LSDfi Reimagined: An Oracleless Future [2/3]
In the previous article, we explored the transformative rise of Liquid Staking protocols, especially following the Shapella network upgrade earlier this year. We discussed the prevailing challenges, centralization risks, and the dominance of players like Lido in the LSD landscape. Now, we’re going to turn our attention to the question of “How can a lending protocol be ‘oracle-free’”?
The Landscape of Lending Protocols
DeFi lending has historically been built around two main architectures:
- Money Markets (e.g., Compound, Aave): These are liquidity pool-based protocols. Lenders deposit assets into a pool, and borrowers take loans from this pool, paying interest to the lenders. The interest rates are algorithmically determined based on supply and demand for the asset in the pool.
- CDP Protocols (e.g., MakerDAO, Liquidity, MIM): Collateralized Debt Positions (CDP) protocols allow users to deposit assets as collateral to mint a stablecoin or another asset. The primary difference here is that there isn’t a direct lender-borrower relationship. Instead, the protocol mints new synthetic assets against the collateral.
Recently, liquidity to fund the supply side of money markets has been hard to come by. In particular, stablecoin liquidity has started exiting DeFi, driven by generally tight macro conditions, as market participants have redeemed stablecoins to chase 5%+ risk-free yields in TradFi. Consequently, CDP-style protocols where synthetic stablecoins are minted have recently grown in popularity, as they don’t require supply-side liquidity at the time of initiating the loan.
A current snapshot of CDP-style protocols is shown here, sorted by collateral TVL:
Maker and Liquity are the incumbents here, but they aren’t growing anymore. All the growth is coming from LSDFi protocols (circled in red) using staked ETH as collateral — i.e. Lybra, crvUSD, Prisma, and Gravita. All these protocols are <5 months old (e.g. Lybra launched in May) and are quickly poised to overtake the incumbents as the available universe of staked ETH collateral continues to expand.
The overall growth of CDP-backed stablecoins is a function of the growth in available collateral. As mentioned in Part 1, staked ETH is the fastest growing collateral type. At Unstable — we wanted to build a protocol that’s uniquely suited to handle the fastest growing collateral type, which is why we’re building a LSDfi lending protocol uniquely capable of supporting any LSD as collateral.
Types of Oracles
Within DeFi lending protocols, oracles serve as mission-critical sources of truth for marking collateral. The type of oracles used by a protocol often dictate the types of collateral that can be accepted, and are the single biggest attack vector for lending protocols. The majority of lending protocol exploits are oracle manipulation attacks. DeFi oracles can be broken down into four main categories: Centralized Market Price Oracles, Decentralized Market Price Oracles, Centralized Fair Price Oracles, and Decentralized Fair Price Oracles.
1. Centralized Market Price Oracles
Definition: Relies on centralized exchange sources for real-time market prices.
Examples: Binance Oracle, Pyth
Pros:
- Rapid data delivery.
- Simpler integration for platforms.
Cons:
- LSDs typically have most of their liquidity on-chain, not on CEXes, making this oracle difficult to obtain for LSDs
- CEX liquidity is transient — market makers tend to pull liquidity and stop quoting during periods of market volatility, making CEX pricing unreliable relative to DEX pricing
2. Decentralized Market Price Oracles
Definition: Gathers real-time market prices from a decentralized set of market sources.
Examples: Chainlink, Redstone, DIA
Pros:
- Price oracle generally represent prices at which collateral liquidation can occur, which helps avoid “bad debt” situations in DeFi lending protocols
Cons:
- Requires substantial on-chain liquidity for eligibility (eg. $3m daily trading volume for >30 days across 3 reputable exchanges, >$25mm DEX liquidity)
- Nearly impossible to obtain for all but the largest LSDs
- LSDs have to waste large amounts of incentives on LSD-ETH liquidity just to gain oracle inclusion — incentives which could otherwise be spent on growing TVL
3. Centralized “Fair Price” Oracles
Definition: A single entity or organization provides the ‘fair’ price based on its own data sources and calculations.
Examples: “Contract rates” reported by LSD protocols and ERC-4626 tokens
Pros:
- Represent the true backing value of an asset that generally can be converted into
- Accurate when the source of the data is reputable or mathematically provable
Cons:
- High risk of data manipulation or single point of failure.
- For LSDs, a compromised centralized oracle can compromise every single protocol that relies on it
4. Decentralized “Fair Price” Oracles
Definition: These oracles use trustless and/or decentralized means to compute and report the fair value backing on-chain assets
Examples: zkOracles (see below)
Pros:
- Minimized manipulation risk as price feeds are not subject to manipulation by a centralized entity
- Ideal for LSDs as they provide a more accurate valuation based on staked ETH validator balances, less susceptible to short-term market volatility.
