In the push to decentralize the financial rails of the world, DeFi has largely been throttled by one thing: Credit. Lenders need to grant credit to borrowers on the basis of trust, but in a decentralized economy, trusting the wrong individual can mean catastrophic losses. TrueFi, RociFi, and ArcX are three protocols that have been working to overcome this DeFi roadblock by making on-chain credit a sustainable reality.

Credit in DeFi is primarily used for leverage or to gain exposure to another asset without the need to sell your current holdings. The issue is that most DeFi credit comes from overcollateralized loans. Given the trustless nature of DeFi, overcollateralized loans are considered a necessary evil, ensuring that lenders do not lose all their funds due to an anon borrower taking the loan with no intention of repayment.

While this protects lenders, overcollateralized lending has poor capital efficiency and fails to provide the same growth opportunities as traditional credit. Take, for example, borrowing DAI on AAVE: There is a 77% maximum loan-to-value ratio (LTV), meaning for every $100 deposited, users can borrow $77. Low capital efficiency inhibits the expansion of DeFi. Traditional finance offers undercollateralized and unsecured loans—such as mortgages, student loans, and credit lines—facilitating entrepreneurship and growth opportunities. Without undercollateralized credit, DeFi faces headwinds on the path to becoming the future of finance.

Alongside the need for undercollateralized credit is an urgent need for credit scoring. Credit scores simplify and automate the process of underwriting loans. TradFi companies that generate consumer credit scores, such as FICO, have access to an abundance of personal and financial information unavailable to blockchain-based protocols. In fact, it is an extremely lucrative business, with FICO valued at over $12 billion while making almost all its profit from its credit score business (Wall Street Journal, 2022). With this market share up for grabs by burgeoning DeFi protocols, we’re now seeing the development of on-chain credit scoring models from RociFi and ArcX. However, considering anonymity and privacy are cornerstones of DeFi, protocols have had to become more creative in deriving on-chain credit scores compared to their TradFi counterparts.

TrueFi offers an ideal product market fit to address the need for undercollateralized credit on chain. As of August 2022, TrueFi has seen over $1.7 billion in total loan originations, making them the most prolific on-chain undercollateralized lending market. By lending only to institutions with at least $10 million in unencumbered assets—many of whom are respected crypto-native firms—TrueFi has been able to avoid any loan defaults (TrueFi, 2022).

TrueFi offers two types of lending pools: DAO-managed pools and Capital Market pools. Both are limited to institutional borrowers. However, DAO-managed pools are open for retail users to lend their stablecoins into either BUSD, USDC, TUSD, or USDT pools. From these pools, approved borrowers can take out loans based on the terms set by TrueFi. The approval process begins with institutional borrowers completing Know Your Business (KYB) forms and master loan agreements, and passing a credit review conducted by TrueFi. The TrueFi credit team determines a credit score based on the borrower’s background, repayment history, trading history, assets under management, and leverage exposure. Based on the credit assessment, the TrueFi team establishes the terms of the loan, which go to the $stkTRU holders for a vote. If the token holders vote to onboard the borrower, the institution can then take out the pre-approved loan. The DAO-managed pools enable retail users to earn over 10% on their stablecoin holdings; far more than the 1-2% offered by AAVE and Compound in return for the higher risk. These pools grant decision-making power to the community, given that $stkTRU holders can veto a loan proposal even if the borrower has passed the TrueFi credit check (though none have been vetoed yet). This is a conscious step towards decentralized, community-based participation.

Capital Market pools have a single borrower per pool, unlike DAO-managed pools, which see a variety of borrowers from a single pool. Borrowers go through a similar credit check but have the ability to choose who can lend to them through allowlists and preapprovals. Since the borrower decides who can lend to them, there is no need for a vote from $stkTRU holders. Still, these pools benefit from the on-chain visibility of funds and the improved efficiency from removing intermediaries in typical TradFi loan originations.

Though the risk is significantly higher than overcollateralized lending, the expected returns greatly make up for the risk. Given the historical success of TrueFi and its “SAFU” fund, which is able to slash up to 10% of the $TRU tokens as a backstop to repay bad debt, TrueFi is arguably the safest iteration of on-chain undercollateralized lending.

