Evaluating Performance of Pendle Liquidity Pools (Part 2)
Summary
- We have performed a study on the performance of the following Pendle pools against the underlying pool: PT-stETH-26DEC24/SY-stETH (Ethereum), PT-GLP-28MAR24/SY-GLP (Arbitrum), and PT-gDAI-28MAR24/SY-gDAI (Arbitrum)
- We found that providing liquidity on Pendle is superior to depositing the asset in the underlying pool in most cases even when excluding PENDLE incentives
- In the worst case, the IL of a Pendle LP compared to the underlying pool was 0.85%
- The performance of an LP using zero price impact mode or otherwise depends on the underlying and implied APY of the pool
Introduction
In the previous study, we evaluated the performance of liquidity providers on Pendle and concluded that Pendle LPs indeed outperform holding the underlying asset without earning yield. However, we have not compared the performance of Pendle LPs against users who earn yield via depositing into the underlying pool. Are Pendle liquidity providers really subject to minimal impermanent loss (IL) as is often claimed? The IL on Pendle pools will be the focus of this study.
Evaluating IL
Traditionally, IL refers to the case when the value of a liquidity provider’s deposited assets in a liquidity pool is lower than the value of those assets if they did not provide liquidity. Due to the nature of Pendle’s PT/SY pool exposure, this definition is not useful because not many will hold SY and PT separately (we will still however perform this study in an appendix). For this study, we shall define IL as PT/SY LP underperformance against holding assets in the underlying pool. For example, this can be a comparison of a Pendle stETH LP against a naked stETH holder.
We have identified 2 possible cases for comparison.
Case A: Providing liquidity on Pendle vs depositing in the underlying pool
Case B: Providing liquidity on Pendle with zero price impact mode vs depositing in the underlying pool
For each case, we also evaluate the impact of PENDLE incentives in these pools, where we either include or exclude PENDLE incentives for PT/SY LPs.
Definition
Intuitively, we compare the yield of an LP on Pendle against a yield farmer who deposits into the underlying pool without interacting with Pendle. For example, we compare the performance of a Pendle GLP pool LP against a GLP holder who does not interact with other protocols.
Much like the LP performance in the previous study, the specifics of each Pendle liquidity position are defined below:
For each case, we shall define an in-pool value and an out-pool value that represent the value of the assets in the LP position (including all yield earned) and the value of the assets deposited in the underlying pool (including all yield earned).
The resulting IL is defined as:
IL = InpoolValue / OutpoolValue
An IL value above 1 indicates the outperformance of using Pendle, whereas an IL value below 1 indicates the underperformance of the Pendle position.
Case studies
For the following case studies, we shall evaluate the same three pools on Pendle as the previous study, namely PT-stETH-26DEC24/SY-stETH (Ethereum), PT-GLP-28MAR24/SY-GLP (Arbitrum), and PT-gDAI-28MAR24/SY-gDAI (Arbitrum).
For each case study, please refer to the data on the following spreadsheets:
- PT-stETH-26DEC24/SY-stETH (Ethereum)
- PT-GLP-28MAR24/SY-GLP (Arbitrum)
- PT-gDAI-28MAR24/SY-gDAI (Arbitrum)
For each sheet, let the date on the y-axis d1 be the date of providing liquidity and the date on the x-axis d2 be the date of withdrawing liquidity. The number on row d2 and column d1 thus corresponds to the IL value defined previously.
Each sheet with the number 1 (e.g. sheet A1) refers to the study without including PENDLE incentives, whereas each sheet with the number 2 (e.g. sheet A2) refers to the study with PENDLE incentives.
Case A: Providing liquidity on Pendle vs depositing in the underlying pool
We define IL for providing liquidity on Pendle (e.g. the Pendle stETH pool) against depositing in the underlying pool alone (e.g. holding naked stETH) as below.
In-pool Value
Note that the PENDLE rewards can be excluded if we do not wish to take these incentives into account.
Out-pool Value
As defined, the IL value is the ratio of the in-pool value to the out-pool value. We shall evaluate the cases of each pool separately.
