In-depth analysis of 5 valuation methods for using NFTs as collateral: What are the pros and cons? What is the outlook?

Author: taetaehoho.eth

Original Title: 《Valuing NFTs as Collateral - Overview, Landscape, Pros/Cons》

Compiled by: Linqi, Chain Catcher

My colleague previously detailed some emerging protocols in the NFT finance space and outlined methods to generate liquidity for NFTs using some of these protocols.

Now, I will focus on one area of NFT financialization, which I believe is currently the most obvious use case—creating liquidity for your NFTs through NFT-backed loans—and explain how protocols in this space evaluate NFT collateral, along with the pros and cons of each method.

Motivation

I have a valuable NFT. I don’t want to sell it, but I want to leverage its value to gain liquidity. What I can do is have someone write me a loan secured by the value of the NFT. Once I receive the ETH, I can go play with DeFi while still retaining my Mooncat.

An important step in this process is assessing the value of the collateral. Lenders will only provide liquidity if they can adequately secure their capital and receive fair compensation for the risks they take—such as your default.

Thus, lenders can either:

1. Determine that the borrower is trustworthy, which is difficult to do in a permissionless and anonymous environment.
2. Or ensure that the collateral's value is sufficient to protect the lent capital in case of user default.

Therefore, establishing robust valuation methods is crucial for platforms connecting buyers and sellers of NFT liquidity.

NFT Loan Ecosystem

Currently, protocols operating in this space use five main vectors to determine valuation.

1. P2P

The P2P model pioneered by NFTfi places the responsibility of valuation on the borrower and lender. The platform acts as a marketplace where participants can meet, negotiate terms, and execute loans. A quick overview:

1. Alice needs liquidity for her Mooncat and uses it as collateral for a loan.
2. Bob wants to earn yield on ETH and submits an off-chain loan offer detailing the principal, term, and APR.
3. If Alice receives an offer she accepts, she will transfer the loan amount on-chain. Meanwhile, her NFT is locked, and ETH is transferred to her EOA (Externally Owned Account).

Among all valuation mechanisms, the P2P market provides participants with the greatest freedom. Borrowers and lenders can dynamically negotiate multiple variables to best match their individual risk preferences and environmental factors. For example, a borrower wanting to execute a 15-day DeFi strategy may prefer a loan with a 15-day term to minimize refinancing risk. In protocols that simplify the user experience, traders cannot execute such strategies due to insufficient freedom (fixed APR, term length).

Protocols created based on this model vary significantly in the options they provide to market participants.

For instance, on Sharkyfi, loan terms are fixed, and APR is determined based on utilization curves. Lenders can only decide the loan size. On the borrower side, they can automatically see the maximum loan size at the top of the loan order book, with uniform APR and term.

Arcade requires borrowers to specify loan terms, which are then filled in by lenders.

Advantages of P2P

• Highly customizable. This allows for negotiations on special transactions (i.e., the borrower and lender understand each other, leading to more favorable terms, and both have preferences for terms and are willing to settle on non-market terms… etc.).

Disadvantages of P2P

• Determining the best parameters can be difficult and resource-intensive.
• Borrowers cannot access liquidity immediately.
• No dynamic valuation adjustments; liquidation is based on LTV.

2. Governance/Evaluation

On JPEG'd, users can mint stablecoins backed by NFT deposits (similar to how Maker mints DAI). At launch, the protocol valued Alien Punks at 4000 ETH and Ape Punks at 2000 ETH. According to the protocol's Medium: ["Project governance can change these values later."](#:~:text=however, governance can change these values at a later date)

The Taker protocol similarly determines the value of NFTs through governance, but not by the protocol's administrators; rather, it is determined by consensus among expert evaluators.

1. CuratorDAO consists of "notable individuals and projects in each NFT category."
2. CuratorDAO provides an evaluation value that all borrowers can use to obtain loans (via LTV buffers).
3. CuratorDAO guarantees loans with its own funds and bears the risks of lending and default, thus providing accurate valuations through self-incentivization. (This design has significant overlap with P2P/rational participants).

Advantages of Governance/Evaluation

• Borrowers receive immediate liquidity.
• Valuation is determined by consensus and verified through a lengthy voting process, making it less susceptible to price manipulation attacks.

Disadvantages of Governance/Evaluation

• Governance processes may slow down adjustment progress.
• Difficult to dynamically adjust valuations.
• Attackers may conduct governance attacks by purchasing votes on-chain.

3. Oracles and P2Pool Using Oracles

Oracles enable real-time dynamic pricing of NFTs based on external price feedback. Protocols that use oracle data for pricing can differ significantly in the following aspects:

1. Data sources.
2. How they aggregate source data.

Among the protocols we studied, two sources are the most widely used.