Higher on the risk curve is RociFi. Like TrueFi, RociFi offers undercollateralized loans to institutions and DAOs, but they are also the first protocol to successfully offer undercollateralized lending to retail users.

In order to scale undercollateralized lending to retail users, RociFi relies on its on-chain algorithmically determined credit scoring system. Before users can borrow from RociFi, they must mint an NFCS (Non-Fungible Credit Score). Similar to a traditional credit score, the NFCS belongs to the user, but the user does not control it. RociFi relies on a 1 to 10 credit scale, with 1 being the most creditworthy. The credit score is determined by on-chain history and wallet activity while cross-referencing RociFi’s native fraud detection database. Users can improve their score by linking other wallet addresses, interacting with other DeFi protocols on chain, and having healthy repayment behavior within RociFi.

The NFCS acts as an access gateway to the three pools within RociFi. There is one undercollateralized pool with a max LTV of 80% for any user with a score 9 or above. Users with scores between 4 to 9 can access the “undercollateralized pool 2,” which has a max LTV of 120%. This pool is primarily designed for “anonymous retail borrowers with good on-chain credit” (RociFi, 2022). The final pool, known as “undercollateralized pool 1,” is restricted to users with scores 1 to 3. This pool is typically used for reputable DAOs, institutions, and high-reputation borrowers, given its max LTV of 140%.

Since the credit scoring and loan origination process is entirely automated, RociFi relies on its own algorithmic interest rate rather than a manually derived term sheet. Based on the TradFi capital asset pricing model (CAPM) equation, RociFi’s interest rate = risk-free rate + risk premium + volatility charge. The risk-free rate comes from an average of AAVE’s stablecoin deposit yield while the risk premium is “based on an estimate of the market’s desired rate of return at a given risk level” (RociFi, 2022).  Based on utilization rates, the risk premium is variable. Finally, the volatility charge is an added charge to benefit borrowers against the possibility of a rapid drop in the collateral’s value, which could potentially lead to unpaid loans. However, the RociFi team plans to potentially use this volatility charge to hedge the collateral with puts in the future rather than giving it to lenders as an added yield. As of writing, the expected lending APYs are 4.05%, 7.42%, and 7.29% respectively for the overcollateralized, undercollateralized 1, and undercollateralized 2 pools.

Given the risky nature of undercollateralized lending, RociFi relies on its liquidation mechanism and social recourse policy to help the protocol avoid bad debt. Each loan is a 30-day fixed-rate loan; if the loan is not repaid on the maturity date, it enters a 5-day grace period for repayment with a doubled interest rate. Overcollateralized loans are automatically liquidated, since these liquidations are profitable and can be executed by profit-seeking bots. However, liquidating undercollateralized loans reaps no profit. The protocol has set aside a fund that covers unpaid undercollateralized loans to minimize the damage. In the event of non-repayment, the user’s credit score drops to a 10, and RociFi disseminates the defaulter’s known information across its social media and community channels as a form of social recourse. The goal is to make a user’s future on-chain reputation more valuable than the potential profit from non-repayment. RociFi presents an innovative case study for the potential success of retail undercollateralized lending on chain coupled with an in-house algorithmic credit scoring system.

Aside from undercollateralized lending, ArcX aims to create the most capital-efficient overcollateralized borrowing platform, improving upon existing solutions like AAVE and Compound.

Based on the assumption that not all borrowers should be treated equally, ArcX uses an algorithmically derived on-chain credit score like RociFi. Instead of a 1 to 10 score, ArcX calculates a credit score between 0 to 999 based on the combination of a Survival Score and a Daily Reward Score. The Survival Score is based on a borrower’s ability to avoid liquidations relative to the market on third-party platforms that ArcX indexes, such as AAVE and Compound (for now). Users who avoid liquidations during periods of high volatility and react appropriately to avoid liquidations can expect a high Survival Score. The Survival Score is based on a 0 to 300 scale, creating a minimum bound for the score.

