Moderation Pays: Optimal Leverage in Lending Protocols
Max leverage on recursive lending in protocols such as Aave, Compound, and their forks isn’t always optimal for max net profit
Lending protocols are one of the cornerstones of DeFi and are often one of the most utilized protocols on any chain. The reason for their success is that they have multiple use cases, such as lending, borrowing, hedging and leveraging, and are often a core component of investing strategies implemented by crypto users. Due to their numerous use cases, They are often among the protocols with the highest total value locked (TVL) on any chain. This can be seen in the chart below showing some of the top lending protocols current TVL.
In this article I will summarize some of the most popular use cases that lending protocols fulfill and follow that with a deeper dive into one specific use case, recursive lending strategy (or often called folding), and how analysis and modeling can help optimize this strategy to improve capital efficiency.
Lending Protocol Use Cases
Among protocols that can be broadly defined as lending protocols, there are two main types that we will define.
- Lending Markets: This is the type of project that comes to mind when someone is talking about a lending protocol. On these platforms you can supply an asset such as ETH or BTC as a lender and receive interest or you can borrow these assets and pay interest. In these markets, the borrowed position is overcollateralized, which means that a borrower will need to supply collateral that is of a larger dollar amount than the amount they wish to borrow.
- Collateralized Debt Position (CDP) Markets: Often set in their own category, CDP protocols work in a similar way and represent a specific use case where a user supplies one asset to mint another. The minted asset is most often a stablecoin with the most well-known CDP protocol being MakerDAO.
These two types of markets offer a wide variety of use cases that can be implemented into investing and asset management strategies. Below are some popular examples.
- Borrowing/Lending: The simplest of use cases, borrowing and lending can provide yields to lenders and also allow users to extract liquidity from assets to make them more productive. An example of this is a user lending an asset they want to continue to hold and use it as collateral to borrow a different asset that they want to deploy for yields in a different protocol.
- Asset Minting: Similar to borrowing and lending, but instead a user deposits their asset to mint another asset (normally a stablecoin like DAI). This is what CDP protocols do and is often a cost effective way to extract liquidity from assets that you don’t want to trade, but want to increase their productivity.
- Hedging: A user who has exposure to an options contract or an LP position on an AMM might want to hedge against potential losses. In an example where the user believes the price of the asset deployed in the option or LP position will fall, they would lend a stable asset such as a stablecoin, ETH or, BTC and borrow the asset that is deployed in the other strategy. If the user believes the asset price will rise, they can do the inverse.
- Leveraging: Borrowing and lending can be used as a method to long or short an asset as well. An example would be if you believe an asset will rise in price against the dollar, you can lend this asset on a lending market, then borrow a dollar stablecoin and trade it for more of the asset you borrowed.
- Farming: Many lending protocols provide incentives to use their platforms. This is often in the form of their own token being given as a reward to lend or borrow assets. A common strategy that is deployed by users is to implement a recursive lending strategy where the user supplies an asset on the lending side, borrows the same asset, and then repeats this process multiple times.
From the use cases listed above, one of the simplest to implement for a crypto user is to use the farming use case in a recursive lending strategy. Especially when a protocol is incentivizing use of their platform, recursive lending can be a good way to capture incentive rewards with a lower risk of having your assets being liquidated compared to some of the other use cases mentioned above. When deploying this strategy, one might believe that the best option is to max your recursive lending by lending and borrowing repeatedly until you have no assets left to borrow. However, the best option might actually involve a more moderate recursive lending strategy.
Recursive Lending: Respect the Gap
To maximize the returns per dollar deployed in recursive lending, there are multiple variables that need to be considered for each pool in a protocol:
- Utilization ratio: It indicates how much of the total supplied assets are currently being borrowed. This ratio is used in determining the current supply and borrow rates. It is important to consider because most lending protocols have a “kink” in their rate models where the interest rates start to increase exponentially.
- Total assets supplied/borrowed: Knowing the current total assets supplied and borrowed in a pool can help determine the depth of the pool. It is important for determining how a new deposit or borrow amount will affect the supply and borrow rates, but also incentives rates.
- Supply, borrow, and incentive rates: knowing what the current supply and supply-side incentive rates as well as the borrow and borrow-side incentive rates will provide a benchmark for a new depositor and also will help determine the best leverage to use.
