders. But that’s not all, a 0.1% fee is taken to open a position and a 0.1% fee is taken to close a position which is also redistributed.

So far, this tokenomic structure has generated around $80m in fees. From all the fees generated by the platform, 70% goes to GLP holders in ETH or AVAX depending on which chain they are using, while also earning esGMX (escrowed GMX). The remaining fees are divided amongst GMX stakers and the GMX price floor fund made to support the GMX price.

The flywheel this model creates is that probabilistically, GLP holders are likely to win most of the time since over 90% of traders lose overtime, this will attract more liquidity to the shared GLP liquidity pool which will allow for more volume, more volume leads into generating more fees for GLP and GMX holders which then incentivizes them to mint more GLP creating a flywheel of growth.

Even though the broader market has been in a bear market, GMX has been one of the safe havens which has seen rapid growth in user activity coupled with price appreciation.

Gains Network  –

Similar to GMX, the Gains network had a meteoric rise this year after their launch on Polygon. Although it launched in January 2021, it didn’t get major traction until its revamp in November 2021. Since then, it has done $13B in total volume generating $13M in fees. Apart from the mechanics, a major differentiator for Gains Network is that they allow leverage trading on stocks (up to 100x) and forex (up to 1000x) in addition to crypto allowing them to tap into a new set of users altogether.

The actual functionality of the protocol is dependent on their main product called gTrade. At the heart of gTrade is the DAI vault which allows all the leverage in the platform to be completely synthetic. Every user is required to deposit DAI as collateral in order to open a trade. Profitable traders get sent money from the DAI vault when positions are closed and negative PnL traders have their collateral sent into the vault when they close positions.

There is no orderbook or sperate liquidity pools for the trading pairs. There is one GNS/DAI pool and the DAI vault which facilitates all trading on the platform. Trading is primarily dependent on their custom Chainlink oracle called the decentralized oracle network (DON). The oracle uses real spot prices and updates them in real time with no funding fee being charged. The idea is to not make their own price like other perp exchanges which charge funding but rather just use existing spot prices, this indirectly makes the platform a bit fairer.

Another noteworthy feature for Gains network is the use of NFTs which no other exchange has done. There are 5 tiers from bronze (lowest) to diamond (highest). The first benefit is that the higher the tier of the NFT you hold, the higher the spread reduction you receive when trading, starting at -15% for bronze going all the way up to -35% for diamonds. The other benefit is that it allows a user to run a bot which executes liquidations and limit orders. This allows the user to generate extra fees because fees from here are given solely to the processor of the liquidation or limit order.

Seeing that the protocol only has 6.5k unique traders, their image of high growth maybe distorted by their tokenomics designed for deflation. The ultimate purpose of GNS is to maximize the amount of DAI that is kept in the system. GNS is paid out as rewards for GNS/DAI LPs, stakers, and NFT bots. To counter this inflation a large chunk is burned every time the DAI vault is above the collateral ratio of 130%. So far, the protocol has burned upwards of 20% of the GNS supply. As lesser DAI leaves the system, more GNS is expected be burnt pushing prices further up.

Despite the rise, it is important to keep in mind that there are still large dependency risks that Gains network faces. First is the dependence on DAI which is the backbone of the entire network. Even though DAI has proven its reliability by surviving multiple cycles, there are no guarantees of the peg being maintained. The lack of diversity leaves the entire protocol susceptible to collapse if any adverse events affect DAI. The other dependence is on Oracles which are the main tool for trade execution. Many have been manipulated in the past which means there is always a risk of them being manipulated in the future. Malicious exploiters could manipulate the oracle to drain the DAI vaults which would be detrimental to the protocol and its users.

Concluding remarks

The need for and potential of decentralised perp markets cannot be denied at this point. It’s been one of the fastest growing sectors in this space with no signs of slowing down anytime soon. Innovation continues with projects such as Kwenta building on top of Synthetix and PerpDEX building the first perp protocol on ZkSync.

The recent resurgence of AMMs is noteworthy as it’s propelled the space closer to the goal of overtaking CEXs when it comes to trading activity. It is impossible to predict whether AMMs or CLOBs will win in the long run but something worth keeping an eye on in the near future is appchains vs L2s. The trend of using existing AMMs such as Uni V3 on L2 as the execution layer gained a lot of popularity, but recently app-chains have started to gain traction due to its unmatched customizability. If dYdX v4 experiences major success, there is a possibility of a major pivot in the space with all the large protocols migrating to their own chain.

Thank you for reading, if you enjoyed then please subscribe to the substack.

Follow me on Twitter

Have a great day. Peace.