In order to reach a score of 999, users need to borrow directly on ArcX. The Daily Reward Score is based on responsible borrowing over the prior 120 days. The score rises at the fastest rate when users borrow at the optimal rate, which is currently set to 60%. This not only rewards safe borrowing but also forces users to borrow on ArcX and generate protocol fees before reaching a score of 999.

ArcX offers three vaults for each underlying asset. Vaults A, B, and C are all denominated with a minimum score requirement to borrow and a maximum loan-to-value range. Vault A is available to every borrower, with a max LTV range of 80% to 90%. Users with a credit score of 0 have a max LTV of 80%, while those with a score of 999 can borrow the maximum LTV of 90% before liquidation. Vaults B and C offer higher max LTV ratios with a peak of 95% and 100% respectively. However, vault B requires more than 500 points, while vault C requires more than 750. As the vaults scale up with LTV ratios, the credit limits scale down such that the credit available for borrowing in vault C will always be less than vault A.

Through this design, ArcX rewards safe borrowers who can prove creditworthiness with higher capital efficiency, improving borrowers’ cost of capital and granting them a competitive advantage. This also gives lenders greater control and clarity, allowing them to supply liquidity preferentially based on borrower’s credit, rather than supplying to a single pool.

Since ArcX does not serve undercollateralized loans, the risk management techniques are much simpler. In the event of liquidation, a borrower’s credit score is deducted by 250 points. For most loans, the LTV ratios will be lower than 100%, making liquidations profitable for the protocol. The protocol will suffer bad debt from the interest owed to lenders only when a user who begins with a score of 300 borrows from vault C at 100% LTV and fails to repay the loan. However, in this worst-case scenario, the protocol will only incur 3.82% of the borrowed amount. This means the protocol cannot ensure profitability with the current design in a worst-case scenario; but based on the aggregate profitability of typical loans and users wanting to increase their scores, the protocol will have enough revenue to cover any worst-case scenarios.

Though TrueFi, RociFi, and ArcX entail higher risk than their predecessors AAVE and Compound, they seek to fill a critical gap in the current DeFi ecosystem. If DeFi wants to compete with TradFi, DeFi must solve the risk/reward challenge of undercollateralized lending. As we see further user adoption and increased loan volumes, these protocols will garner more data and continuously improve their risk management. Though in its nascent stage, the innovations being made in undercollateralized lending and on-chain credit scores lay the foundation for the expansion of consumer credit in DeFi. While institutional lending on-chain with legal recourse is more efficient and transparent than TradFi, there is still work to be done to make retail-focused undercollateralized lending as safe as its traditional counterparts.

AnnaMaria Andriotis. August 2021. FICO Score’s Hold on the Credit Market Is Slipping.

https://www.wsj.com/articles/fico-scores-hold-on-the-credit-market-is-slipping-11627119003

RociFi. June 2022. RociFi Protocol LitePaper.

RociFi Protocol Litepaper

RociFi Protocol Litepaper "Perhaps the largest financial value built directly on reputation is credit and uncollateralized lending. Currently, the web3 ecosystem cannot replicate simple forms of ...

https://blog.roci.fi

RociFi. July 2022. RociFi Docs.

Introduction | Lending Platform

RociFi is a decentralized credit economy with non-transferable, blockchain-native credit scores designed to facilitate zero and under-collateralized lending. At its core, the protocol leverages on-chain data, machine learning and loan risk-management to effectively and profitably facilitate under-collateralized loans via the blockchain.

https://docs.roci.fi

TrueFi. September 2022.

TrueFi. November 2022. Introducing TrueFi, the DeFi Protocol for Undercollateralized Lending.

https://blog.trusttoken.com/introducing-truefi-the-defi-protocol-for-uncollateralized-lending-9bfd6594a48

ArcX. August 2022. ArcX Wiki

ARCx Credit Introduction | ARCx

ARCx Credit is a decentralized credit market on the Polygon network that offers dynamic maximum-LTV loans on ETH collateral based on a borrower's DeFi Credit Score. The DeFi Credit Score provides a credit risk assessment for individual wallet addresses based on historical on-chain borrowing activity.

https://wiki.arcx.money

In the push to decentralize the financial rails of the world, DeFi has largely been throttled by one thing: Credit. Lenders need to grant credit to borrowers on the basis of trust, but in a decentralized economy, trusting the wrong individual can mean catastrophic losses. TrueFi, RociFi, and ArcX are three protocols that have been working to overcome this DeFi roadblock by making on-chain credit a sustainable reality.

