Primary Lending Platform: Collateral asset parameter modeling
2022 - Alex Pasfield, Brian Pasfield.
Introduction
This document describes the modeling that is applied to arrive at the lending parameters for each collateral asset on Fringe Finance’s Primary Lending Platform.
The key parameters for each collateral asset are as follows:
- Liquidator Reward Percentage (LRP)
- Maximum Borrowing Capacity (MBC)
- Loan-to-Value Ratio (LVR)
The aim of deriving suitable values for these collateral asset parameters is to balance the competing aims of maximizing both the platform’s stability and its long-term user adoption.
Each collateral asset is assessed to ensure its unique characteristics are considered in deriving these parameter values.
Determining Loan-to-Value Ratio (LVR) for a collateral asset
Determining a suitable Loan to Value Ratio (LVR) for a collateral asset plays an important role in maintaining platform stability.
Each collateral asset has an LVR that defines the ratio of loan value that a user can take out against a given value of collateral. Loans are overcollateralized, which results in the LVR always being less than 100%. For example, a collateral asset with an LVR of 60% would allow users to take out a loan of up to 60% of their collateral value.
The aim of the LVR is to best assure the loan remains overcollateralized even when the collateral asset experiences adverse market price action. Accordingly, determining the LVR for a given collateral asset involves analyzing the asset’s historical price volatility and then assuming that volatility will tend to extend into the future. The more volatile a collateral asset’s price is historically, the lower the LVR will be set, and vice-versa.
The method to determine LVR is to analyze the price history of the collateral asset and gauge its maximum price falls over the period of time during which liquidations are reasonably likely to take place. We have used a period of four hours to assess an asset’s volatility.
Why four hours?
We choose four hours due to liquidators having to compete with one another to be first to liquidate a loan. This results in them having an imperative to perform liquidations without undue delay.
Liquidators may not always be in a position to immediately perform a liquidation when a loan position falls below its minimum collateralization level. This could be either due to their capital already being deployed elsewhere or more attractive liquidation opportunities being available. The four hour time window takes both these factors into account. Additionally, we performed the same volatility analysis for different time periods and the results were not sufficiently different to warrant using a time basis other than four hours.
In order to carry out this analysis, we analyze the maximum price movements within every four hour period since trading began for each collateral asset. The LVR is then chosen as 100% minus the worst historical four hour drawdown, which is then multiplied by a conservative factor.
The below diagram approximates our approach by displaying discrete four hour periods, whereas our actual methodology uses continuous four hour periods. This simplification was made with the goal of making our diagramatic depiction more intuitive.
As can be seen in the example above, the blue lines point to the open and close price points for each four hour time window. Periods where the price rises are ignored because we are only interested in situations where price falls. The highlighted segment shows where the price fell the most, by 60%. This is then used as the basis for the LVR, given the price history tells us that the asset has experienced price falls of up to 60% during any four hour period.
When analyzing historical price data, we sometimes ignore an asset’s initial weeks of trading due to the extremely volatile nature of such periods of low liquidity. Such early stage volatility is generally not reflective of long-term later patterns.
Implementing the conservative factor
When loans secured by a particular collateral asset are liquidated, the collateral is subsequently sold by liquidators onto the market, resulting in the price of that collateral asset often decreasing due to the increased supply (i.e. the liquidator will experience slippage).
To take into account the impact of these loan liquidations on the price of the collateral asset, we multiply the worst expected volatility over a four hour period by a slippage threshold. The slippage threshold is calculated based on the maximum slippage we expect to result from all loans secured by a collateral asset being liquidated simultaneously.
This is, additionally, one of the primary rationales for why we have implemented a Max Borrow Capacity for each collateral asset. It helps ensure that slippage is bounded and that the available liquidity in markets for each collateral asset is capable of absorbing this volume without causing a precipitous decline in the collateral asset’s price.
As mentioned, a conserative factor is then applied to the maximum price drawdown including the slippage threshold to arrive at the final LVR used by the Primary Lending Platform. This conservative factor considers a 5% slippage, therefore reducing the LVR to 0.95 times its original calculation. In the example presented in the diagram, this results in the following calculation:
Using a Maximum Borrowing Capacity for a collateral asset
Fringe Finance sets the maximum aggregate borrowing amount for each collateral asset. This is aggregated across all borrowers.
This is for multiple reasons:
One is to protect the platform from price manipulation attacks. Assume an attacker manipulates the prices of an asset on external marketplaces by any means. Price manipulation attacks allow attackers to inflate the price of an asset, often by several orders of magnitude, which they then use as collateral to take out a large loan position. When the price manipulation event is over, the price returns to its previous level, resulting in very little collateral backing a large loan, making the loan insolvent. This type of attack is an existential threat to lending platforms. With a maximum borrowing amount in place, Fringe can make this kind of attack unprofitable and therefore eliminate every incentive to conduct it.
