Spectral Finance – Credit Risk Oracle for Pseudonymous Undercollateralized Lending in DeFi

Spectral Finance is Rarestone’s best bet to solving undercollateralized loans in DeFi without doxing a user’s identity while incorporating practical recourse measures.

Read on to learn more about why Spectral excites us about the future of credit in DeFi.


Highlights

Spectral Finance benefits from the first-mover advantage. They are the only team we are aware of that is aiming to tackle on-chain credit assessments and undercollateralized lending primarily on-chain and without doxing their users.

Pre-seed investors include Rarestone Capital, Galaxy Digital, Maven 11, New Form and the DeFi Alliance. These investors will also support Spectral with bootstrapping its network effects when launched through contributing liquidity to the pool. Rarestone Labs (a subsidiary of Rarestone Capital) will support Spectral Finance with the token launch and marketing efforts.


Background

DeFi has grown into a vibrant ecosystem offering users the ability to access a plethora of different financial instruments without the need for any trusted intermediary.

Growth has been predominantly catalyzed by borrowing and lending platforms, yet due to the trustless nature of DeFi, capital efficiency is problematic. Currently, participants are faced with overcollateralized debt obligations when borrowing due to the risk being transferred to the user by the protocol. Such a feature is not conducive to true decentralization as it hinders those without ample assets from fully participating while concentrating utility to those with large amounts of capital.

As such, unsecured lending is arguably one of the biggest market opportunities in DeFi. However, it remains the least penetrated. Those that have been trying to tackle this problem aim to do so through combining elements of traditional finance (TradFi) with DeFi. In short, they aim to enable a user to access an under-collateralised loan by assessing their financial history from their bank. By doing so, the users are essentially doxing themselves in the process.

We believe that the approach to unsecured lending described above not only fails to scale by being unable to serve the unbanked population, it also fails to align with the spirit of DeFi. It’s now widely accepted that the DeFi ethos promotes equal rights, openness, self-custody and — most importantly — privacy.

One of DeFi’s defining characteristics is composability. However, composability is indicative of modular and compounding future value creation. There is a growing opportunity for capturing this value retrospectively, by looking at past user behaviour within the ecosystem.

Participants of DeFi are building wallet-based financial histories and thereby establishing creditworthiness (or lack thereof). Yet still, there is no adequate solution to rewarding good behaviour while providing recourse for bad behaviour.

Therefore, the logical solution for undercollateralized lending in DeFi (while preserving pseudonymity) is through on-chain credit risk analysis. In such a system, users would essentially be scored based on previous interactions with DeFi products and money markets, how well they’ve managed their finances and debt obligations over time.


What is Spectral?

Spectral Finance is developing a — first-of-its-kind — credit risk oracle and loan underwriter that will serve as a middleware solution for the entire DeFi ecosystem, namely the credit markets. The credit ratings generated by Spectral will determine the amount of collateral required by a user.

Spectral’s credit ratings are determined by three factors:

  1. On-chain financial history in DeFi
  2. Credit delegation by other users
  3. Off-chain data points (optional)

Empowered by Spectral, borrowers will now be able to purchase under-collateralized loans from leading DeFi protocols such as Aave or Compound. By doing so, Spectral leverages the robustness and supply-side network effects of the existing DeFi credit markets to optimize capital efficiency.

The Spectral protocol will be responsible for posting the collateral needed to service the over-collateralization ratio required by most DeFi lending protocols.

Spectral will manage its own liquidity pool. Liquidity providers are motivated to provide the financial capital needed to post collateral, in return for an interest rate that corresponds with a borrowers credit risk.


Why?

Existing attempts at solving undercollateralized lending aim to do so by integrating the TradFi world with DeFi and, therefore, rely on legacy recourse measures. This approach falls short since they:

  1. Cannot scale exponentially — The protocol must be able to process credit data from a variety of different countries, take nomenclature differences into account, and “trust” certain actors at the edges of the network.
  2. Fails to cater to underserved portions of the market — Most credit systems rely almost entirely on historical debt repayment information and therefore cannot easily accommodate users who are new to credit.
  3. Fails to preserve privacy — Porting financial history from a user’s bank account to access undercollateralized loans would entail doxing oneself.
  4. Expose single points of failure or attack — Reliance on TradFi will expose the undercollateralized lending application to centralization risks facing the legacy financial system e.g, identity fraud. Borrowers must expose all of their personal information when applying for a loan – the same info an attacker can use to open new lines of credit.
  5. Expose users to censorship threats — Doxing a user’s identity and financial history may expose risks of unwarranted censorship e.g., geo-restrictions.

