Find Smart Money from eth-usdc-005 LP

Introduction

In last year’s annual report, we conducted an in-depth analysis of the behavior strategies and returns of the Uniswap V3 asset management product (fund), presenting the state of the liquidity-providing market. This year, in addition to continuing the analysis of the fund’s performance, we will also screen high-net-worth and long-term liquidity providers to research on their behavior and return performance.

This post will provide statistics and categorization of addresses actively participating in Uniswap V3 liquidity provision in the first half of 2023. By applying the CAPM model, we will analyze the relationship between risk and return for these providers. Additionally, we will continue to track the returns of the fund from last year to ensure that these insights are fully considered when making optimal market-making investment decisions.

You can view our code and data on GitHub. We welcome discussion. Feel free to contact: zelos@antalpha.com or make an issue in our git.

Stay tuned to Zelos, and we will continue tracking Defi products’ returns in the 2023 second half year.

Data

By our open source tool, Demeter, we conducted a detailed statistical analysis of the Uniswap liquidity pool “eth-usdc-0.05” (0x45dda9cb7c25131df268515131f647d726f50608) on the Matic network. The analysis covered a range of data from the pool’s creation date (December 20, 2021) to June 20, 2023, including key metrics such as yield and volatility. This analysis ultimately encompassed a total of 15,769 liquidity provider addresses, along with their corresponding historical net asset values and returns. In the future, another post will cover engineering and technical aspects, explaining the calculation process and optimizing algorithms by identifying common computational components for comprehensive data statistics.

In the first half of 2023, we calculated two key metrics for liquidity providers: Holding Time Percentage and Max Net Value. Through our analysis, we excluded addresses with relatively small maximum net values (within the 0–0.5% range) and also removed liquidity providers with short holding times (within the 0–10% range of holding time percentage). This resulted in a sample of 223 high net worth and long-term participating liquidity providers for further analysis of their behavior performance. We will conduct a detailed analysis of these providers to evaluate their behavior performance.

Market Performance

The market returns per hour were calculated by weighting the hourly returns of the 223 participants with their respective net asset values. It can be observed that around March 11th, there was significant volatility in both the market and ETH returns, indicating a period of turbulence.

Selected Address Performance

For high net worth and long-term participating users in Uniswap liquidity provision, there is no significant correlation between their annualized returns and net values, with a correlation coefficient of -0.03401. The average annualized return for these users is 13.46%.

By analyzing the distribution of the holding time percentage, we categorized these 223 users into four groups with thresholds of 31%, 60%, and 80%.

Mirroring last year’s report, we continue using the ETH price as our benchmark. In the first half of this year, ETH achieved a cumulative semi-annual return of 9.47%, cumulating an annualized return of 24.81%, amid a volatility of 40.99%. By plotting the net asset values based on Ethereum (ETH) addresses’ inception dates, we’ve crafted a net asset trend chart. This graphic portrays asset dynamics among the top 20 addresses, ranked by net asset value. Curve colors align with previous categorization by holding duration. Notably, a majority of higher net asset value addresses failed to surpass ETH’s price returns, particularly those with greater assets and longer holding spans. Conversely, smaller net asset value addresses demonstrated relatively stronger performance.

Funds performance

The high-yield EOA addresses that were identified last year did not persist in offering liquidity throughout 2023. The subsequent graph portrays the net asset value trends of the five aforementioned funds spanning from last year to the present. Amid abrupt downturns in ETH prices, funds capable of swift response retain a competitive edge in performance. For instance, arrikis and dHEDGE funds adeptly adjusted their market-making strategies during the March 11 ETH price drop, while Unipilot encountered more pronounced losses.

Find Alpha

Assumption

We employ the CAPM model to assess the performance and risk sensitivity of liquidity providers’ market-making strategies relative to the overall market. By calculating the alpha value (excess returns), we identify liquidity providers who surpass market performance.

The market’s performance is depicted by the weighted average return rate of liquidity providers (LPs). To reflect the varying influence of different LPs, proportional to their final net asset values, we employ the maximum net asset value among LPs as the weight for calculating the market return rate (Rm).

The risk-free rate (R0) is assumed to be zero. While one could consider borrowing rates from platforms like Aave as a risk-free rate, or even account for the yield on beth, disparities in data and statistical standards present challenges. These discrepancies encompass factors such as loans and stake. Given these complexities, our present approach leans towards setting the risk-free rate at 0.