PT-stETH-26DEC24/SY-stETH (Ethereum)
For the case of SY stETH, yield is compounded into the position itself, where 1 SY stETH (equivalent to 1 wstETH) is worth more stETH as the pool accrues yield from the ETH staking position on Lido.
Sheet A1 shows that stETH liquidity providers on Pendle outperformed naked stETH holders (stETH holders who did not provide liquidity on Pendle) on several cases even without PENDLE incentives. The figure below illustrates the performance of Pendle LPs on certain timeframes.
In the diagram above, each line represents the date when a user provided liquidity, while the dates on the x-axis represent the time when this user withdraws liquidity from the pool. IL appears to be the worst for Pendle LPs who exited in mid-June 2023. Between 24 April 2023 and 21 July 2023, the worst IL was 0.30%, which was the case when a holder provided stETH liquidity on Pendle on 24 April 2023 and withdrew on 23 June 2023.
The inclusion of PENDLE incentives as shown in Sheet A2 led to all Pendle LPs outperforming naked stETH positions. The figure below illustrates the performance of Pendle LPs on certain timeframes.
In all cases, IL has been eliminated with the inclusion of PENDLE incentives.
Generally, Pendle stETH LPs have outperformed naked stETH holders.
PT-GLP-28MAR24/SY-GLP (Arbitrum)
For the case of GLP, yield is only issued in the form of WETH rewards. Thus, the SY index remains constant at all times, but we need to account for the WETH rewards when calculating the yield.
Sheet A1 shows that in most cases, GLP liquidity providers on Pendle outperformed naked GLP holders even without PENDLE incentives. The figure below illustrates the performance of Pendle LPs on certain timeframes.
IL appears to be the worst for Pendle LPs who exited in mid-June 2023. The worst IL was 0.78%, which was the case when a holder provided GLP liquidity on Pendle on 15 June 2023 and withdrew on 23 June 2023.
The inclusion of PENDLE incentives as shown in Sheet A2 led to an improvement in performance for LPs on Pendle. The figure below illustrates the performance of Pendle LPs on certain timeframes.
We can observe a significant improvement in IL in many cases. However, the worst IL remains around 0.78%, also when a holder provided GLP liquidity on Pendle from 15 June 2023 to 23 June 2023.
Generally, Pendle GLP LPs have outperformed naked GLP holders.
PT-gDAI-28MAR24/SY-gDAI (Arbitrum)
For the case of gDAI, yield is compounded into the position itself, where 1 gDAI is worth more DAI as the pool accrues fees from traders on gTrade.
Sheet A1 shows that in most cases, gDAI liquidity providers on Pendle outperformed naked gDAI holders even without PENDLE incentives. The figure below illustrates the performance of Pendle LPs on certain timeframes.
IL appears to be the worst for Pendle users who deposited in late May 2023 and withdrew in early June 2023, along with those who deposited in late June and withdrew in mid-July. The worst IL was 0.10%, which was the case when a holder provided gDAI liquidity on Pendle on 30 May 2023 and withdrew on 3 June 2023.
The inclusion of PENDLE incentives as shown in Sheet A2 led to all Pendle LPs outperforming naked gDAI positions. The figure below illustrates the performance of Pendle LPs on certain timeframes.
In all cases, IL has been eliminated with the inclusion of PENDLE incentives.
Generally, Pendle gDAI LPs have outperformed naked gDAI holders.
Case B: Providing liquidity on Pendle with zero price impact mode vs depositing in the underlying pool
We define IL for providing liquidity on Pendle (e.g. the Pendle stETH pool) against depositing in the underlying pool alone (e.g. holding naked stETH) using zero price impact mode on the Pendle Trade UI as below.
In-pool Value
Note that the PENDLE rewards can be excluded if we do not wish to take these incentives into account.
Out-pool Value
We shall evaluate the cases of each pool separately.
PT-stETH-26DEC24/SY-stETH (Ethereum)
Sheet B1 shows that there are several cases where Pendle LPs (using zero price impact mode) underperform naked stETH holders if we exclude PENDLE incentives. The figure below illustrates the performance of Pendle LPs on certain timeframes.
