These researches have been made with Redacted Labs as part of our ongoing exploration of AMMs.

We have already conducted a study that was released last month, published in the Redacted Labs’ Mirror which was based on data limited to a 90-day history. In this article, we extend our analysis to cover a 365-day historical period to assess the significance and relevance of different approaches.

---

We are embarking on a new series of analyses that will delve into LPing in AMMs. The dominance of Uniswap has prompted us to focus our studies on this protocol, using it as a reference for LP strategy creation.

Several analytical tools are being used, including Nuant*, which allows us to access a vast database and cross-reference it in a customized manner to obtain answers to our questions.

Explaining the workings of AMMs and conducting a comprehensive study of the DEX market are beyond the scope of this article.

The main goal is to understand how to construct LP strategies that outperform the market while maintaining a neutral or directional exposure to ETH. The creation of such highly customized strategies is made possible through Maverick and, soon, Uniswap v4 Hooks, which we eagerly anticipate.

Table of Contents:

1. The AMM Market Dominated by Uniswap

2. Scope of Our Study

3. Identifying the Most Attractive Pools

4. Conclusion and Key Takeaways

---

1) The AMM Market Dominated by Uniswap

Liquidity provision is a flagship activity in DeFi, ensuring competitive trading on decentralized platforms while serving as a source of income for providers.

Today, this market is largely dominated by Uniswap, accounting for over half of the volume and having 28.50% of the total TVL of DEXs (06/02/2024)

Concentrated liquidity has brought significant innovation in terms of capital customization and efficiency, a concept that will be adopted numerous times thereafter by projects like Cowswap, Ambient, Maverick and Trader Joe.

---

2) Scope of Our Study

In order to create strategies that meet our requirements, we need to filter pools that :

• Are neutral ; stable/stable pools and ETH/ETH-variant pools.

• Are correlated with ETH ; the volatility of the asset paired with ETH outperforms ETH itself, allowing for reduced IL while being directional on ETH.

The overall study does not go into detail on stable/stable pools because the stakes are lower; the degree of customization is lower, and "the dynamics of stable pairs are completely different; for example, the lowest fees are much more attractive" (quoted from Atiselsts.eth's article).

The study of ETH/ETH-variant pools is very interesting and will be addressed in due course.

So, we started by finding out : which pools have high statistics relative to their TVL, and which ones have the highest degree of correlation with ETH ? In other words, are there one or more pools from which we can profit due to their high on-chain activity while having reduced IL ?

---

3) Identifying the Most Attractive Pools

For scalability of yield in setting up our strategies, as well as to define the scope of our study, we have only considered pools with over 2M TVL, excluding those related to USDC, USDT, DAI.

To make the study more relevant and more “fair” among all the assets, we also decided exclude new tokens for which the TGE took place recently. This includes : MKR_100bps, PORK, RBX, INST, ELON, GEL, ONDO, PRIME, 1CAT, GALA, MUBI

New trendy token such as PORK, ONDO, PRIME, 1CAT and MUBI are pretty interesting to study because they provide astronomical volumes and fees.

We also plan to extend our study to all Uniswap’s pools later.

Our definition of “Attractive pools" here, is the ones which have high on-chain statisticS and strong correlation with ETH.

3.1) Volume vs TVL

Recently, SOL/ETH and INJ/ETH outperformed the market for couple of week due to the recent hype :

I agree that the graph is quite messy but allow us to discern fleeting trends. Here are the top 5 assets :

The metrics are pretty insane for INJ which largely leads the market. Seeing wBTC here is not surprising, but I would be surprise to see it in the fees/TVL leaderboard as the the wBTC/ETH has only 5 bps.

3.2) Fees vs TVL

Obviously Fees / TVL graphs are similar with the Volume / TVL because the pools mainly have 30 bps and among the token excluded from the analysis (MUBI, 1CAT …), they have 100 bps which may impact a lot the LP profitability ;

Same here, take a glance to the top 5 assets

3.3) First interpretation of the results

INJ is largely outperforming the market by both the volume (average at 14,59%) and the fees (average at 0,053%). The gross yield is 73% APR, really impressive. But, you can have an IL if INJ price under-performed ETH price.

INJ keeps the top position because of its insanely high volume and we see RDNR and AGIX appear, which both have 100 bps.