Credit in DeFi is primarily used for leverage or to gain exposure to another asset without the need to sell your current holdings. The issue is that most DeFi credit comes from overcollateralized loans. Given the trustless nature of DeFi, overcollateralized loans are considered a necessary evil, ensuring that lenders do not lose all their funds due to an anon borrower taking the loan with no intention of repayment.

While this protects lenders, overcollateralized lending has poor capital efficiency and fails to provide the same growth opportunities as traditional credit. Take, for example, borrowing DAI on AAVE: There is a 77% maximum loan-to-value ratio (LTV), meaning for every $100 deposited, users can borrow $77. Low capital efficiency inhibits the expansion of DeFi. Traditional finance offers undercollateralized and unsecured loans—such as mortgages, student loans, and credit lines—facilitating entrepreneurship and growth opportunities. Without undercollateralized credit, DeFi faces headwinds on the path to becoming the future of finance.

Alongside the need for undercollateralized credit is an urgent need for credit scoring. Credit scores simplify and automate the process of underwriting loans. TradFi companies that generate consumer credit scores, such as FICO, have access to an abundance of personal and financial information unavailable to blockchain-based protocols. In fact, it is an extremely lucrative business, with FICO valued at over $12 billion while making almost all its profit from its credit score business (Wall Street Journal, 2022). With this market share up for grabs by burgeoning DeFi protocols, we’re now seeing the development of on-chain credit scoring models from RociFi and ArcX. However, considering anonymity and privacy are cornerstones of DeFi, protocols have had to become more creative in deriving on-chain credit scores compared to their TradFi counterparts.

TrueFi offers an ideal product market fit to address the need for undercollateralized credit on chain. As of August 2022, TrueFi has seen over $1.7 billion in total loan originations, making them the most prolific on-chain undercollateralized lending market. By lending only to institutions with at least $10 million in unencumbered assets—many of whom are respected crypto-native firms—TrueFi has been able to avoid any loan defaults (TrueFi, 2022).

TrueFi offers two types of lending pools: DAO-managed pools and Capital Market pools. Both are limited to institutional borrowers. However, DAO-managed pools are open for retail users to lend their stablecoins into either BUSD, USDC, TUSD, or USDT pools. From these pools, approved borrowers can take out loans based on the terms set by TrueFi. The approval process begins with institutional borrowers completing Know Your Business (KYB) forms and master loan agreements, and passing a credit review conducted by TrueFi. The TrueFi credit team determines a credit score based on the borrower’s background, repayment history, trading history, assets under management, and leverage exposure. Based on the credit assessment, the TrueFi team establishes the terms of the loan, which go to the $stkTRU holders for a vote. If the token holders vote to onboard the borrower, the institution can then take out the pre-approved loan. The DAO-managed pools enable retail users to earn over 10% on their stablecoin holdings; far more than the 1-2% offered by AAVE and Compound in return for the higher risk. These pools grant decision-making power to the community, given that $stkTRU holders can veto a loan proposal even if the borrower has passed the TrueFi credit check (though none have been vetoed yet). This is a conscious step towards decentralized, community-based participation.

Capital Market pools have a single borrower per pool, unlike DAO-managed pools, which see a variety of borrowers from a single pool. Borrowers go through a similar credit check but have the ability to choose who can lend to them through allowlists and preapprovals. Since the borrower decides who can lend to them, there is no need for a vote from $stkTRU holders. Still, these pools benefit from the on-chain visibility of funds and the improved efficiency from removing intermediaries in typical TradFi loan originations.

Though the risk is significantly higher than overcollateralized lending, the expected returns greatly make up for the risk. Given the historical success of TrueFi and its “SAFU” fund, which is able to slash up to 10% of the $TRU tokens as a backstop to repay bad debt, TrueFi is arguably the safest iteration of on-chain undercollateralized lending.

