Research on Overlapping Multi-Position LP Returns
Overview
Our study aims to investigate the strategy of liquidity providers (LPs) on Uniswap known as the “overlap long position” strategy. This strategy involves LPs holding one or more market-making positions with different time spans in a transaction pair pool simultaneously. The intent behind employing this strategy is to achieve risk diversification, maximize returns, and gain flexibility, thereby potentially reducing financial losses stemming from position price fluctuations.
We have computed the percentage of time that LPs overlap their multiple positions within their total market-making duration and assessed the impact of the overlap long position strategy on LPs’ returns. You can access our codes here.
TLDR:
- The majority of LPs do not employ the overlapping long positions strategy
- Among LPs operating on the ETH chain with substantial funds and trading volume, those utilizing the overlapping long positions strategy demonstrate a higher average return.
- The return risk of the overlapping long positions strategy is similar to the risk of the ordinary strategy.
Date Description
Data Collection
Our statistical data comes from Uniswap’s usdc-eth-005 and usdc-weth-005 liquidity pools on the ETH and Polygon networks. The time range of the data is from January 1, 2023 to December 11, 2023。
By grabbing tick data, we get LP’s market-making behavior and net asset value situation, You are welcome to find our code of data processing at：https://github.com/zelos-alpha/uniswap-annual-review-2023. As well as our data processing article: Uniswap Data Processing of zelos research
Percentage of Overlapping Position Time
We calculated the total time of overlapping LP positions as a percentage of its total market participation time. This concept is illustrated in the diagram below. In the figure, each bar represents an LP position, where type1 indicates employment of the overlapping position strategy, while type2 represents independent position strategies. The percentage of overlapping position time is labeled as the ratio of the blue time period to the red time period in the figure (for cases involving three or more overlapping positions, we treated them as a single overlapping time period during the calculation process).
Overlapping Position LP Group and Return Rate
We utilized Demeter to compute the return rates of LP addresses. In order to investigate potential significant differences in returns among LPs with varying percentages of overlapping participation time, we employed one-way analysis of variance. This method aims to determine the meaningfulness of categorizing LPs based on the percentage of overlapping position time for return analysis. If the grouping is meaningful, we intend to identify the grouping method that yields the largest mean difference in returns.
We examined the return performance of these distinct groups, calculating metrics such as maximum net value, return risk, and the number of overlapping positions, to delve into the influence of LPs’ specific overlapping position strategies on returns.
Analysis
Distribution of LPs based on Overlapping Positions
In our analysis of the time-frequency distribution of overlapping positions, we observed that on ETH, 87.17% of LPs employ a completely non-overlapping position strategy, while on Polygon, 84.04% of LPs do the same.
The predominant trend among LPs is the adoption of a non-overlapping position strategy.
LP Grouping based on Overlapping Positions
We computed the mean return rates separately for LPs with overlapping and non-overlapping positions. On ETH, LPs utilizing an overlapping position strategy demonstrated a higher average return compared to those with a non-overlapping strategy.
We employed the one-way variance method to analyze the overlapping and non-overlapping groups, calculating the F-statistic and P-value. The F-statistic measures variance differences between groups, while the P-value gauges the likelihood of the observed F-statistic value being due to random factors. The P-value for ETH fell well below the standard significance level, whereas the F-statistic value for Polygon was too low and the P-value too high.
In summary, we observed significant return disparities between overlapping and non-overlapping groups on ETH, characterized by large funds and trading volumes. However, such disparities were not evident on Polygon, which is associated with smaller funds and frequent operations. Our focus now shifts to a more detailed examination of LPs’ overlapping position strategies on ETH.
Moving forward, our next step involves refining the groupings to scrutinize the specific impact of overlapping position duration percentages on returns.
When a third group classification is introduced, the above figure illustrates the influence of the selected percentage on group disparities. We identified the percentage value corresponding to the maximum F-statistic (14.75291, 4.07792e-07) as 87%.
Upon analyzing the overlapping time groups and their respective mean returns, it becomes evident that LPs employing an overlapping position strategy achieve higher returns, with the return increasing as the proportion of overlapping positions relative to all positions rises (1.10484→1.15721→1.24339).
Analysis of Factors Influencing Returns
In this section, we will examine the combined impact of the proportion of time in overlapped positions, the maximum net value, return risk, and the number of overlapped positions on returns. Firstly, a high maximum net value may afford LPs greater funds and flexibility to initiate new position combinations or adjust existing positions. Secondly, a higher proportion of time in overlapped positions and a greater number of overlapped positions may contribute to reducing the overall volatility of the investment portfolio.
Based on the scatter plot above, LPs employing a higher proportion of overlapped position strategies tend to hold greater net assets. Furthermore, for these LPs, a positive correlation exists between their net asset holdings and their return rates.
Additionally, for users with multiple overlapped positions, a similar positive relationship is observed between net asset value, the number of overlapping positions, and the returns.
In terms of return risk, it broadly adheres to the principle that higher risk is associated with higher returns. In comparison to the non-overlapping position strategy, the overlapping position strategy exhibits smaller variance in return risk, with marginal differences in the mean return risk.
Conclusion
We found the situation of liquidity providers on Uniswap adopting overlapped long position strategies and the impact of this strategy on their returns.
- Most LPs do not adopt the overlapped position strategy, and LPs adopting this strategy on the high-volume ETH chain achieve higher return rates;
- The higher the proportion of overlapping positions of LP, the higher the return rate;
- LPs adopting overlapped long position strategies have similar return risks to ordinary strategies.
There are many future research directions to consider, such as combining the ETH price period with the long position period to explore when the long position strategy yields high returns; in addition to polygon and ETH, what impact do different trading pairs or market conditions in different time periods have on this strategy.
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.
