Infinity's Risk Framework
At the heart of Infinity is a powerful risk management system and methodology more commonly used within an Institutional Finance context. DeFi protocols, and more specifically, the risk management systems that Lending protocols use, are limited not only in their computational abilities and inability to run simulations, but also their ability to actively hedge (i.e. liquidate) positions as appropriate given their reliance on 3rd parties (e.g. Oracles, Liquidators/Keepers).
We've employed a hybrid risk management system where we run our risk off-chain, but execute our hedges on-chain (although this may change in the future).

Advances in Risk Management

Oracles providing a single price are grossly insufficient to rely on for financial risk management purposes. It's one thing to make the status of a baseball game available on-chain, it's another to make the price of a fast-moving financial asset instantly available, alongside it the liquidity or depth of the market, and an understanding of any second-order effects that may result from liquidation.
Risk isn't a Price - a Price is a static, one-dimensional metric. Risk is a dynamic set of multi-dimensional arrays and it's with this change in approach that we built our risk management system. Risk management is (broadly) a combination of Price Determination, Risk Evaluation, and Liquidation. When the markets are moving fast, having three different parties to perform these functions exposes the protocol to market or gap risk - you need all three of these components in a single, fully-integrated system (with external redundancy/failovers where appropriate).

Dynamic Replication for Pricing and Liquidation

Infinity's prices are sourced from a handful of CeFi entities, DeFi protocols, and Oracles. We find CeFi typically moves faster than DeFi, but despite this, more information across not just prices, but depth, and volatility enable better visibility. Infinity takes the following types of assets as collateral:
  • Basic tokens: e.g. USDC, USDT, DAI, ETH, and WBTC
  • Deposit tokens: e.g. aETH (Aave), cETH (Compound)
  • Complex tokens: e.g. Curve LP Tokens, Uniswap NFTs
When tokens are actively traded, for example Basic Tokens, hedging is relatively straight-forward once we've taken market depth into account. With Deposit and Complex Tokens however, there is effectively a one-sided market and our ability to both price and liquidate such tokens is at the discretion, or design, of the protocol itself.

Managing Complex Risks

For such tokens, we've had to reframe the 'price' determination process to one known as Dynamic Replication. The underlying principle is to 'look through' the complex product and evaluate the risks that drive the price/value of the asset itself. In the case of an aETH token, the price is driven by that of ETH, and to a lesser extent, the credit risk of Aave and prevailing interest rates. Why do we do this?
Let's take for example, a situation in which we have $80mm of Aave's aETH tokens, they only have $10mm ETH available (i.e. it's a $100mm market, and $90mm has been lent out), and we need to liquidate our full position. It's simply not possible on short notice.
As most of the risk is driven by ETH, what we can do is short $80mm ETH so we are directionally hedged, and then wait for more liquidity in Aave's protocol until such time that we can then ultimately reduce our basis risk. This assumes that Aave itself isn't the reason for the change in price.
Dynamic replication isn't limited to Deposit-like tokens - take for example Uniswap's V3 LP positions which by definition are unique (i.e. they are ERC-721s, and as such are NFTs), have no market and as such, have no liquidity. Uniswap's LP positions are driven by the underlying markets, with their price being 100% as one asset, 100% as the other asset, or somewhere in between (ignoring fees). Rather than taking the boundary-value conditions (i.e. 100% one asset or 100% the other asset), our system can 'read' the NFT, its tick values, currency pair, and associated pricing parameters and almost completely reverse engineer the unwind price. While Uniswap doesn't have the same liquidity risks that Aave has, knowing how to price, and thus value, such super-unique NFTs enables us to provide financing to LP providers.
​
​
​
Last modified 23d ago
Copy link
Outline
Advances in Risk Management
Dynamic Replication for Pricing and Liquidation
Managing Complex Risks