Higher on the risk curve is RociFi. Like TrueFi, RociFi offers undercollateralized loans to institutions and DAOs, but they are also the first protocol to successfully offer undercollateralized lending to retail users.

In order to scale undercollateralized lending to retail users, RociFi relies on its on-chain algorithmically determined credit scoring system. Before users can borrow from RociFi, they must mint an NFCS (Non-Fungible Credit Score). Similar to a traditional credit score, the NFCS belongs to the user, but the user does not control it. RociFi relies on a 1 to 10 credit scale, with 1 being the most creditworthy. The credit score is determined by on-chain history and wallet activity while cross-referencing RociFi’s native fraud detection database. Users can improve their score by linking other wallet addresses, interacting with other DeFi protocols on chain, and having healthy repayment behavior within RociFi.

The NFCS acts as an access gateway to the three pools within RociFi. There is one undercollateralized pool with a max LTV of 80% for any user with a score 9 or above. Users with scores between 4 to 9 can access the “undercollateralized pool 2,” which has a max LTV of 120%. This pool is primarily designed for “anonymous retail borrowers with good on-chain credit” (RociFi, 2022). The final pool, known as “undercollateralized pool 1,” is restricted to users with scores 1 to 3. This pool is typically used for reputable DAOs, institutions, and high-reputation borrowers, given its max LTV of 140%.

Since the credit scoring and loan origination process is entirely automated, RociFi relies on its own algorithmic interest rate rather than a manually derived term sheet. Based on the TradFi capital asset pricing model (CAPM) equation, RociFi’s interest rate = risk-free rate + risk premium + volatility charge. The risk-free rate comes from an average of AAVE’s stablecoin deposit yield while the risk premium is “based on an estimate of the market’s desired rate of return at a given risk level” (RociFi, 2022).  Based on utilization rates, the risk premium is variable. Finally, the volatility charge is an added charge to benefit borrowers against the possibility of a rapid drop in the collateral’s value, which could potentially lead to unpaid loans. However, the RociFi team plans to potentially use this volatility charge to hedge the collateral with puts in the future rather than giving it to lenders as an added yield. As of writing, the expected lending APYs are 4.05%, 7.42%, and 7.29% respectively for the overcollateralized, undercollateralized 1, and undercollateralized 2 pools.

Given the risky nature of undercollateralized lending, RociFi relies on its liquidation mechanism and social recourse policy to help the protocol avoid bad debt. Each loan is a 30-day fixed-rate loan; if the loan is not repaid on the maturity date, it enters a 5-day grace period for repayment with a doubled interest rate. Overcollateralized loans are automatically liquidated, since these liquidations are profitable and can be executed by profit-seeking bots. However, liquidating undercollateralized loans reaps no profit. The protocol has set aside a fund that covers unpaid undercollateralized loans to minimize the damage. In the event of non-repayment, the user’s credit score drops to a 10, and RociFi disseminates the defaulter’s known information across its social media and community channels as a form of social recourse. The goal is to make a user’s future on-chain reputation more valuable than the potential profit from non-repayment. RociFi presents an innovative case study for the potential success of retail undercollateralized lending on chain coupled with an in-house algorithmic credit scoring system.

Aside from undercollateralized lending, ArcX aims to create the most capital-efficient overcollateralized borrowing platform, improving upon existing solutions like AAVE and Compound.

Based on the assumption that not all borrowers should be treated equally, ArcX uses an algorithmically derived on-chain credit score like RociFi. Instead of a 1 to 10 score, ArcX calculates a credit score between 0 to 999 based on the combination of a Survival Score and a Daily Reward Score. The Survival Score is based on a borrower’s ability to avoid liquidations relative to the market on third-party platforms that ArcX indexes, such as AAVE and Compound (for now). Users who avoid liquidations during periods of high volatility and react appropriately to avoid liquidations can expect a high Survival Score. The Survival Score is based on a 0 to 300 scale, creating a minimum bound for the score.

