• Interest rate arbitrage in the DeFi space refers to taking advantage of interest rate differences across DeFi platforms. DeFi-CeFi arbitrage also exists when there are interest rate differences between centralized and decentralized platforms.

• Conversely, carry trade strategies may be pursued, i.e., it is possible to borrow with a low-interest asset to invest in an asset yielding higher returns.

• However, all existing DeFi platforms rely on over-collateralization requirements from the borrower’s perspective, which limits potential arbitrage opportunities.

• Several potential market inefficiencies may subsist both within DeFi and with CeFi platforms, which can be explained by the following reasons:

  • Nascency of the DeFi space: illustrated by low liquidity, visible on the size borrowable amounts and daily volumes transacted (relatively to CeFi).

  • Platform-specific risks: interest rate volatility, potential smart-contract issues, loan matching uncertainty, redemption risks, etc.

  • Different platform & protocol characteristics: all protocols and platforms, whether they are decentralized or centralized, have different expected characteristics, collateralization requirements, and terms of the agreement.

• Eventually, as the DeFi space matures, new opportunities are expected to become more scarce. Interest rates across platforms should also converge, as long as assets and platforms share similar risk characteristics. New platforms and protocols like interest rate swaps may contribute to a broader range of trading opportu

DeFi (Decentralized Finance) has been growing significantly since mid-2018, with Ethereum being one of the most prominent blockchains for platforms and developers1. This report describes trading strategies, namely arbitrage and “carry trade”, within the DeFi space and between DeFi and CeFi2 platforms. Besides, this report aims to discuss potential price inefficiencies along with risks and constraints that must be considered when building these advanced positions.

1. General definition(s)

In the financial industry, arbitrage has relied on market price inefficiencies. Namely, greater opportunities tend to exist in less mature markets. As a result, the crypto-industry exhibited many price efficiencies in its early days3.

Arbitrage is defined by Binance Academy as:

“Arbitrage is the practice of buying and selling assets over two or more markets as a way to take advantage of different prices. [...] Arbitrage exists as a result of inefficiencies in the markets. ”

In the DeFi space, arbitrage opportunities occur when there are differences in interest rates among platforms. In general, interest rate arbitrage links to situations where the lending rate is below the borrowing rate on another platform. We can define “pure interest arbitrage” in the DeFi space as:

“DeFi interest rate arbitrage refers to taking advantage of interest rate differences across decentralized platforms.“

The most straightforward case is when the borrowing rate on one platform is lower than the lending rate on another platform.

However, arbitrage opportunities may also occur with custodial platforms (e.g., custodial crypto-lending services, like BlockFi and Nexo, and non-crypto financial services providers).

Opportunities also exist with non-blockchain financial providers like commercial banks or lending companies.

“DeFi-CeFi interest rate arbitrage refers to taking advantage of interest rate differences between decentralized and centralized platforms.“

Furthermore, carry trade strategies are also discussed in this report. According to Investopedia, they are defined, such as:

“A carry trade is a trading strategy that involves borrowing at a low interest rate and investing in an asset that provides a higher rate of return. A carry trade is typically based on borrowing in a low-interest rate currency and converting the borrowed amount into another currency, with proceeds placed on deposit in the second currency if it offers a higher rate of interest.”

However, in this report, carry trades are referenced such as benefiting from interest rate differentials between stablecoins, supposedly pegged to the same fiat currency (i.e., the US dollar).

The next two sections discuss both carry trade and arbitrage opportunities4 that have occurred between platforms through the analysis of case studies, both from an existing and an empirical perspective.

2. DeFi-CeFi case studies

2.1 Retail platforms and DeFi protocols

2.1.1 Trade description

Table 1 - Comparison of interest rates on retail platforms assuming a FICO score > 7205

CeFi borrowing rates on USD typically do not require users to post any initial collateral. However, short-term maturities are typically expensive.

As a result, a simple strategy could potentially be coordinated around DeFi and CeFi platforms.

For this strategy, Compound is used as a representative example for DeFi as this protocol has the most significant amount of USDC borrowed/supplied6 (more than 30 million as of September 20th 2019) while offering immediate compounding of interest rates.

Table 2 - Set of steps

This strategy would generate a positive annual interest rate as long as:

However, a more realistic version would be to move some USD from a savings account to Compound/dYdX to maximize expected returns.

Then, it would be worth moving USD to Compound from a saving account as long as:

2.1.2 Risks and considerations

In the previously described strategy, several key elements need to be considered along with potential existing risks that may hinder the expected returns of this particular strategy. Some of these relate to how interest rates on Compound are defined and evolve. Specifically, borrowing and lending interest rates are modeled based on a linear function, with both a floor and a cap for each asset, which respond to real-time demand and supply in the protocol.