The worst IL was 0.03%, which was the case when a user provided stETH liquidity on Pendle on 24 April 2023 and withdrew on 23 June 2023.
The inclusion of PENDLE incentives as shown in Sheet B2 led to all Pendle LPs outperforming naked stETH positions. The figure below illustrates the performance of Pendle LPs on certain timeframes including PENDLE incentives.
In all cases, IL has been eliminated with the inclusion of PENDLE incentives.
When comparing the performance of providing stETH liquidity on Pendle via zero price impact mode against the case without using zero price impact mode, IL using zero price impact mode is lower. The figure below illustrates the historical implied and underlying APYs for the Pendle pool.
In the figure above, the green line represents the underlying yield for stETH, whereas the blue line represents the implied yield on PT stETH. The underlying APY for stETH has been higher than the implied APY until 18 June 2023, when this situation reverses.
stETH LPs depositing into Pendle using zero price impact mode significantly before 18 June 2023 have earned significantly higher yields as a result of their YT exposure, causing outperformance compared to the LPs that did not use zero price impact mode. After 18 June 2023, this was no longer the case.
The reasons for this outcome are:
- Providing liquidity on Pendle decreases the exposure of the LP to the yield of the underlying pool.
- The lack of the YT exposure of an LP without using zero price impact mode causes it to have less exposure to the underlying yield compared to an LP using zero price impact mode.
- When the underlying yield is high, the YT exposure of the LP using zero price impact mode results in superior performance.
On average, LPs generally benefited from zero price impact mode.
Regardless, Pendle stETH LPs have generally outperformed naked stETH holders.
PT-GLP-28MAR24/SY-GLP (Arbitrum)
Sheet B1 shows that there are several cases where Pendle LPs (using zero price impact mode) underperform naked GLP holders if we exclude PENDLE incentives. The figure below illustrates the performance of Pendle LPs on certain timeframes.
We observe that there is a brief period of IL incurred for users who provided liquidity on Pendle between 18 May 2023 and 11 June 2023. If we exclude PENDLE incentives, the worst IL was 0.85%, which was the case when a user provided GLP liquidity on Pendle on 24 April 2023 and withdrew on 13 July 2023.
The inclusion of PENDLE incentives as shown in Sheet B2 led to an improvement in performance for LPs on Pendle. The figure below illustrates the performance of Pendle LPs on certain timeframes including PENDLE incentives.
In most cases, Pendle LPs have outperformed naked GLP holders if we include PENDLE incentives. The worst IL was 0.84%, which was the case when a user provided GLP liquidity on Pendle on 24 April 2023 and withdrew on 13 July 2023.
Unlike stETH, when comparing the performance of providing GLP liquidity on Pendle via zero price impact mode against the case without using zero price impact mode, we arrive at an interesting observation: the underperformance against naked GLP is worse when using zero price impact mode. We explain this by studying the historical implied APY and underlying APY in the figure below.
Since the underlying yield has underperformed the implied yield on Pendle most of the time besides the spike in mid-June 2023, the zero price impact mode LP on Pendle generally underperforms the non-zero price impact mode LP due to its YT exposure.
This trend noticeably reversed in mid-June during the spike in GLP yield, where we observe non-zero price impact mode LPs underperforming zero price impact mode LPs.
Regardless, Pendle GLP LPs using zero price impact mode have generally outperformed naked GLP holders.
PT-gDAI-28MAR24/SY-gDAI (Arbitrum)
Sheet B1 shows that there are several cases where Pendle LPs (using zero price impact mode) underperform naked gDAI holders if we exclude PENDLE incentives. The figure below illustrates the performance of Pendle LPs on certain timeframes.
We observe that the worst IL seems to be incurred between late April 2023 to late June 2023. The data indicates that the worst IL was 0.03%, which was the case when a user provided gDAI liquidity on Pendle on 3 May 2023 and withdrew on 24 June 2023.
The inclusion of PENDLE incentives as shown in Sheet B2 led to all Pendle LPs outperforming naked gDAI positions. The figure below illustrates the performance of Pendle LPs on certain timeframes including PENDLE incentives.
In all cases, IL has been eliminated with the inclusion of PENDLE incentives.
