Pendle PT tokens from
@pendle_fi are increasingly being used as collateral in lending markets to create high-yielding positions.
Yet there are substantial challenges for creating and managing these kinds of markets using existing solutions.
Here’s how Euler vaults can make life simpler, safer, and more efficient.
--
Pendle PT tokens represent the principal of a yield-bearing asset. They typically trade at a discount to their underlying, giving rise to a fixed interest rate for the period over which they mature.
For example, a PT token for sUSDe which matures on 24th October currently yields ~12.5% APY.
Using this as collateral to borrow, for example, USDC, enables borrowers to earn a multiplier on their earned interest.
Borrowers can deposit the sUSDe PT token as collateral, borrow USDC at a cost of e.g. 5%, and then loop.
At a loan-to-value of 0.75, this enables a multiplier of around 4x, meaning borrowers can earn an upper limit of 4x * (12.5 - 5) ~= 30%.
Meanwhile, lenders of USDC benefit from receiving yield on a relatively low risk market. They earn 5% whilst lending to the holders of a collateral asset which is highly correlated in price (and therefore relatively unlikely to trigger liquidations). As the PT token matures, the value of the collateral increases over time as well.
Overall, one can see why it is a popular trade.
Today these kinds of positions are typically being created on lending protocols using ungoverned isolated pairs. There’s several reasons why that is a sub-optimal approach in terms of simplicity, risk management, and efficiency.
--
First, creating and managing lending pairs for this kind of trade is deeply inefficient and will require a lot of housekeeping. If there are several high-yielding PT tokens USDC lenders want to lend to, a separate USDC pool is required for each PT token.
This means USDC lenders need to fragment and manage their liquidity inside multiple pairs at once. This is especially challenging when considering that new PT-USDC pairs need to be recreated every time a PT token is close to maturity, with lenders having to migrate frequently between pools.
In a best case scenario, all borrowers will repay their loans close to maturity, but in practice this never happens. For some lenders, this means their USDC could be stuck inside an illiquid pool for a prolonged period of time after maturity. Earning yield, yes, but still unable to withdraw.
Even if lenders use a risk curator to manage this process on their behalf, the fragmented liquidity across different USDC vaults creates uncertainty for borrowers, with more volatile rates making profit calculations harder.
Euler vaults can be used to overcome the inefficiencies of pairs very easily. A single vault can be created that is designed to accept any number of collateral assets. This means a vault creator can add multiple PT tokens as collateral at once, or gradually add and remove PT tokens as they are born or reach maturity.
This makes life much easier for both lenders and borrowers, who can more easily roll their positions as the world moves on.
--
Second, using ungoverned markets for PT tokens raises challenges for risk management. In the rare event that something goes wrong, USDC lenders might appreciate having additional checks and balances on vaults to help manage risk.
Euler vaults can be deployed in an ungoverned manner if people wish, but they also allow risk curators to govern and help manage risk. A governor would have the option to be able to slowly ramp down LTVs, switch interest rates, modify supply and borrow caps to curb further borrowing activity. Alternatively, vault creators can add custom hooks which can be used to allow permissioned deposits or liquidations, withdrawal delays, and more.
These governance and more advanced risk management features are completely optional and likely won’t to everyone’s taste, but for many users will add value.
--
Third, using isolated pairs with fixed LTVs and liquidation bonus parameters can pose risks to both lenders and borrowers.
In the event of liquidation, borrowers should have to pay no more than is needed for someone to close out their position. Yet on the vast majority of lending protocols, the liquidation bonus paid to liquidators is a fixed percentage of the borrower’s collateral – typically more than 5%!
This means borrowers who have looped 4x stand to lose up to 20% of their collateral, just to pay for the privilege of being liquidated by an MEV bot.
This is obviously bad news for borrowers, but it poses risks to lenders too. If 20% of a borrower’s collateral is going towards liquidation costs, that’s 20% that is no longer helping to secure loans. Fixed liquidation bonuses can, in some circumstances, therefore elevate the risks of bad debt to lenders as well as punishing borrowers.
In contrast, Euler uses a Dutch-auction based approach to price the liquidation bonus. This leads to much fairer liquidation bonuses that tend to only be a little higher than the fixed cost to liquidate someone. I wrote about how this is the fairest liquidation mechanism in DeFi recently here:
x.com/euler_mab/status/18414….
--
Finally, most people building and managing looped positions on PT tokens today are doing so manually. This costs gas and takes a significant amount of time. Unwinding positions in a hurry is a nightmare.
Euler’s powerful batch transaction system with liquidity deferral checks enables looping PT token’s as collateral in a single transaction. Moreover, borrowers have the option to use intents to enable operators to manage positions on their behalf, giving them the option to automatically close out positions which are no longer profitable, execute more advanced stop-loss and take profit orders, and much more.