In order to reach a score of 999, users need to borrow directly on ArcX. The Daily Reward Score is based on responsible borrowing over the prior 120 days. The score rises at the fastest rate when users borrow at the optimal rate, which is currently set to 60%. This not only rewards safe borrowing but also forces users to borrow on ArcX and generate protocol fees before reaching a score of 999.

ArcX offers three vaults for each underlying asset. Vaults A, B, and C are all denominated with a minimum score requirement to borrow and a maximum loan-to-value range. Vault A is available to every borrower, with a max LTV range of 80% to 90%. Users with a credit score of 0 have a max LTV of 80%, while those with a score of 999 can borrow the maximum LTV of 90% before liquidation. Vaults B and C offer higher max LTV ratios with a peak of 95% and 100% respectively. However, vault B requires more than 500 points, while vault C requires more than 750. As the vaults scale up with LTV ratios, the credit limits scale down such that the credit available for borrowing in vault C will always be less than vault A.

Through this design, ArcX rewards safe borrowers who can prove creditworthiness with higher capital efficiency, improving borrowers’ cost of capital and granting them a competitive advantage. This also gives lenders greater control and clarity, allowing them to supply liquidity preferentially based on borrower’s credit, rather than supplying to a single pool.

Since ArcX does not serve undercollateralized loans, the risk management techniques are much simpler. In the event of liquidation, a borrower’s credit score is deducted by 250 points. For most loans, the LTV ratios will be lower than 100%, making liquidations profitable for the protocol. The protocol will suffer bad debt from the interest owed to lenders only when a user who begins with a score of 300 borrows from vault C at 100% LTV and fails to repay the loan. However, in this worst-case scenario, the protocol will only incur 3.82% of the borrowed amount. This means the protocol cannot ensure profitability with the current design in a worst-case scenario; but based on the aggregate profitability of typical loans and users wanting to increase their scores, the protocol will have enough revenue to cover any worst-case scenarios.

Though TrueFi, RociFi, and ArcX entail higher risk than their predecessors AAVE and Compound, they seek to fill a critical gap in the current DeFi ecosystem. If DeFi wants to compete with TradFi, DeFi must solve the risk/reward challenge of undercollateralized lending. As we see further user adoption and increased loan volumes, these protocols will garner more data and continuously improve their risk management. Though in its nascent stage, the innovations being made in undercollateralized lending and on-chain credit scores lay the foundation for the expansion of consumer credit in DeFi. While institutional lending on-chain with legal recourse is more efficient and transparent than TradFi, there is still work to be done to make retail-focused undercollateralized lending as safe as its traditional counterparts.

AnnaMaria Andriotis. August 2021. FICO Score’s Hold on the Credit Market Is Slipping.

https://www.wsj.com/articles/fico-scores-hold-on-the-credit-market-is-slipping-11627119003

RociFi. June 2022. RociFi Protocol LitePaper.

RociFi Protocol Litepaper

RociFi Protocol Litepaper "Perhaps the largest financial value built directly on reputation is credit and uncollateralized lending. Currently, the web3 ecosystem cannot replicate simple forms of ...

https://blog.roci.fi

RociFi. July 2022. RociFi Docs.

Introduction | Lending Platform

RociFi is a decentralized credit economy with non-transferable, blockchain-native credit scores designed to facilitate zero and under-collateralized lending. At its core, the protocol leverages on-chain data, machine learning and loan risk-management to effectively and profitably facilitate under-collateralized loans via the blockchain.

https://docs.roci.fi

TrueFi. September 2022.

TrueFi. November 2022. Introducing TrueFi, the DeFi Protocol for Undercollateralized Lending.

https://blog.trusttoken.com/introducing-truefi-the-defi-protocol-for-uncollateralized-lending-9bfd6594a48

ArcX. August 2022. ArcX Wiki

ARCx Credit Introduction | ARCx

ARCx Credit is a decentralized credit market on the Polygon network that offers dynamic maximum-LTV loans on ETH collateral based on a borrower's DeFi Credit Score. The DeFi Credit Score provides a credit risk assessment for individual wallet addresses based on historical on-chain borrowing activity.

https://wiki.arcx.money