We study the profitability of a hypothetical delta-neutral Dynamic E-CLP that combines an LP position in ETH/USDC with a hedge of the ETH exposure. The result is a simple measure of LP profitability: real yield achievable on USD after hedging costs.

TLDR:

• The hedged LP yielded 134% APY over the backtest (2024–12–01 to 2025–08–01)

• The hedged LP generated positive profit in ~73% of hours in the analysis (~87% of 30-day rolling periods)

• The backtest is fairly conservative and based on the real yields of E-CLPs that have existed over this period

Motivation

The profitability of LPing is complicated to understand with measures like ‘markouts’ (a measure of market making profitability) and ‘impermanent loss’ (a historical measure of LP profitability). LPs have lacked a simpler measure to understand LP profitability: how much an LP position makes in USD terms after hedging price exposure. From this perspective, an LP position then clearly either makes money (earns more in trading fees than it loses due to arbitrage and price hedging) or doesn’t.

Prior analysis has shown how Gyroscope’s pools have excelled as measured by markouts (see here and here). This analysis complements markout profitability from the further perspective of hedged LP profitability.

Gyroscope’s Dynamic CLPs

Gyroscope’s Dynamic CLPs are non-custodial liquidity pools that auto-adjust with the market. Dynamic CLPs solve an issue faced by many LPs: it is difficult to manage positions well. The average LP often ends up out of range (earning nothing) or too concentrated, dragging down performance. Dynamic CLPs auto-adapt for LPs, designed with strategies designed to optimize passive LPing. Read more about the technology in the Gyroscope docs.

Setup

The study measures the profitability of a hypothetical dynamic WETH/USDC E-CLP where the LPer invests $1M into the E-CLP and attempts to fully hedge the ETH exposure in each time step by going short in the perp market (delta hedging) on an hourly basis. We backtest based on the historical time frame from 2024–12–01 to 2025–08–01, where simulated volume (for fees) is based on the real performance of three static ETH/USD{C,T} E-CLPs that were live during that time period.

Dynamic E-CLPs have been live since June 2025 but follow the same general mechanism as the preceding static E-CLPs.

Note that this only results in a single trajectory; no attempt is made to model a probability distribution of returns / VaR / etc, though we can get something similar by considering different entry/exit times.

1 Resulting PnL

The following chart shows

• Top: The ETH price and the price range [alpha, beta] of the ECLP, and the times where the dynamic ECLP updates its price range (vertical lines). Recall that the dynamic ECLP updates its price at discrete intervals, whenever it goes out of range vs. the global market price. The pool then shifts its price range such that the edge is just the current market price.

• Middle: the value of the LPer’s pool position without hedging

• Bottom: The total cumulative PnL of the LPer’s whole strategy with hedging (bottom).

Over the 8-month time frame, the LPer made a profit of $893k, corresponding to a yield of 89.3% over 8 months or 134.1% APY.

Note that there is a period of persistently small negative PnL between 2025–03 and 2025–05. This is the result of imperfect hedging together with relatively low volumes and ongoing costs for perp trading (see below).

PnL Components

The following chart shows the individual components of cumulative PnL (top). For easier readability, we also show the pool TVL position + the perp position together and other costs separately (bottom).

We can see that:

• Perp funding rates have a negligible effect in either direction. They are not zero and the pool is actually overexposed to them (because we tend to go more short when prices fall and funding rates tend to be negative) but they are by and large so close to 0 that they do not matter in the grand scheme of things, and they do cancel out over time.

• Perp trading creates a moderate, steady cost us, but the yield of the strategy itself is much larger.

• Even after hedging the pool balances, PnL of (pool TVL + perp) can be negative at times. This is a result of imperfect hedging: the perp position only hedges the (delta) exposure to the assets in the pool at the start of each period (hour) but not the (gamma) exposure resulting from the pool acquiring/selling ETH as prices move. This imperfection is always to the pool’s detriment: when prices drop, the pool acquires more ETH during the period and the short position makes insufficient profit; when they increase, the pool sells off ETH and the gain from ETH is too small to cover the losses in the short position.

2 Distribution of returns over time

The simulation style results in a single trajectory but we get get some quasi-distributional results by considering an LPer who joins/exits the strategy at different points in time. The following table shows two variants of this. The left-hand column shows different APY values. The middle column shows the percentage of hours where the strategy made at least that equivalent APY. The right-hand column shows the share of 30-day rolling (overlapping) periods where an LPer joining at the beginning and exiting the strategy at the end would have made at least that equivalent APY.

For example, the strategy is weakly profitable (APY ≥ 0) 73.25% of the hours and 87.79% of the rolling 30-day windows. The difference between the two numbers means that, roughly speaking, losses tend to cancel out with profits after some while.