--
Overall, Euler is a very flexible lending platform where builders can deploy vaults that can make life simpler, safer, and more efficient for traders.
Pendle PT integrations with Euler may unlock a whole new world of possibilities.
Euler still has the fairest and most efficient liquidation mechanism for borrowers in all of DeFi in my humble opinion:
- the lowest liquidation penalties
- no liquidation fees
- the option to use intents to customise partial liquidations, stop-losses, and take profit options
Here's how it all works with some discussion of the trade-offs associated with different designs.
Feel free to disagree!
--
When a user is at risk of becoming insolvent, you want liquidators to come in and close out their position. The liquidators aren't going to do this out of the goodness of their heart, they want to get paid. That payment comes in the form of a bonus, taken from the user's collateral.
But how big should the bonus be? In a world of fluctuating gas prices, asset prices, and liquidity, how do we even know how much it will cost to perform a liquidation so that we can compensate people fairly?
We generally don't. So, instead, what most lending protocols do is take a fixed % of a user's collateral and give that to the liquidators. The problem with this is two fold.
First, there is no fixed percetage that works fairly in all cases. Is 10% fair? Well 10% on collateral of $200 often isn't going to cut it, meaning positions won't get liquidated, but 10% on collateral of $2m is a payment of $200k. Imagine paying $200k to your friendly neighbourhood liquidator for the privelege of being liquidated! There are many such cases.
Second, most large bonuses don't even go to the liquidators, they end just generating MEV. Everyone can see a liquidation is coming, and profitable ones will generally get sniped by bots who bid up the gas price to get priority inclusion.
So what we do on Euler to avoid this is use reverse Dutch auctions. These are based on health scores, rather than based on time. A user with health of 0.99 pays a 1% bonus. A user with health of 0.95 pays a bonus of 5%, and so on. As a user's health score falls, the potential profit to liquidators rises. Since anyone can liquidate at any time, there's a Dutch auction on the liquidation bonus, which in a copetitive environment like Ethereum, usually settles close to the marginal level of profitability for liquidation.
This means that if it costs $20 to liquidate, then the liquidation bonus on collateral of $200 will be found at close to 10%, whereas on collateral of $2m, will be found at close to just 0.001%.
--
On Euler there are also no liquidation fees. Many lending protocols implement these as a source of extra revenue. These fees can be incredibly punishing, and create a conflict of interest between the protocol's governors and their borrowers. The protocol wants you to get liquidated so that it can get paid. Sophisticated borrowers should be factoring these fees into their expected cost of borrowing. Something like this:
extraBorrowingCost = P(liquidation) * liquidationFee / initialCollateral
A fee of 1% sounds innocent enough, but for people looping, the fee gets taken on the looped amount. So 1% can quickly become 5% or more. If you're a large borrower why risk paying this?
--
Finally, Euler makes partial liquidations optional by giving borrowers the option to use intents to customise partial liquidations, stop-losses, take profits and more.
Rather than impose an opinionated model for partial liquidations onto borrowers, like we did in Euler v1, in v2 borrowers can recruit operators to perform actions on their accounts under tightly specified conditions.
Example: "if the chainlink oracle for BTC reports a price of $55k or less, close out 50% of my position for me and you will get a reward of 50 USDC"
Operators can be smart contracts or literal people in an office clicking buttons. It doesn't really matter. The allowable actions and incentive to perform those actions on an account is pre-specified by the borrower. In the scenario in which they choose nothing, the only actions that can be taken are through liquidations.
You can even implement a kind of LLAMA-like liquidation using this mechanism. You want an operator that scales out of your position as the price falls, and scales back into the position if it begins to rise again.
--
Some lending protocols group positions into bands so that they can be co-liquidated together, which can lead to gas savings. Something like this could also be built on top of Euler by tokenising user positions and allowing transfers between them. Does this really add value though?
An important consideration here is that this type of banding will only increase the efficiency of liquidations if users naturally choose similar collateralisation ratios for their positions so that they are actually banded together often.
On some pairs of assets this might be common if the lending protocol is popular and the trade has a natural schelling point for collateralisation ratio, but for volatile trades where risk/reward tolerances vary, borrowers will tend to vary in their ratios, rendering banding and co-liquidation less beneficial or potentially more costly than a regular liquidation.
It really depends on the popularity of the trade. With banding, borrowers may be incentivised to look for popular bands they can share with other users in order to optimise and reduce expected costs.
--
All in all, I think Euler has an incredibly flexible and efficient system that is hard to beat in terms of value for borrowers.