On the other hand, CeFi platforms often have variable interest rates, which fluctuate over time based on the macro-economic environment along with specific terms of agreements that often vary from one platform to another.

Hence, different considerations must be considered when building a CeFi-DeFi arbitrage position:

• Variable-rate volatility: in most of the DeFi platforms, interest rates fluctuate based on protocol demand and supply mechanisms. As a result, it introduces volatility risk related to the lending rate. As the DeFi industry remains small, platform-specific demand and supply dynamics can diverge significantly from the off-chain equilibriums.

• Prepayment penalty on the CeFi loan: many centralized platforms do not allow an individual to repay loans early or there may be prepayment penalties for early repayment of the loan. Hence, if the DeFi lending rate were to move lower than the CeFi borrowing rate, traders would be faced with two main choices:

  1. To wait and maintain a negative yielding-profit strategy.

  2. To pay the prepayment penalty.

• Gas fees: these fees must be considered and incorporated in the assessment of the profitability of the strategies. For practical reasons, gas fees are omitted from this analysis. As gas fees are typically fixed (i.e., irrelevant of the amount transferred from one address to another), this omission would only marginally impact the profitability of a strategy, as long as the amount transacted is significant.

Eventually, other risks are related to Compound (or the DeFi protocol)7:

• Withdrawing funds risk: as funds lent on Compound don’t have any maturity date; it may be hard to redeem instantly depending on the borrowing-lending differentials.

• Central administrator: Compound decides and updates interest rate models8. However, it is worth noting that the model specifications are fully transparent, but admin keys could potentially be compromised.

• Smart contract risks: like most DeFi smart-contract-based protocols, there may be some risks from smart-contracts. However, the development of insurance platforms may mitigate some of the risks that are smart-contract specific for every single platform. For instance, it is possible to subscribe an insurance on Nexus Mutual for a Compound smart-contract for a fixed time-length and for a particular amount.9

2.2 BlockFi and dYdX

2.2.1 Trade description

BlockFi is a custodial crypto-platform that allows users to borrow and lend large market-cap cryptoassets such as XRP, Bitcoin (BTC) or Ethereum (ETH).

As the lending interest rate on BlockFi is higher than the borrowing rate on dYdX, it appears that these interest rate differences may be arbitrage-able.

The table below describes the two-step process required to create an arbitrage position between dYdX and BlockFi for large DAI holders.

Table 3 - Set of steps (as of September 23rd 2019)

2.2.2 Risks and considerations

In a similar fashion to other positions, several risks need to be considered, such as:

• CeFi interest rates are set manually: BlockFi sets borrowing and lending interest rates (see terms and conditions10). Historically, previous rate changes were significant11 with near immediate effect. Hence, this may impact the profitability of strategies.

• Price volatility risks: if the price of Ethereum were to increase, the position of dYdX could be liquidated. As a result, the initial collateralization ratio is essential to consider and traders need to monitor it over time.

• Smart contract risks: like most DeFi smart-contract-based protocols, there may be a risk of issues in the general behavior of the smart-contract. Similar to the finding in subsection 2.1.2, some insurance platforms can be used for insuring against platform-specific smart-contract risks on dYdX12.

• Uncertainty regarding loans: BlockFiloan matching process is not immediate, which could impact the profitability of the strategy. Specifically, it could lead to negative returns if the matching process took too much time.

Specifically, the loan would not be profitable if:

: the time it takes for a loan to be matched on BlockFi.

: expected time length for the trading opportunity to remain valid.

2.3 dYdX and Binance Lending

2.3.1 Trade description

Binance Lending allows users to hold funds in subscribed Binance Lending products, with fixed maturities, and to grow holdings by accruing interests.

Table 4 - Set of steps (as of September 23rd 2019)

As the Binance’s ETH lending interest rate is higher than dYdX’s borrowing rate, this trade would return a positive profit.

Someone could use DAIs as collateral to borrow ETH on dYdX and then transfer onto Binance to subscribe to a lending offer for Ethereum. Once again, collaterals parked on dYdX also receive lending interest rates and any Dai, used as collaterals, would receive lending rates too.

2.3.2 Risks and considerations

Before establishing this position, there are three main potential risks which may impact this position:

• Price volatility risks: if the price of Ethereum were to increase, the position of dYdX could be liquidated. As a result, the initial collateralization ratio is vital to consider and traders need to monitor this ratio over time.

• Smart contract risks: like most DeFi smart-contract-based protocols, there may be a risk of issues in the general behavior of the smart-contract.