CAPM

Within each hour, we calculate the return rate for each address as well as the market and ETH return rates. We employ linear regression to estimate the alpha value for each LP:

In this context, the alpha value signifies the surplus return of the LP in contrast to the market, while the beta value quantifies the responsiveness of the LP’s returns to market fluctuations.

We also analyzed the alpha and beta performance of different addresses when considering ETH returns as the market return rate. The following two scatter plots illustrate the relationship between alpha and beta values under ETH and market return rates, with larger points representing clustered points from different holding time groups.

By observing these clustered points, we can draw the following conclusions:

1. LP addresses with longer holding times (80–100%) exhibit the lowest market volatility sensitivity. The excess returns of these addresses are mostly positive.
2. On the other hand, LP addresses with shorter holding times (0–31%) show higher sensitivity to market volatility.

Sharp

We observe the “risk-reward” situation of these LP addresses through the Sharpe ratio. The slope of the scatter plot represents the excess returns obtained per unit of risk. The larger points in the scatter plot represent the clustered points from each grouping.

By examining the clustered points from different holding time groups, we can observe that LP addresses with the longest holding time (80–100%) have the highest Sharpe ratio: 0.9586. The LP addresses with the shortest holding time (0–31%) have the next highest Sharpe ratio: 0.4271. The Sharpe ratio for the holding time range of 31–60% is 0.0829, and for the range of 60–80%, it is -0.2030.

These Sharpe ratios indicate the risk-adjusted performance of the LP addresses. A higher Sharpe ratio suggests a better risk-reward trade-off, while a lower or negative Sharpe ratio indicates a relatively poorer risk-adjusted performance.

Based on these findings, we will focus on observing LP addresses that have high alpha values under the condition of longer holding times, as well as LP addresses that have high alpha values and a positive correlation between “risk-reward” under the condition of shorter holding times.

Find Strategy

To analyze the actual strategy of the fund, we analyzed EVENT data. The following graph depicts the LPs’ positions on ETH prices, where a narrow red range indicates a short holding time or a small price range. The gray line represents the 95% confidence interval of ETH prices. We also plotted the cumulative returns based on the fund’s net asset value and the seven-day rolling annualized volatility of ETH.

Funds

Compared to last year, we selected four funds that have been consistently market-making in the first half of this year. It can be observed that the Unipilot fund has the narrowest market-making price range. During periods of high volatility and ETH price declines, this fund chose to adjust its market-making range, resulting in significant losses. Other funds either opted for longer holding times or wider price ranges, which allowed them to achieve relatively good returns. Among them, Popsicle had the highest excess returns.

Large Alpha (Participating exceeding 80%)

LPs with high participation rates tend to engage deeply in the market, and some of these addresses exhibit high alpha values. These addresses often hold multiple market-making positions simultaneously. Another notable characteristic is that positions are held for longer periods, with relatively wider price ranges. These LPs typically adjust their positions dynamically based on volatility. They generate profits by collecting long-term fees.

Large Alpha (Participating rate between 0~31%)

When we shift our focus to addresses with participation rates below 31%, there is a clear emphasis on capital efficiency, leading to a preference for narrow market-making ranges. These addresses frequently adjust their positions, actively taking on short-term risks. The other strategy is to pay close attention to timing and withdraw from the liquidity market when volatility changes. Both of these strategies can be observed among makers with low participation rates.

Conclusion

We analyze the behavior and returns of liquidity providers (LPs) in Uniswap V3, focusing on high net worth and long-term participants. By utilizing the CAPM model, their relative performance against the market is evaluated, and the risk-reward trade-offs of different holding times are analyzed. LP addresses with longer holding times exhibit lower market risk and higher excess returns per unit of risk, making them attractive targets for tracking.

Furthermore, we examine the actual strategies employed by funds and addresses with significant excess returns. Due to varying levels of participation, LPs adopt different market-making strategies. This analysis provides insights into the diverse approaches taken by LPs based on their level of involvement.

Disclaimer

This is a working paper representing research in progress. The report is the production of a professional study, and its contents are intended to be informational only. This article is not and should not be considered as providing investment advice. No representation or warranty is made herein regarding the information’s fairness, accuracy, reasonableness, or completeness.

Interesting Address

In addition, we have also identified a market participant from the perspective of micro-market structure. This participant engages in market-making for a very short duration of only 1 minute and 30 seconds, constructing and withdrawing 14 positions.