Spectral’s Target Market

Demand Side
  • DeFi Users (B2B and/or B2C) — pre-existing individuals or businesses using DeFi products/services e.g., Aave, Compound, Synthetix, etc.
  • Unbanked Population — individuals that do not have access to banking services nor the credit history that permits undercollateralized loan issuance.

Supply Side
  • Liquidity Providers (LPs) — risk-taking investors seeking higher yields on their assets compared to LPing to the likes of Compound, Aave and others.

Spectral’s Value Propositions

  • Permissionless undercollateralized loans — users of Spectral do not need to have a track record in legacy credit systems.
  • Capital efficiency — Since less financial capital is needed to attain a loan, there are improvements to market liquidity overall. It also opens up opportunities for an increase in leveraged yield farming due to lower collateralization requirements.
  • Scalability through leveraging existing network effects — Spectral will rely on blue-chip DeFi lending protocols to provide the last-mile lending services to the borrowers on their platform.
  • Highly composable — Spectral’s credit underwriting system can integrate with an array of DeFi lending protocols, unlocking access to the best borrowing and lending rates in the market. Additionally, other protocols or applications in the web3 ecosystem can choose to utilise Spectral’s credit scoring system for more unconventional use-cases.
  • No KYC — Credit scoring in Spectral’s ecosystem does not depend on doxing a user’s identity or examination of their financial history in the legacy world.


How does it work?

1. Data Collection & Analysis

a.) On-chain history

Most users manage several wallet addresses when dealing with DeFi products and services. To ensure that Spectral is able to cater to the majority, users will be able to bundle all of their wallets together. When a user bundles their wallets, it will be represented as an NFT. This will add a layer of obfuscation so their wallet addresses and transactional history remains private and still can be added to a gamified leaderboard.

Spectral will use a machine learning model to parse multiple on-chain data points from blue-chip DeFi protocols users have previously engaged with. The model will assess several factors such as payment history, debt owed, length of credit history, new credit, and credit mix. Each of these variables will be broken down into many granular sub-variables to ensure the accuracy of the credit scoring.

The sub-variables will ultimately be more nuanced, taking into account factors such as wallet balances, income/debt ratios, liquidation levels, risk management, and more.

The variable groups (and their respective scores) will be publicly visible for users to see. This is done to encourage users to take the appropriate actions to try and improve their credit scoring, ultimately encouraging responsible risk-taking.

Meanwhile, the sub-variables and their respective weightings will remain closed-source in the near- to medium-term, to ensure the machine learning algorithm is not gamed by bad actors.

To guarantee robustness and accuracy of the chosen parameters and the corresponding weightings, the Spectral team will be continually back-testing. At present, DeFi “whale” wallets are primarily being used as training data as they provide rich data-sets.

b.) Credit delegation / Credit Unions

In a bid to cater to users with no historical activity in DeFi, Spectral also leverages information through other novel means that avoid explicit doxing while enriching its analysis of stability and risk across the credit lifecycle.

For users that do not have a rich credit history, Spectral aims to take into consideration the financial history of a user’s peer network as an indicator of repayment behaviour. Peers can be friends, family or affiliates in the DeFi space with known or unknown identities. This will also allow Spectral to create credit unions or groups where a DAO, corporation or any entity can leverage collective creditworthiness from its members.

Individually, members are beholden to each other to preserve their creditworthiness so the collective groups buying power and beneficial borrowing terms are not damaged. For instance, a company could create a credit union so its employees can get better terms for borrowing than if they went individually to a money market. Their actions will reflect the greater group’s creditworthiness.

In practice, users with rich DeFi credit history will be encouraged to stake their $SPEC tokens and nominate their peers they consider to be worthy of undercollateralized lending. Encouragement is accomplished through both a staking reward as well as improvements to one’s own Spectral credit score. The staking reward will be inflation-based. In the event of a credit default, credit delegators (stakers) will be at risk of slashing and their personal credit scores being negatively impacted. These measures taken by the Spectral Protocol is done to encourage good behaviour and social recourse in the event of a liquidation.

c.) Off-chain data points

Spectral Finance will give users the option to source alternative data using existing identifiers from the legacy world. The team is exploring alternative sources of data that could be used to demonstrate creditworthiness, however, are not currently accounted for in legacy credit analysis. For instance, Spectral could take into account a user’s Spotify monthly subscription commitments and payment history to demonstrate creditworthiness.