• Volatility in DeFi rates: if the borrowing interest rate on dYdX were to increase above the lending interest rate on Binance Lending, this position would lead to negative returns. However, the differential in interest rates is currently big enough to prevent such a case from occurring.

Furthermore, there are other additional major components and constraints to consider:

• No prepayment on Binance Lending: unlike many platforms, funds provided on a subscription cannot be redeemed before the maturity date. Hence, the funds cannot be used to reduce a position in the adversity of any of the risks mentioned previously.

• “Missing the subscription”: as Binance Lending subscription offers are prevalent and competitive, users may be prevented from participating as there are caps per subscription, both a total subscription size and respective individual limit.

• Binance withdrawing fees: like many centralized exchanges, Binance has a fixed fee policy to withdraw assets14. However, as the maximum size of the subscription is quite large, this would only marginally impact the profitability.

• Gas fees: gas fees must, once again, be considered and incorporated in the assessment of the profitability of the strategies. For practical reasons, gas fees were omitted from this analysis. As gas fees are typically fixed (i.e., irrelevant of the amount transferred from one address to another), this omission would only marginally impact the profitability of a strategy.

The next section discusses “pure DeFi” strategies. Namely, carry trade strategies and arbitrage strategies are investigated in the next section, through the analysis of several case studies.

3. Pure DeFi arbitrage and carry trade strategies

3.1 USDC and DAI interest rate differentials

One carry trade opportunity exists if it is possible to obtain a lending interest rate higher on one platform than on another platform.

3.1.1 Trade description

Despite both cryptocurrencies being pegged to the US dollar, DAI and USDC display extremely different lending interest rates, with DAI exhibiting much higher lending rates.

Chart 1 - Historical lending rates on USDC and DAI on Compound v2

Sources: Binance Research, LoanScan, Compound

Table 5 - Set of steps (as of September 23rd 2019)

3.1.2 Risks and considerations

Like other carry trade strategies, it displays potential high risks mostly owing to peg deviation with DAI and USDC prices deviating from their respective peg. Interestingly, if the price of DAI was to appreciate against USDC (e.g., August 2019 on the chart below), it could lead to an additional source of profit.

Chart 2 - OHLC price of DAI on Coinbase Pro (in USDC)

Sources: Binance Research, Coinbase

Here are some of the largest potential reasons which could lead to fluctuations in DAI price:

• Supply cap of DAI: with a cap at 100 million units in its existing system, DAI supply cannot increase to respond to market demand. Hence, this strategy may lead to higher price uncertainty if the circulating supply increases, getting close to this boundary16.

• Downward movements in the price of ETH: if ETH price goes down, there may be a demand shock on DAI from CDP-holders as they may prefer to close existing CDPs, instead of parking more ethers. As a result, it may lead to a potential short-term price increase owing to market demand dynamics.

• Lack of liquidity to exchange DAI to USD (and collateralized stablecoins): reasons such as thin orderbooks or lack of credible trading venues may prevent the proper execution of the strategy.

As this strategy only relates to USDC-holders, the next subsection discusses a more advanced trading strategy, which can be executed by ETH-holders.

3.1.3 Alternative case for an ETH-holder

Ethereum’s lending interest rates on Compound and dYdX are respectively lower than 0.03% and 0.10% APR, as of September 23rd 2019.

As the borrowing rate on USDC is inferior to the lending rate on DAI (see chart 3), ETH-holders could potentially consider engaging in a carry trade to earn an additional profit.

Chart 3 - Historical USDC borrowing rate and DAI lending rate on Compound v2

Sources: Binance Research, LoanScan, Compound.

Chart 4 - Spread between DAI lending rate and USDC borrowing rate on Compound v2

Sources: Binance Research, LoanScan, Compound.

Spread is the difference between the DAI lending rate minus the USDC borrowing rate.

Table 6 - Set of steps (as of September 23rd 2019)

However, the most accurate version of the previous inequality includes expectations of the length of this opportunity. Hence, it would be worth engaging in this trade if:

: expected time length for the trading opportunity to remain valid. : transaction fee (in % of the amount traded) at step 3.17

3.2 Maker-Compound arbitrage (historical case)

The Maker/Compound arbitrage opportunity refers to a historical opportunity that occurred in late 201818.

DAI stablecoins are minted in the MakerDAO’s ecosystem, through the creation of CDPs which are backed solely by ethers (ETH). Specifically, the mechanism relies on over-collateralization. DAI are backed by ethers (ETH), parked in CDPs, which are over-collateralized by at least 150% of the equivalent value in USD worth of ethers.