These data show us how important it is to keep in mind the fee tiers level. We won't delve further into the details regarding the choice of fees, Ambient already provided in-depth researches about it in their articles

If we compare these results with the previous article in which we took 90 days of history, we notice that none of these assets except AGIX appear in the top positions. Two reason ; 1) because we exclude new tokens here and 2) because the market dynamic changes a lot;

If INJ, RNDR and SOL are able to give high yield, are they strong enough vs ETH to be used in LPing strategies ?

3.4) Study of the degree of correlation

Three data sets were used to study the correlation: Pearson coefficient, scatter plots, and positive price evolution compared to ETH.

3.4.1) Pearson Coefficient :

The Pearson coefficient is a statistical measure that assesses the degree of linear relationship between two quantitative variables. Often denoted by the letter "r", it varies between -1 and 1 and is calculated by taking the difference relative to the means:

Interpretation:

• r = 1: Perfect positive correlation. When one variable increases, the other also increases

• r = -1: Perfect negative correlation. When one variable increases, the other decreases.

• r = 0: No linear correlation. The variables do not appear to be related

Values between -1 and 1 indicate the degree of correlation : the closer the absolute value of r is to 1, the stronger the correlation.

In our case, we calculated this coefficient over 30d, 90d and 365d :

Which coefficient should we look at ?

30d ; we already look at this metric last month and the results change a lot ; LINK is actually uncorrelated over the 30 past days whereas last month it was in top position. In fact, look at the Paerson coefficient over 30d doesn’t make sense because it depends too much on the trend. Moreover, our approach would be to say that LPing strategies have a medium-term horizon.

So, 90 days seems to be more suitable metric to analyse because short-term ephemeral movements have less impact. 365d Paerson coefficient is quite too long to my opinion but it can be use as a complementary metrics for 90d Paerson.

wNXM have INST have surprising metrics as the correlation level is close to 1 for both 90d and 365d ! Hypothetically, they can largely be consider in our strategies. But, will it still be interesting if on-chain statistic are low ?

Look at this data :

• wNXM : 90d Paerson = 0,982 ; 365d Paerson = 0,888 ; Vol/TVL = 2,77% ; Fees/TVL = 0,028%

• INST : 90d Paerson = 0,921 ; 365d Paerson = 0,904 ; Vol/TVL = 1,32% ; Fees/TVL = 0,013%

On-chain performance are really poor. Honestly, I will never LPing if the pool provides less yield than other more simple DEFi strategies, such as PT_Pendle or ETH/ETH-variant pools. “More risk” for less rewards …

Let's remember one thing : Paerson coefficient is a great tool if you’re looking for passive LP strategies with a confortable range. If you’re using it to set-up strategies, you may have a long-term horizon and no emergency to exit your LP, because some ephemeral decorrelations can implied IL. Take a glance to the next graph displayed to understand it.

Let’s compare top 3 tokens which have the higher correlation with ETH with top 3 tokens in term of on-chain statistics :

These graphs highlight if the assets tend to frequently be decorralated ; in other words, how often I risk having significant impermanent loss if I’m forced to close my LP position.

Obviously, as I said before, decorrelation can be good if the asset linked to ETH outperforms.

3.4.2) Scatter Plots

A Scatter Plot is a graph where the abscissa and the ordinate represent the two assets for which we want to determine the correlation. Each point on the graph represents an individual observation, with its position determined by the values of these two variables.

Interpretation:

• If the line is ascending, the two assets have a positive correlation.

• If the line is descending, the correlation is negative.

• If the points are scattered without a clear shape, there is little or no correlation

For the sake of simplicity and synthesis, let's see how this looks for the 6 tokens studied above :

Scatter Plots graphs show us the high correlation of wNXM, INST and wBTC. These graph don’t show other interesting insight as we already have this data with Paerson. To our point of view, for both Scatter Plots and Pearson coefficient alone are not efficient ; our goal is to outperformed ETH. What are the current limitations ?

• An assets which often outperformed ETH will not appear in Paerson top position

• Most of the time, assets which are highly correlated to ETH doesn’t perform well in term of on-chain statistic

3.4.2) Positive price evolution vs ETH

Here we are : does the asset paired with ETH in the pool pose a risk of devaluation compared to ETH ? In other words, am I at risk of IL ?

Let’s take a glance for the 6 tokens ;

All these tokens have outperformed ETH during 2023. What difference do we notice between the two categories?