We can also plot the 30-day rolling APYs over time. Note that the time window is looking back, so the first 30 days of the period do not have any data.

3 Methodology

The simulation assumes that the LPer acts as follows in each time period, where a time period is one hour.

• The LPer first swaps some USDC to ETH and enters a perp short position to hedge the ETH exposure.

• In each time step, the pool makes some organic (uninformed) volume and some arbitrage volume, earning fees and losing money to adverse selection. If it is out of range, the update-to-edge procedure is executed to bring it back into range. At the end of the period, the LPer adjusts their perp position to match the new ETH exposure of the pool. They pay a cost for trading in the perp market and they pay/earn the funding rate.

• In the final time step, the trader closes their perp position.

• Note that the hedge is not perfect because the pool changes its composition across each time period (i.e., we hedge delta, not gamma). We study this below. Gamma hedging would be more complex and would require derivatives that are not easily available. Shorter periods would help here (but increase trading cost and obviously operational cost).

Organic volume is inferred as follows: we consider the three static ECLPs that have been live with nontrivial TVL since 2024–12. For each hour and each pool that is in range, we compute the pool’s liquidity provided measured by liquidity density. This is a measure for liquidity independent of the pool type and configuration. We compute the theoretical arbitrage volume, infer organic volume, and consider average organic volume through these pools relative to liquidity density. To estimate organic volume through the hypothetical dynamic ECLP, we multiply the value from the previous step by the liquidity density of the dynamic ECLP in each time step. This approach is sound if we assume that organic volume is linear in liquidity provided at market price in each time step (while trading demand can vary across time steps), which is a major assumption. It also ignores a couple details (e.g., arbitrage volume within the period as a result of organic trading). The main benefit of the approach is that we do not need a model of trading demand over time or demand through ECLPs specifically, and we avoid a modeling artifact that could otherwise dominate the result.

There is a period of about 3 weeks between Apr and May 2025 where none of the ECLP was in range and had significant balances. We fix that longer period by just assuming the average volume / LD for the 3 weeks before. All other (briefer) such periods are forward-filled from the last observation.

The simulation does not account for cost of capital to margin perps, price impact in the perp market, or perp/spot divergence, all of which would likely be small unless the deployed capital is very large. We also do not study optimal rebalancing intervals or partial delta hedging.

Data Sources

We use:

• Hourly spot prices from Binance.

• Funding rates from Binance at their actual update intervals (every 8 hours).

• Trade amounts and pool balances from three static WETH/USDC and WETH/USDT ECLPs at different price ranges

Assumptions

We assume:

• A dynamic ECLP configured to +/- 10% around its mid price with an amplification of lambda=4. This corresponds to dynamic ECLPs we have deployed in the meantime.

• NB: A wider range or less amplification would reduce gamma exposure and thus the cost from imperfect hedging and perp trading costs, but it would also reduce liquidity provided and thus revenue from fees.

• Swap fee = 30bp. This corresponds to most static and dynamic ECLPs we have deployed in the past.

• Initial capital = $1M

• Perp trading cost = 6bp on the USD value. This corresponds to 5bp Binance trading fee + 1bp spread.

What to do next

• Follow Gyroscope on Twitter/X @GyroStable.

• Join the Gyroscope Discord.

• Explore the live Dynamic CLPs in the Gyro app.

About Gyroscope

Gyroscope is a non-custodial liquidity protocol that combines efficient passive concentrated liquidity and stablecoin settlement infrastructure. The initial development team of the Gyroscope protocol is FTL Labs, including the authors of this article (Ariah Klages-Mundt and Steffen Schuldenzucker, PhDs), with backing led by Placeholder and Galaxy. FTL Labs members have authored seminal papers on stablecoin design and DeFi risk. For more information, please visit https://gyro.finance/ or review the documentation.

Disclaimer.

The information provided in this communication is for informational purposes only and should not be considered financial advice. It does not constitute an offer to sell, a solicitation of an offer to buy, or a recommendation to purchase any securities or investment products. All investments involve risks, including the potential loss of principal. Past performance is not indicative of future results. The projections and estimates presented herein are based on assumptions and subject to change without notice. This communication has been prepared based upon information, including market prices, data and other information, from sources believed to be reliable, but such information has not independently been verified and this communication makes no representations about the enduring accuracy of the information or its appropriateness for a given situation.

LP returns are variable and depend on many factors, including volumes, asset price changes, and hedging performance.

Users should conduct thorough research and consult with a financial, legal and technical advisor before interacting with Gyroscope pools. It is the responsibility of each participant to understand the risks of Gyroscope pools and to comply with all applicable laws and regulations in their jurisdiction. See the Terms of Service for more information.

FTL Labs