Part 2: Quantitative Crypto Insight: Stablecoins and Unstable Yield

By George Liu and Matthew Turk

Tl;dr: This weblog analyzes centralized stablecoin lending yield for Compound Finance and shares our insights on efficiency, volatility, and elements that drive this yield on collateralized lending of stablecoins in DeFi. The evaluation exhibits that this lending yield can outperform the risk-free yield within the TradFi market.

In half two of this quantitative analysis piece, we’ll study stablecoin lending yield for the Compound Finance V2 decentralized finance (DeFi) protocol and share our insights on yield efficiency, volatility, and what elements are driving yield on collateralized lending of stablecoins by way of DeFi protocols. We additionally examine the “risk-free” yield in conventional finance (TradFi) to the idea of “low-risk” yield in DeFi, which we launched partially one.

ACKNOWLEDGEMENT: While we’re conscious of the current collapse of Terra’s algorithmic stablecoin TerraUSD (UST), our evaluation right here is on the world of collateralized lending yield for centralized stablecoins. We’re targeted particularly on Compound for USDC and USDT (fiat-backed stablecoins), which have disparate dangers and alternatives.

We conclude on this piece that utilizing stablecoins for low-risk (inside DeFi) collateralized lending might outperform the risk-free funding within the conventional monetary market.

USDT/USDC Yield Analysis

As talked about in part one of this blog post, a Compound consumer who has positioned their property right into a liquidity pool can calculate whole lending yield utilizing exchangeRate, which is a sign of the worth of the curiosity that the lender can anticipate to obtain over time, and the return from time T1 to T2 could be merely obtained as

R(T1,T2)=exchangeRate(T2)/exchangeRate(T1)-1.

Additionally, annualized yield for any such collateralized lending (assuming steady compounding) could be calculated as

Y(T1,T2)=log(exchangeRate(T2)) — log(exchangeRate(T1))/(T2-T1)

While the Compound liquidity swimming pools assist many stablecoin property such USDT, USDC, DAI, FEI and many others, we’re solely going to research the highest 2 stablecoins right here, i.e USDT and USDC, which have a market capitalization of $80B and $53B respectively. Together, they make over 70% of the overall market of the stablecoins.

Below are the plots of the annualized every day, weekly, month-to-month and biannual yields generated based on the formulation within the earlier part. The every day yield is considerably risky, whereas the weekly, month-to-month and biannual yields are respectively the smoothed model of the prior granular plot. USDT and USDC have comparatively related patterns within the plot, as they each expertise excessive yield and excessive volatility through the begin of 2021. This signifies that there are some systematic elements which are affecting the stablecoin lending market as a entire.

Source: The Graph

One speculation of the systematic elements that might have an effect on the lending yield are crypto market knowledge (just like the BTC/ETH costs) and its corresponding volatilities. When BTC and ETH are in an ascending development, some bull-chasing buyers could borrow from the stablecoin swimming pools to purchase BTC/ETH, after which use the bought BTC/ETH as collateral to borrow extra stablecoins and repeat this cycle till their leverage reaches the specified stage. Additionally, when the market enters right into a excessive volatility regime, there are extra centralized and decentralized crypto transactions which might improve the demand for stablecoins as properly.

Now, to examine the connection of the stablecoin yield and the crypto market knowledge, we carry out a easy linear regression evaluation to see how a lot variation within the yield could be attributed to the value and volatility elements utilizing the next method:

To measure the magnitude of those elements’ contribution, we use the R-Squared rating, which has a spread of [0, 100%]. A rating of 100% would imply that the yield is totally decided by the contributing elements.

Regression of USDC/USDT on the BTC market and the ETH market respectively lead us to the next R-Squared desk:

ETH market knowledge has a greater explanatory energy (18% & 17%) than the BTC market knowledge (16% & 11%) in figuring out the yield of USDC and USDT. This is unsurprising, significantly attributable to ETH’s elevated reputation and expanded footprint within the DeFi market because the begin of 2021. As seen with these outcomes, crypto value and volatility elements didn’t absolutely clarify the yield in stablecoins. We can conclude that there have to be different elements that assist to enhance the rating from the fundamental mannequin.

We carried out additional exploratory evaluation by introducing the historic stablecoin provide knowledge and MACD technical indicator value knowledge to the mannequin. The stablecoin provide (the overall variety of stablecoins provided to Compound liquidity swimming pools) ought to — intuitively — have an effect on the supply/shortage of the stablecoins and not directly affect the yield. MACD is a crucial momentum buying and selling sign (subtracting the 26 interval EMA from the 12 interval EMA — on this case on value) because it might assist momentum buyers to resolve when to leverage and when to deleverage.