3.3.1 Description

In late 2018, Compound’s lending interest rate on Dai was higher than the respective interest rate in the MakerDao ecosystem (i.e., stability fee). As a result, it was possible to open CDPs collateralized by Ethereum to earn an annualized interest rate higher than the lending interest rate on ETH.

Chart 5 - Daily lending DAI rates on Compound v1 and stability fee on Maker (%) between November 1st 2018 and May 1st 2019

Sources: Binance Research, LoanScan, MakerDAO, Compound.

Table 7 - Set of steps (as of December 1st 2018)

This strategy generated a positive profit as long as the following inequality holds:

Unlike Compound and dYdX, it is also worth noting that ether parked in a CDP do not return any interest, thus representing an opportunity cost.

Hence, it would have been worth conducting this trade as long as the following inequality holds:

3.3.2 Risks and constraints

In the strategy defined in the preceding subsection, there were some key constraints to consider to assess the profitability of the profitability strategy:

• MakerDAO’s over-collateralization requirement: to mint Dai in the ecosystem, it is required to post collateral with a required initial collateralization ratio equal to 150%. Furthermore, CDPs are at risk whenever the collateralization ratio goes below 150%. Hence, the median collateralization within the system is much higher (368% as of September 23rd 2019). As ethers parked in a CDP do not return interest rate, increase protection from collateral price decrease ties to lower expected return.

• Gas fees: this strategy would also need to incorporate elements such as gas fees to open a CDP, deposit ethers, mint DAI, park DAI on Compound, and withdraw DAI from Compound and close CDPs.

Similarly, some risks needed to be considered:

• Liquidation risk: despite this strategy being profitable, if the price of ether had gone down further, it would have put CDPs at risk, requiring the user to either transfer some of the DAI to reduce the CDP, or to increase the collaterals.

• Compound redemption risk: there were historical cases where more than 98% of the funds parked on Compound were borrowed19, hence preventing the protocol from processing immediate redemptions of all borrowed funds. However, this could lead to spikes in interest rates owing to this, which could increase the expected rate of return on this trading strategy.

• Platform-specific risk: all these platforms could have had price oracle mismatches as MakerDAO & Compound use two different price oracles20.

Eventually, this opportunity ended once the stability fee was hiked several times, through a series of votes by MKR token-holders.

Without considering historical gas fees, this trading position would have resulted in a “pre-gas” positive profit for three months, starting on December 1st 2018 and ending on March 9th 2019, and would have yielded a total return of around 0.87% i.e., 3.25% annualized.

Chart 6 - OHLC price of Ethereum (in USD)

With a close price at $119 on December 1st 2019, ETH price reached a low of $83 on December 15th 2019. As a result, the minimum initial collateralization ratio that would have been required to prevent any liquidation, assuming no additional deposit of ether in the CDP, was around 215%.

4. Conclusion

Part of the DeFi Series, this report discussed several arbitrage strategies along with carry trades in the Ethereum space.

Arbitrage opportunities appear to be longer-lasting between CeFi and DeFi than within DeFi. Potential reasons may explain this such as the lack of maturity of the DeFi space, the importance of over-collateralization in all protocols, and interest rate setting mechanisms that reflect protocol supply/demand factors (rather than global macro-environment dynamics). On the other hand, different risk profiles between DAI and USDC may explain this persisting divergence between their respective interest rates.

Other important aspects are also worth to be noted:

• Long-term asset allocation perspective: some trades are profitable, based on the asset(s) someone wishes to retain in the long-run. For instance, Ethereum lenders receive extremely low-interest payments denominated in ether21. As a result, strategies appealing for ETH-holders may not attract other token-holders. For example, someone holding DAI/USDC may be disincentivized to build complicated positions.

• Ability to diversify the interest exposure: arbitrage positions such as the one described between Maker and Compound can allow individuals to get exposure to new assets.

Furthermore, it is worth noting that “second-layer” DeFi protocols and platforms have been built, which introduced new features like:

• Platform bridges: platforms like InstaDapp22 offer an automated way of moving loans between Maker and Compound.

• Protocols and contracts to automatically allocate to the best platforms such as MetaMoneyMarket23 or RAY (Robo-Advisor for Yield)24

• Mutualization of interest rates across protocols: platforms offering an interest rate aggregated from different DeFi protocols could help to offset risks from one platform, without the need to pay recurring gas fees. Hence, it could further lead to interest rate convergence amongst protocols.

Furthermore, new aspects must also be considered for future analysis like oracle price arbitrage opportunities or additional decentralized platforms like Uniswap. Ultimately, interest rate swaps25 or markets to hedge against the direction of interest rates26 could further bring a whole new range of trading opportunities.