1. NFTX Floor Price.
2. Opensea API--- This information is brought on-chain via Chainlink oracles.

This data is typically aggregated in the form of TWAP. The data from different sources is then combined into a weighted average final price.

For example, Drops DAO uses three data sources: Drops NFT Floor TWAP, NFTX Floor Price TWAP, and Chainlink NFT Oracles, and aggregates the data into a weighted average floor price for the collection. Borrowers then take loans at the LTV % based on this dynamically adjusted valuation.

Pine Protocol uses the min (7-day average trading price, collected floor price) obtained from the Opensea API. Generally, TWAP is the most common data aggregation method we see in NFT loan protocols.

Advantages of Oracles

• Dynamic valuation.
• Borrowers receive immediate liquidity.

Disadvantages of Oracles

• Can be manipulated— the less liquid the market, the easier it is to manipulate. Malicious actors can continuously list NFTs at low prices and buy them themselves, leading to liquidation of that specific collection. This only works when the discussed NFT liquidity is extremely poor and arbitrage bots are scarce. Therefore, ensuring listing requirements (through governance or automatically) is crucial.

4. Rational Agents

Incentivizing profit-maximizing agents to ensure correct valuation of NFTs. The most common practice is to establish a "shared risk" mechanism among valuation providers, where agents incur losses in case of user default or valuation errors, and profit in the opposite case. We will explore two different approaches, but there are many different options.

Abacus Spot

Abacus uses an "optimistic proof of stake" valuation method. (White Paper)

1. Alice is a profit-maximizing trader. She sees an open pool for a rare punk and decides to lock ETH in the pool (we will explain why later). The longer she locks ETH, the more protocol tokens (ABC) she receives.
2. Because she is quick, she receives the first "ticket"—the first 0-1 ETH in the pool is hers.
3. Bob and his friends also lock ETH in the pool.
4. NFT holder Charlie sees that there are 20 ETH locked in the pool by Alice and Bob, but Charlie believes his NFT is worth less than this value.
5. Charlie immediately "closes" the pool (only Charlie can do this because he owns the NFT), transferring all ETH in the pool to Charlie and auctioning the NFT for 48 hours.
6. If the NFT sells for more than 20 ETH, the profits will be shared with Alice, Bob, and their friends. Those who locked in later will receive a proportionally higher profit. This is because…
7. If the NFT sells for less than 20 ETH, the profits will be distributed on a first-come, first-served (FIFO) basis. Alice, who locked in at 0-1 ETH, will receive 1 ETH back, but Bob, who locked in at 19-20 ETH, will receive nothing.
8. Thus, in step 6, Bob receives a greater return for taking on more risk.

There are some complexities that occur before and after the expiration date, but overall, the trader's motivation is to lock enough ETH such that the potential profit from the sale + token release = opportunity cost of the capital (locked ETH).

Once Abacus determines the valuation, other protocols can provide loans based on this valuation. Gradient is one such example.

Pilgrim

Thus, the protocol uses profit-maximizing rational actors to determine the valuation of NFTs, in this case, the total liquidity in the pool.

Advantages of Rational Agents

• Establishes forward-looking valuations. This is also true for P2P and governance-led valuations, but oracle valuation methods are the opposite of forward-looking.

Disadvantages of Rational Agents

• Protocols must attract traders to the platform.
• Currently, these protocols establish valuations for individual NFTs, but this is difficult to scale.

5. Machine Learning

Valuation protocols use ML to predict valuations by using past transactions and features as inputs. NFTBank and Banksea Finance are prime examples.

To better understand NFTBank's algorithm, check out How to value items in NFT projects? --- Part 1. As of November 2021, their model achieved single-digit average absolute percentage error accuracy for Axie.

Performance of the NFTBank model as of November 2021.

NFTBank has announced a partnership with Chainlink to bring their predicted prices on-chain, allowing protocols needing real-time NFT valuations to use their data feeds.

Banksea Finance stated in its initial funding proposal that it aims to incorporate "NFT creator information, attributes, historical transactions, media coverage, community status, popularity, and other information to assess NFT value trends and NFT sentiment trends" to determine prices.

Advantages of Machine Learning

• Real-time dynamic data, updating with every transaction occurring in the collection.

Disadvantages of Machine Learning

• NFTs within the same trait group may have a wide error margin due to outlier sales.
• May not predict systemic market trends (i.e., changes in game NFT metas, changes in the market conditions of general NFTs). Experienced traders/evaluators may catch this.
• Difficult to respond to "game changers."
• Currently non-forward-looking.

Valuation Methods Pros and Cons Summary

The financial season belonging to NFTs is approaching. Finding liquidity for NFTs will become as seamless and widely used as DeFi CDPs.