Another is to ensure liquidators will be willing to perform liquidations. With this kind of limit in place, there is sufficient readily-available liquidity in the markets on which the asset is traded. If there is too little liquidity, liquidators will be reluctant to perform liquidations, as they’ll experience excessive slippage when disposing collateral assets. This would compromise efficient liquidation operations and risk destabilizing the Fringe Finance platform.
Therefore, calculating a suitable Maximum Borrow Capacity for a collateral asset is crucial for the smooth operation and stability of the Fringe Finance platform.
Calculating Maximum Borrowing Capacity
To determine an asset’s Maximum Borrow Capacity, we analyze the aggregate liquidity of the asset in the markets in which it is traded. This aims to determine how much liquidity is readily available where there is relatively little slippage. Thanks to this, we can define a Maximum Borrow Capacity which not only ensures liquidators are willing to perform liquidations, but also deters attackers by making price manipulation attacks unprofitable.
The following example diagrammed below illustrates how the Maximum Borrow Capacity protects the Fringe Finance platform. It presents a hypothetical bid order book which has a large proportion of its liquidity near the last traded market price, as is typical of order books. As liquidity dries up progressively at prices further away from the recently traded price, any subsequent trades would experience significant slippage, seriously affecting the price at which an asset could be disposed.
The reason we use only 50% is because we do not wish to assume all orders in the order book will persist in any market movement. Choosing 50% is effectively a ‘conservative factor’ for reasons of prudence.
The slippage threshold we have used for all collateral assets so far is 5% due to it being a point of diminishing returns. Allowing for higher slippage brings little increased liquidity, and therefore is impractical.
The example in the diagram shows that up to $50M of loans on the Fringe Finance platform could be liquidated and disposed of on the available markets within a reasonably acceptable effective slippage. Of course, during usual operations of the platform, only a very small portion of all loans will need to be liquidated at any point in time, resulting in liquidators experiencing much lower slippage than our worst case scenario figure of 5% slippage.
To maximize the protection of the Fringe Finance platform, we configured the parameters for worse case scenarios to take into account rapid, adverse market conditions that could cause cascading liquidations to occur simultaneously.
Determining the Liquidator Reward Percentage (LRP)
Hand-in-hand in with determining the Max Borrowing Capacity, we can also derive the Liquidator Reward Percentage.
The Liquidator Reward Percentage is the percentage of additional collateral assets awarded to the liquidator when they liquidate a loan position. When a liquidator pays off a loan, they receive collateral asset to the value they paid PLUS the Liquidator Reward Percentage.
There’s a simple rationale behind paying liquidators such a reward: Only paying them what price oracles state as the value of the capital which they provide to liquidate a loan is insufficient to fully cover their costs. Examples of such additional costs include slippage, operational overheads, and other opportunity costs which liquidators must incur.
By far the largest and most variable of these additional costs is slippage. We set the Liquidator Reward Percentage to be sufficient to cover the slippage costs, which would result from all loans secured by a given collateral asset being liquidated simultaneously. As a result, we calculate the Liquidator Reward Percentage for liquidators to always find it profitable to liquidate loans to maintain platform solvency, even in the worst market conditions.
Conveniently, as mentioned earlier, we limit the maximum slippage, which all loans secured by a given collateral asset liquidated simultaneously could cause, to 5% which is achieved by limiting the value of all such loans to be below the Max Borrowing Capacity.
As a result, we can arrive at the Liquidator Reward Percentage by simply adding a conservative factor to the slippage threshold of 5%. Currently, we have used a conservative factor of 1.3x which results in a Liquidator Reward of 6.5%. Given that a high Liquidator Reward Percentage is costly for borrowers if they are liquidated, as it is paid from their collateral, we must ensure it is as low as possible without compromising platform stability.
As the Max Borrowing Capacity for a given collateral asset is increased, the slippage which liquidating all such loans would result in, also increases. Consequently, a higher Liquidator Reward Percentage is required to compensate liquidators for added slippage costs.
Jointly, the Max Borrowing Capacity and Liquidator Reward Percentage have the simultaneous goals of minimizing liquidation costs for borrowers while keeping Max Borrow Capacity as high as possible.
Summary
Fringe Finance forges new frontiers in enabling holders of smaller, more volatile tokens to deploy their capital by borrowing stablecoins against it. To achieve this, we have established a set of mechanisms within the platform to maintain its stability. These mechanisms rely on a suitable set of parameters for each collateral asset. As described in this paper, deriving suitable parameters employs a systematic and logical approach with a rationale that reflects the characteristics of each collateral asset.
Fringe Finance will monitor the changing characteristics of each collateral asset on an ongoing basis and will continue to assess the performance of the platform. As required, we will adjust the parameters accordingly to achieve the best balance of platform stability, capital efficiency for our users, and to maximize long-term adoption through incentive-compatible mechanics.
Fringe Finance. DeFi for Everyone.