References

• Amber Group. Interest Rate Arbitrage on Ethereum (2018). https://medium.com/amber-group/interest-rate-arbitrage-on-ethereum-621813ae7c0

• Compound. Compound: The Money Market Protocol (2019). https://compound.finance/documents/Compound.Whitepaper.pdf

• Dharma (Max Bronstein). Dharma Markets Reports # 6 - Trading Strategies in DeFi (2019) https://medium.com/dharma-blog/dharma-markets-report-6-trading-strategies-in-defi-b7651378bedb

• dYdX (Antonio Juliano). dYdX: A Standard for Decentralized Margin Trading and Derivatives (2017). https://whitepaper.dydx.exchange/

• Kyber Network. Kyber: An On-Chain Liquidity Protocol v0.1(2019) https://files.kyber.network/Kyber_Protocol_22_April_v0.1.pdf

• Pintail. Understanding Uniswap Returns (2019). https://medium.com/@pintail/understanding-uniswap-returns-cc593f3499ef

• Placeholder (Alex Evans). DeFi Liquidity Models (2019). https://www.placeholder.vc/blog/2019/4/9/defi-liquidity-models

• Uniswap. Uniswap Whitepaper (2019). https://hackmd.io/@477aQ9OrQTCbVR3fq1Qzxg/HJ9jLsfTz?type=view

1. See our recent report about “The World of Tokenization”. https://research.binance.com/analysis/tokenization↩

2. CeFi stands for Centralized Finance. For this report, it includes both non-crypto financial institutions and custodial platforms like centralized exchanges (e.g., Binance, Coinbase) or centralized lending platforms (Nexo, BlockFi).↩

3. Our past report about “Diversification Benefits with Bitcoin” briefly discusses cross-exchange price inefficiencies that have become almost non-existent owing to the increasing presence of sophisticated arbitrageurs. https://research.binance.com/analysis/bitcoin-diversification-benefits↩

4. It is worth noting that these are largely arbitrage-like opportunities. Arbitrage implies price differences between the same asset and thus a guaranteed profit opportunity. In this report, arbitrage opportunities also refer to price differences between very similar assets (e.g., USDC and USD).↩

5. Data retrieved from The Simple Dollar on September 23rd 2019, assuming residence in California with an excellent FICO credit score above 720.↩

6. DeFi Pulse ranks Compound as the second-largest DeFi application as of September 23th 2019.↩

7. An example of the risks associated with Compound. https://medium.com/@ameensol/what-you-should-know-before-putting-half-a-million-dai-in-compound-fafdb2645f77 Similar risks exist in most of the platforms. Platforms are audited by third parties. Example of audits with Compound or dYdX.↩

8. Compound plans on migrating to a decentralized governance.↩

9. As of September 10th 2019, it would cost to insure 5 ETH in Compound around 0.30% for 90 days or 1.30% for 365 days.↩

10. https://blockfi.com/terms/↩

11. https://www.coindesk.com/crypto-lending-startup-blockfi-slashing-interest-rates-on-ether-deposits↩

12. For instance, on Nexus Mutual, it would cost to insure 5ETH around 2.17% for 90 days and 8.57% for 365 days It is worth noting that insurance for dYdX is much more expensive than for Compound.↩

13. https://www.binance.com/en/lending/products?utm_source=BinanceResearch↩

14. For instance, ETH has a fixed withdrawal fee of 0.01 ETH. https://www.binance.com/en/fee/schedule↩

15. It would arguably lead to a hike in the stability fee, which could potentially lead to a subsequent interest in this carry trade strategy as Compound market rates would likely also move upward.↩

16. It would need to be paid twice: to buy DAI with USDC and sell DAI back to USDC.↩

17. This opportunity was also covered in a research report from Amber Group.↩

18. https://medium.com/@ameensol/what-you-should-know-before-putting-half-a-million-dai-in-compound-fafdb2645f77↩

19. Compound: https://github.com/compound-finance/compound-oracle/blob/master/docs/Oracle-Specification.pdf MakerDao: https://github.com/makerdao/medianizer↩

20. Interestingly, it can be explained by a persistent mismatch between lenders and borrowers. There is little demand to borrow Ethereum, potentially, because nobody wants to short ETH.↩

21. https://medium.com/instadapp/bridge-protocols-eb14b2dd0fe7↩

22. https://medium.com/primeradiant/introducing-metamoneymarket-62b23969c7ea↩

23. https://medium.com/@staked/https-medium-com-staked-announcing-robo-advisor-for-yield-ray-9d7395e84769↩

24. https://cherryswap.now.sh/↩

25. Example of Suma built at ETHBoston in 2019. https://devpost.com/software/suma↩