• INJ has highly outperformed ETH by more than 1700%. The price vs ETH has been quite flat during most of the 2023 year and then skyrocketed. We can see the same kind of chart for SOL but with a lower amplitude

• RNDR has been more volatile and largely outperformed ETH as well.

• wNXM is the perfect use case of the pattern we’re looking for ; increase slowly vs ETH making the rebalancing easier to handle

• INST seems a bit less suitable for LPing

• wBTC volatility is strong, but the amplitude is the lower one. Set up tiny range to earn most of the fees into this competitive pool may be quite hard to manage

Having a decorrelation vs ETH is not a bad thing in our case. Charts like INJ, SOL and wNXM are pretty interesting because they offer an “easy’ way to manage an LP position while still earning against ETH.

Here is a summary of the data studied :

The tools like Pearson's coefficient or scatter plot are useful for determining the flexibility that an LP strategy can have, but they are not sufficient to confirm a relevant correlation between two assets. Other tools are necessary, and the approaches vary depending on the strategies.

We are comparing 2 very interesting case studies based on their respective criteria. Now the question is : should we prioritize returns at the expense of correlation ?

---

4) Conclusion and Key Takeaways

These results show that certain pools appear to be more suitable for LPing strategies.

Liquidity providers view returns as a source of income to counter impermanent loss, hoping to profit without truly anticipating the challenges involved in this DeFi investment.

Hedging is possible directly through concentrated liquidity within the same pool by applying mirrored out / in ranges, but greatly reduces capital efficiency as a portion will always remain dormant. Hedging via options and derivatives poses the same question of efficiency.

Therefore, we are seeking more efficient solutions, and here is what we have learned in this first part of the study ;

1. The Pearson coefficient and scatter plots, widely used in finance, are not sufficient to confirm a strong correlation. However, they are useful for determining the flexibility of an LP strategy ; can you close your range at any time without incurring too much IL ?

2. In the case of a ETH directional strategy, the approach differs slightly. It is important that the pool's asset be resilient against ETH as the goal here is to accumulate ETH.

3. On-chain metrics are the ones we should take attention to. Most of the tokens with high statistic outperformed ETH

4. Even if you determine the largest pools in terms of Volume vs TVL and Fees vs TVL, this is not enough as some of the liquidity may be out-of-range. Therefore, a pool generating less fees/volume but having a good degree of correlation and most of its liquidity out of range may present a good opportunity to introduce your concentrated liquidity optimally.

Correlation between tokens, resilience vs ETH and basic on-chain metric are great to quickly see if some pools have advantage over the other for LPing ; but it’s still too risky.

In the upcoming articles, we will delve deeper into which liquidity is really used into a pool. We will go further in determining the volatility of the assets and the risk they represent if implemented in LP strategies.

We will detail the concentrated liquidity of the chosen pools, integrate new metrics such as IV/HV, Sharpe ratio, Delta / Gamma / Vega and more.

Feel free to subscribe to receive notifications about upcoming publications :)

---

Disclaimer:

This article is provided for informational purposes only and does not constitute legal, commercial, or investment advice. Do not base your investment decisions on this article and do not consider it as an accounting, legal, or tax guide. Mention of specific assets or securities is for illustrative purposes only and not an endorsement. The author's opinions may not reflect those of their affiliations and are subject to change without notice.

About Redacted Labs:

Redacted Labs is a DeFi desk offering customized services to professionals looking to profit from decentralized finance with institutional risk management.

Our goal at Redacted Labs is to provide our clients with the decentralized finance experience while controlling risks, offering a fund management service tailored to their needs. We firmly believe that decentralized finance is the future, and our expertise and experience enable us to meet the growing needs of businesses looking to invest in this area.

About Nuant:

Nuant QIS (Quantitative Intelligent System) is an advanced integrated solution dedicated to the decentralized finance (DeFi) industry, offering tools for quantitative analysis, risk management, and research. Its three-tier structure, consisting of Data Stream (DS), Quantitative Layer (QL), and Integrated Environment (IE), enables efficient quantitative analysis and management of the DeFi market. Leveraging cutting-edge technologies for real-time processing of large datasets, QIS ensures accurate analyses, thereby supporting strategic decision-making. It also emphasizes customization, allowing users to develop strategies tailored to their objectives and risk profiles, making QIS ideal for optimizing returns and minimizing risks in the DeFi ecosystem.

---