2. Credit Scoring

Once the synthesis of all the data points worthy of consideration is completed, a credit score — termed Multi-Asset Credit Risk Oracle (MACRO) score — is attributed to each individual user upon request.

The rating will use a similar variable weighting as the FICO grading model and the MACRO score is normalized to the 300 – 850 range which users will be able to interpret easily. 

FICO grading model
FICO grading model

In Spectral’s case, the credit score given to each borrower will determine the amount of collateral they will be required to post when taking out a loan. The higher the score, the lower the collateral requirements.

During the early phases, the best possible outcome for users with a stellar credit score will be approximately a 120% collateralization ratio (subject to change), whereby the remaining 30% needed to take out a loan on the likes of Compound will be financed by Spectral liquidity providers.

As the credit rating machine learning algorithm improves over time, the Spectral protocol will be more accepting of risk and thereby lower the collateralization ratio for borrowers.

3. Undercollateralized Loan Issuance

To cater to reduced collateral loan issuance, Spectral rely on their community of liquidity providers (LPs) to service the collateral requirements on behalf of the borrowers in return for an attractive yield.

The process involves LPs initially depositing their financial capital into a Spectral pool. To avoid complexity and user experience constraints, LPs will deposit their financial capital into a single pool with the expectation of a blended yield in return.

In the back-end, the Spectral protocol will funnel the pooled capital into several tranches that distinguish themselves from each other by the level of risk (risk corresponding to the credit scores) and expected interest rate returns.

When a borrower with a pre-defined credit score requests a loan, liquidity from the corresponding liquidity tranche/pool is used as collateral in the chosen lending protocol (e.g., Compound).

Spectral will collect interest payments from borrowers periodically. The amount of interest paid by the borrower will be contingent on their credit score.

Since LPs invest in a single liquidity pool, the interest payments they earn will be a blend of all the interest payments borrowers pay with respect to their credit score.

DeFi Composability

Credit scores attributed to each individual user will be represented with the same NFT generated when wallet addresses are bundled. The creation of the NFT will allow for creditworthiness to now become a fungible asset.

Eventually, new protocols (or existing ones) will accept Spectral NFTs instead of capital or collateral, representative of creditworthiness without any doxing.


$SPEC Token – Value Accrual Thesis

Based on our analysis, we believe that the $SPEC token will capture value due to the following features:
  • Staking for Credit Delegation Purposes— inflation rewards combined with the incentive to improve one’s own credit score delivers what we believe to be a strong incentive to take supply off the market and stake it in the Spectral network.
  • Governance — The Spectral network has the opportunity to capture fees from the interest payments paid by borrowers to LPs. As the Spectral network grows and progressively decentralizes, we expect token holders to vote for a portion of the fees to compensate them in some capacity.

Token Metrics

100M total supply of $SPEC tokens, of which:

  • 12% — Pre-seed investors
  • 13% — Seed investors (to be allocated)
  • 24% — Team
  • 40% — Network emissions
  • 6% — Future advisors and team members
  • 5% — Insurance / emergency pool


Founding Team

Seasoned cryptocurrency professionals and deep tech computer science engineers. The team stem from reputable universities and have working experience at the likes of Loopring and Google. Two out of the three co-founders are completing a PhD in Computer Science at NYU, one with a focus on cryptography/blockchain and the other on machine learning.

The founding team members asked to remain anonymous for the purpose of this research report. For an introduction to the team contact us directly.


Stage of Maturity

Spectral plan to launch its MVP in Q2 2021. 

Current Stage
  • Tech architecture and system design still in refinement phases.
  • MACRO Credit Score – parsing on-chain behaviour to output a credit risk score, based on 50 variables and split into groups of 5. The model is currently in development. Credit scoring will be primarily based on a user’s Aave and/or Compound historical activity.

Future Developments
  • Refining MACRO Credit Score – At the start, the collateralization ratio will be around 120%, with iterative refinements to the algorithm to scale down to reach undercollateralized loans. 
  • Credit scoring improvements – Spectral will expand the scope of DeFi protocols it assesses to reach a credit score, namely Synthetix in the near-to-mid term.


Disclosures

Rarestone Capital has recently taken a position in $SPEC based on the analysis provided in this report. This is meant to disclose any perceived conflict of interest and should not be mistaken as a recommendation to purchase $SPEC tokens. This overview has been prepared solely for informational purposes and is not to be considered as investment advice. It does not purport to contain all of the information that may be required or desirable to evaluate all of the factors that might be relevant to a potential investor, and any recipient hereof should conduct its own due diligence investigation and analysis to make an independent determination of the suitability and consequences of any action.