Mastering perp dexes - Part II
Let s now review the different models, briefly describing them and reporting their main features. The concept map of different types of perp dex cost me 72 hours of work, very few hours of sleep and much, too much coffee, I hope you can appreciate it (if you don't like it, please appreciate it anyway haha)
As can be seen from the image above, there are 5 main types of infrastructure developed:
Intent based / centric architecture
AMM-based models, which in turn are divided into the following subtypes: a) vAMMs; b) Counterparty LP based; c) Synthetic assets based; CDPs (technically speaking it is not an AMM, but the synthetic assets creation is performed through LPs - more about Synthetix later)
CLOBs, which are divided into hybrid or fully on-chain based on the "location" of the order book and matching engine system, whether on-chain or off-chain
Hybrid models, meaning that they support multiple models simultaneously with varying degrees and patterns of coordination between them
Aggregation protocols, which integrate the liquidity of different perp dex and essentially offer a routing service (more or less customizable depending on the project), along with other complementary tools/features
1. Intent centric architectures focusing on derivatives are relatively few, but there are nevertheless numerous projects working on building infrastructures which can simplify and generalize on-chain creation and management of intents (the most fascinating are undoubtedly Anoma, SUAVE by Flashbots, Essential and Across - Uniswap). Projects to keep an eye on in this subcategory are:
a) @symm_io, which has developed a new primitive for symmetrical (aka intent based) OTC based derivatives creation, trading and settlement, called the automated Markets for Quotes / Intents (aMFQ) based model. The infrastructure built by Symmio is basically a comprehensive framework, offering:
an intent centric execution environment (through the definition of a set of standard rules for communicating intents)
a bilateral, peer-to-peer Quotes settlement system that isolates each position from the others, reducing overall risks to the infrastructure
a network of Solvers specialized in the type of service offered (mostly MM activities, thus Market Makers)
a network of MEV researchers economically encouraged to act in the interest of stakeholders
Unique features:
no need for LPers (although it supports them through Hedgers, LPing solutions automating MM strategies)
peer-to-peer matching and settlement
Liquidity aggregation by default (Symmio’s main trading face, @IntentX_, leverages the @LayerZero_Labs stack to expand this feature to multiple chian simultaneously )
Request for Quote price finding mechanism
ease of B to B product integration through SDK (+ team support)
potentially oracleless
broad assets offering
high leverage
ease of deployment on other EVM networks
b) @CadenceProtocol, which implemented a model similar to Symmio's at the architecture level (thus remaining intent based and RFQ based) but more advanced because:
it gives Solvers the ability to search for the best routing path wherever it is located (cross chain finding path), regardless of the source chain (thanks to smart wallets with integrated AA, the Intent Pooling Engine collecting users' most recent Intents and allows Solvers to access them, and lastly Symphony, the intent execution network, not far from mainnet launch)
supports LPing yield bearing solutions such as CLPs (because yields are autocompounded) and others (such as RWAs and LSDs) as collateral on the platform
uses decentralized keepers to execute trades (developed by Redstone)
Key features:
cross chain liquidity sourcing & path finding
chain agnostic trading
enhanced UX thanks to AA smart wallets
decentralized keepers
near instant execution speed
proof of lock - Symphony
c) @dappOS_com, an intent execution network which can assign the intent centric nature to any web3 project compatible with one of the developed frameworks. Not a few perp dex have chosen to use this option: these include @GMX_IO, @KiloEx_perp, and @perpprotocol. It differs from previous models because:
it offers a plug and play stack (a network with optimized security mechanism for managing and executing intents, Optimistic Minimum Staking, a growing ecosystem of integrated service providers, and a decentralized governance model)
is not focused on derivatives stuff but is targeted more generally at web3 projects interested in cost efficiency and improving the UX of their user base
Main features:
plug and play stack
intent based UX for web3 needs
chain agnostic functionalities
overall cost reduction
enhanced capital efficiency
AMM-based models were among the first to be developed and still enjoy a very large user base, thanks largely to the early contribution of Perp Protocol (which launched the first vAMM in DeFi) and the vertical growth of GMX in 2023, one of Perp's dexes that most benefited from the FTX collapse (and still holds the first place on Defillama by overall TVL, as well as 12th in the special ranking of the most forked protocols).
The AMM-based model has been revised by many teams and over time has evolved into an articulated subcategory, which we can further divide into:
a) vAMMs, virtual AMMs, which are automated market maker-based models designed to simulate the liquidity depth of an AMM and its operational mechanics (for pricing in and out of trading positions) through the definition of virtual parameters.
What it means? It means that the real assets representing the collateral are not deposited within the vAMM, but are instead managed by a smart contract (a vault) that precisely "simulates" the AMM behavior in the price discovery mechanism with an amount of assets that has been predetermined by the vault creator.
A quick practical example to understand how this works:
The most interesting protocol in the vAMM subcategory is undoubtedly @perpprotocol with its second iteration, V2 (Curie), which unlike the model illustrated above is not based on constant product but instead leverages the concentrated liquidity of Uni V3 as an execution layer to exponentially increase capital efficiency (and drastically reduced price impact in closing and opening trades); for these reasons, it has been defined as a LP driven derivatives AMM (dAMM) based protocol; the benefits are not only for traders, but also for LPers who can thus for the first time provide leveraged liquidity in CL and get insane returns.
Unique features:
extremely high capital efficiency
permissionless market creation
potential listing of TradFi assets
default maker strategies
more fee tiers than currently offered by Uniswap
b) the multi-asset LP - oracle based model developed by @GMX_IO, which relies on oracle price feeds to execute trades and leverages an AMM based pool consisting of a basket of assets (50% stables, 50% approximately ETH and BTC) as a counterparty for spot swaps and leveraged trades opened on the platform; The leverage offered on GMX is somewhat spot because it is created through the borrowing of real assets deposited within GLP.
Unique features:
0 price impact
multi asset LP which pays 70% of trading and liquidation fees to LPers in ETH
Integration of GLP on third-party protocols for developing structured financial services (leverage farming, auto compounding, delta neutral strategies, and so on), which increase composability
c) the synthetic, oracle-based peer to pool AMM model proposed by @perenniallabs, which offers a comprehensive and easily integrated framework (thanks to the SDK developed by the team) for creating two-sided, customizable payoff derivatives as long as it can be derived from the oracle feed itself: indeed the payoff function for a market can be any function over an available oracle feed (a payoff function essentially describes the logic the smart contracts should use to divide the money between the two market sides)
Unique features:
LPers risk exposure limited to the resulting imbalance between long and short positions;
permissionless market creation
customizable payoff functions
every trade is settled in DSU, Digital Standard Unit (100% backed by USDC)
composable nature (low fees, fully on chain operation, ease to integration, team support, etc)
d) the Collateralized Debt Positions model, which is represented at the highest level by @synthetix_io with its new V3 architecture. This is the one and only Liquidity as a Service infrastructure for both the creation and trading of synthetic derivatives accepting collateral from LPers, delegates it to different collateral pools (based on the preferences expressed by the LPers themselves) which then use it to generate a stable synthetic snxUSD (with different collateralization rates based on the types of assets deposited) that will be allocated to different markets. Pool owners enjoy exclusive powers and that's why most users prefer to rely on representative and authoritative entities (such as the spartan group within Synthetix) rather than random ones.
Unique features:
Modular Architecture
Optimized Gas Efficiency
Enhanced Security Features
Decentralized Governance
Atomic Execution
Dynamic Debt Pools
Higher customisation degree for synths
Cross-Chain Capabilities
Streamlined Liquidation Mechanisms
Worldwide reputation