These variables will help determine three key metrics when deciding what type of leverage to use.
- Net Gap Rate: This is simply the gap between the net supply rate (supply rate + supply incentives rate) and the net borrow rate (borrow rate + borrow incentives rate) of the pool. This rate helps to determine the profitability of the recursive lending strategy on the pool.
- Net Borrow Rate: Combining the borrow rate and the borrow incentives gives us the net borrow rate which is the realized cost of borrowing assets on the lending protocol. It is important to determine if the net borrow rate is negative or positive because it will impact the leverage (how many times to perform the recursive strategy) we want to use. If the rate is negative, it means that each time we borrow more assets, we will be diminishing the size of the net rate gap. Depending how large the negative borrow rate is, high leverage could make the net rate gap negative, thus making the strategy outcome a net loss.
- Proximity to the kink: A lending protocol’s kink for each pool can be found on their website dashboard or their documentation. We can determine how close we are to the kink by looking at the difference between it and the current utilization rate. This helps to determine the optimal amount of assets to allocate to the pool. When the utilization rate passes the kink, supply and borrow rates both begin to increase substantially which will change the optimal leverage used in the recursive lending strategy. Below is an example from Aave’s ETH pool on Ethereum. It shows the current utilization rate (50.18%) and the optimal rate (kink) at 70%. In the Aave example, if the utilization rate passes the kink at 70%, interest rates will start to increase.
To best understand how these variables play a role in creating a recursive lending strategy, we will use the example shown in the table below. The example will be implementing recursive lending on the BTCb pool in Benqi, by lending and borrowing BTCb.
Benqi BTCb Pool: Lower leverage can be optimal
The BTCb pool in the Benqi protocol on Avalanche is currently a perfect example of a pool where it is best to have lower leverage than the maximum possible in order to get the highest return per dollar deposited. The first items to look at for this pool are in the table above where we see that the current net borrow rate is negative (-4.14%) and the net rate gap is 1.01%. This indicates that the more leverage that we use (meaning more borrowed) at a certain point we will reduce the net rate gap to zero and subsequently will start incurring a loss on our deposit subject to the deposit size.
The decision to be made is what will be the best leverage ratio to use to earn the highest return? As an example, we will look at a hypothetical depositor that has $3M to deploy. The chart below plots out an array of potential deposit amounts (between $1K and $3M) and leverage possibilities. Each dot in the scatter plot represents a deposit at a given leverage. To make the chart easier to interpret, the leverage used has been categorized by color with leverage 1 indicating no leverage used and leverage x-y indicating the leverages in that range (ex. In leverage 1–2 you will find leverage 1.1, 1.2, etc.). Furthermore, specific paired deposit examples have been highlighted where the color represents the size of the deposit and the shape represents which deposit is using higher leverage.
The best way to interpret this plot is to look at it from right to left. In the bottom right corner, we see APRs and net profits for smaller deposits. As we move to the upper left, the deposit size becomes gradually larger. The general shape of the lines created by the consecutive dots in the scatter plot is due to increasing deposit size. As deposit size increases, the incentive rates are diluted across the total size of the pool (incentive rates are linearly decreasing compared to supply and borrow rates that are increasing), thus decreasing the realized APR per dollar deposited. When the incentive rates reach a certain level of dilution, the net rate gap becomes zero and the net profits start to decrease.
If we focus back on the bottom right, we see the two blue markers that represent a deposit of $10K using either 3.2x or 2.0x leverage. With smaller deposits, we see that we receive a higher realized APR (APR after removing costs of borrowing) at the higher leverage compared to the lower, resulting in annualized net profits of $738 and $614, respectively. However, looking at a $3M deposit into the pool (black markers), it becomes more profitable to reduce leverage with a 1.1x leverage generating $116,425 in annualized net profits and 2.0x leverage only generating $90,547.
Summary
Recursive lending is an effective strategy to generate yields on deployed assets whilst also reducing liquidation risks. However, depending on the size of the deposit to be made into the pool in question, it is worthwhile to do additional analysis to determine the optimal leverage that will provide the returns per dollar deposited. Additionally, positions with reduced leverage can provide secondary benefits by further reducing risk of liquidation for the depositor and increasing borrowable assets in the pool which can be beneficial to the health of the lending protocol.