We see a noticeable improve in R-Squared scores, as each USDC and USDT bought a bump to a stage round 60%-70% as proven under.

From this knowledge we will conclude that stablecoin provide is a considerable contributing issue, because it alone is ready to carry the rating to round 60% for each stablecoins in any of the 2 markets. It appears to counsel that [supply] is a significant component in affecting the yield within the stablecoin lending market. This is similar to the TradFi world, the place credit score provide by the Federal Reserve will have an effect on the overall rate of interest of the entire system.

The introduction of MACD knowledge (on BTC and ETH value) brings combined enchancment. In the case of the BTC market, its unbiased contribution is much lower than the availability issue, and the marginal profit over the shoulder of provide is only some share factors. We observed within the ETH market, nevertheless, that MACD has a larger unbiased contribution to the R-Squared worth as in comparison with the BTC market. This means that stablecoin lending yields are extra correlated with momentum primarily based buying and selling exercise in ETH than in BTC.

An instance of the regression coefficients for USDC lending yield within the ETH market are displayed under. The desk means that increased ETH costs, volatility and [stable coin supply] are usually related to decrease USDC lending yield. At the identical time, the stronger the MACD sign is, the upper the yield would go.

Comparison to the Traditional Risk-Free Yield

While it’s attention-grabbing to disclose what has pushed the low-risk yield on stablecoin lending, additionally it is vital to check these yields with the counterpart within the TradFi market.

Because stablecoin lending yields are derived from the realized floating rates of interest for collateralized loans on the Compound platform, we chosen the General Collateral (GC) price used within the conventional cash market because the comparable risk-free price, as a result of additionally it is a floating price with treasury debt because the mortgage collateral.

Below is a plot of the portfolio worth of the investments that earn USDC lending yield, USDT lending yield, and GC price yield respectively. The investments all begin with $100 preliminary worth on 2020–05–01, and finish on 2022–05–01. As seen under, yield on USDT and USDC collateralized lending is increased than the GC price by a big margin. On the opposite hand, risk-free funding that earns GC price hardly grows for a similar interval.

The common rate of interest within the desk under additionally confirms that GC price is on common round 0.08%, whereas USDC and USDT lending yields are respectively 3.71% and 4.51% for this era as seen under. (We additionally checked the 2Y time period yield on the treasury debt on 2020–05–1 which is merely 0.2%)

For the foreseeable future, it’s affordable to conclude that the low-risk price, throughout the crypto market not less than, will proceed to outperform the risk-free price within the TradFi market. One motive for that is the good contract danger, or liquidation danger talked about in part one of this weblog. However, a bigger motive is the slower development within the stablecoin provide relative to the expansion within the crypto financial system as a complete. By comparability, the TradFi market has seen main credit score development because the begin of the Covid-19 pandemic, which has helped to drive the risk-free price to historic lows (see Fed steadiness sheet development under).

Conclusions

This weblog supplied a broadly indicative evaluation of the low-risk yields out there from collateralized lending of stablecoins by way of DeFi protocols. While these yields could also be very risky each day, their basic development could be defined comparatively properly by BTC/ETH costs, volatilities, stablecoin provide and MACD (momentum buying and selling actions). We additionally in contrast these yields with the risk-free price within the TradFi market the place we see constant outperformance within the crypto market. To reiterate, this isn’t monetary recommendation.

Next steps

We, as a part of the Data Science Quantitative Research staff, intention to get a holistic understanding of this house from a quantitative perspective. We are on the lookout for individuals which are passionate on this effort to affix our rising staff. If you have an interest in Data Science and specifically Quantitative Research in crypto, come join us.

The evaluation makes use of the Compound v2 subgraph made out there by way of the Graph Protocol. Special due to Institutional Research Specialist, David Duong, for his contribution and suggestions.

NOT INVESTMENT ADVICE

The content material is for informational functions solely, is basic in nature and shouldn’t be relied upon or construed as authorized, tax, funding, monetary, a promise, assure, or different recommendation. Nothing contained herein constitutes a solicitation, advice, endorsement, or provide by Coinbase or its associates to purchase or promote any cryptocurrency or different devices in any jurisdiction wherein such solicitation or provide can be illegal underneath the legal guidelines of such jurisdiction.

Part 2: Quantitative Crypto Insight: Stablecoins and Unstable Yield was initially revealed in The Coinbase Blog on Medium, the place persons are persevering with the dialog by highlighting and responding to this story.