Cons:
- Can be slower and more expensive due to the computational complexity of trustless methods (i.e. zkSNARKS)
The Rise of “Oracle-free” Protocols
As the DeFi lending landscape has evolved, the concept of “oracle-free” protocols has started to pick up steam. These protocols have little or no reliance on external data sources, which have come to be seen by many as a growing centralized point of failure in the ecosystem. As Dan Elitzer from Nascent puts it in his recent article,
“As soon as any external dependency has been introduced into a contract, the contract inherits any risks associated with that dependency. Governance over contract functionality implies a form of upgradeability, requiring threat models to factor in both the range of possible changes and current and future conditions under which they could be made. Oracles by definition introduce external data, and with it, all dependencies that go into delivering that data and all potential variability in its accuracy and timeliness.”
He goes on to assert that any protocol that calls itself a primitive should abide by the definition of a primitive:
“[…] In order to qualify as a primitive, a contract must have zero external dependencies, other than those of the blockchain it is deployed upon.
This means no governance, no upgradeability, no oracles.”
Recently, a number of protocols have attempted to implement this paradigm, such as Ajna. However, such endeavors have faced challenges, particularly around liquidity. The limited liquidity in these platforms limits their scale.
Another approach explored is the use of isolated margin. Platforms such as Silo Finance have championed this concept. Still, much like Ajna, they too grapple with a lack of liquidity, especially when it comes to long-tail assets. This issue underscores the complexity and challenges in striving for a fully decentralized and reliable DeFi lending system.
Ultimately, being truly “oracle-free” is probably not feasible for the wide range of scalable Defi applications. However, it’s important to be trust-minimized, decentralized, censorship resistance, and have oracle data sources be impervious to market conditions to avoid manipulation. That’s what we’re building with zkOracles at Unstable.
zkOracle: A Leap Forward with Succinct Labs and unshETH
In our pursuit of a decentralized and trustless DeFi landscape, we’re excited to announce collaboration with Succinct Labs and unshETH. Together, we’re introducing the zkOracle for LSD fair value. The zkOracle falls under the fourth category of oracles — a decentralized, fair price oracle for LSDs.
By using Succinct’s state of the art ZK technology, we are able to generate a zk-SNARK proof of validator balances on the Beacon Chain. Succinct’s accelerated hash function primitives, enabled by their open-source Curta library, and the Succinct SDK’s easy support for a “map-reduce” style computational model which allows us to shared proving by batches of validators, are key to enable this type of validity proof of large-scale computation. With this zk-SNARK oracle, our goal is to supplant the need for LSD-specific market price oracles. This oracle, trustless by design, will pave the way for a more accessible integration experience of LSDs as collateral on lending protocols, significantly reducing both the time and cost barriers for onboarding.
A working proof-of-concept for the zkOracle has been achieved internally, and is being prepared for day one integration within the Unstable Protocol. Protocol developers can look forward to a testnet version rolling out in the coming weeks. Our vision for the zkOracle extends beyond our own platform — it will be an open-source, public good for the ecosystem at large. By doing so, we encourage other protocols to adopt and integrate this oracle, further enhancing the LSD onboarding experience across the entire spectrum of lending protocols.
The Unstable Paradigm
What does the holy grail of lending look like? A protocol that seamlessly integrates cross-margined collateral; boasts an uncomplicated interest rate model; ensures protection against systemic risks like “bank runs” or “looping,” and supports long-tail assets without an excessive dependency on centralized or market-price oracles (i.e. “oracle-less”). This culmination of features is no small feat and remains a challenging problem to tackle, especially while ensuring economic robustness.
At Unstable, we’re solving this for the largest and fastest growing category of DeFi collateral and the most popular use case today — borrowing dollars against staked ETH. We are able to achieve this with our comprehensive LSDfi-native risk framework and by using the decentralized, trustless zkOracles built in partnership with Succinct Labs and unshETH.
Our model at Unstable aims to go beyond just LSDs. We’re venturing into assets that, while not strictly liquid-staked derivatives (LSDs), mimic their behavior and trading patterns. At launch, we will be supporting LSDs, but our model is extensible to support:
- Index LSDs (e.g. unshETH, Yearn yETH)
- Liquid restaking tokens (i.e. liquid wrappers of LSDs are re-staked on Eigenlayer)
- Structured LSD products (e.g. Origin oETH, Sommelier ryETH)
- LSD-ETH LP positions
Stay tuned for Part 3, the final article in the series introducing Unstable Protocol, in which we will lift the curtain behind the core protocol.
In the meantime, follow us on Twitter!