Much like stETH, when comparing the performance of providing gDAI liquidity on Pendle via zero price impact mode against the case without using zero price impact mode, IL using zero price impact mode is lower. We once again turn to the historical implied and underlying APYs for the Pendle pool in the figure below.
Unlike GLP, gDAI’s underlying APY does not show consistent underperformance against the implied APY on Pendle. In such a case, we observe that zero price impact mode LPs do outperform LPs without using zero impact mode.
Regardless, Pendle gDAI LPs have generally outperformed naked gDAI holders.
Conclusion
From our study of the three pools on Pendle above, we can conclude that:
- LPs on Pendle generally outperform those who do not use Pendle, even when excluding PENDLE incentives.
- In the worst case, IL on Pendle pools is below 1% (the worst case was observed to be 0.85%) even when excluding PENDLE incentives.
- When including PENDLE incentives, IL is no longer present in the stETH and gDAI pools.
- The performance of an LP using zero price impact mode or otherwise depends on the underlying and implied APY of the pool.
These results do not factor in the impact of vePENDLE holdings for the LP, which may lead to further outperformance of Pendle liquidity providers against the underlying pool. The impact of vePENDLE holdings is a subject that we may perform a further study on.
Appendix
In our case studies above, we have defined IL as PT/SY LP underperformance against holding assets in the underlying pool. However, it is also possible to define IL in the classic terms where we compare PT/SY LP underperformance against holding PT and SY separately without providing liquidity in Pendle. This section performs this theoretical study despite the unlikely scenario that anyone would choose to hold PT and SY separately.
In-pool Value
The in-pool value can be calculated with the same formula in Case A mentioned above.
Note that the PENDLE rewards can be excluded if we do not wish to take these incentives into account.
Out-pool Value
PT-stETH-26DEC24/SY-stETH (Ethereum)
Sheet Appendix 1 shows that in many cases, stETH liquidity providers on Pendle underperformed those who held PT and SY gDAI separately. The figure below illustrates the performance of Pendle LPs on certain timeframes.
The worst IL was 0.03%, which was the case when a user provided stETH liquidity on Pendle on 24 April 2023 and withdrew on 23 June 2023.
The inclusion of PENDLE incentives as shown in Sheet Appendix 2 led to an improvement in performance for LPs on Pendle. The figure below illustrates the performance of Pendle LPs on certain timeframes including PENDLE incentives.
In all cases, IL has been eliminated with the inclusion of PENDLE incentives.
PT-GLP-28MAR24/SY-GLP (Arbitrum)
Sheet Appendix 1 shows that in most cases, GLP liquidity providers on Pendle underperformed those who held PT and SY GLP separately. The figure below illustrates the performance of Pendle LPs on certain timeframes.
The worst IL was 0.88%, which was the case when a user provided GLP liquidity on Pendle on 24 April 2023 and withdrew on 13 July 2023.
The inclusion of PENDLE incentives as shown in Sheet Appendix 2 led to an improvement in performance for LPs on Pendle. The figure below illustrates the performance of Pendle LPs on certain timeframes including PENDLE incentives.
We can observe a noticeable improvement in IL in many cases. However, the worst IL remains around 0.87%, also when a holder provided GLP liquidity on Pendle from 24 April 2023 to 13 July 2023.
PT-gDAI-28MAR24/SY-gDAI (Arbitrum)
Sheet Appendix 1 shows that in many cases, gDAI liquidity providers on Pendle underperformed those who held PT and SY gDAI separately. The figure below illustrates the performance of Pendle LPs on certain timeframes.
The worst IL was 0.03%, which was the case when a user provided gDAI liquidity on Pendle on 24 April 2023 and withdrew on 19 June 2023.
The inclusion of PENDLE incentives as shown in Sheet Appendix 2 led to an improvement in performance for LPs on Pendle. The figure below illustrates the performance of Pendle LPs on certain timeframes including PENDLE incentives.
In all cases, IL has been eliminated with the inclusion of PENDLE incentives.
We can note that even when applying the classic definition of IL on PT/SY LP underperformance against holding PT and SY separately, the IL remains below 1%.
