Table of Contents
Do not index
Introduction
Static fees work in calm markets, but break down as volatility rises. Algebra’s Dynamic Fee adapts fees to market volatility—keeping fees low to attract organic flow and raising them when arbitrage risk increases. In simulations using real ETH–USDT data, Dynamic Fee generated ~3.5% more cumulative fees than a static fee with the same average level.
Fees drive DEX performance, but not all swap flow is equal. Noise traders generate real revenue and prefer low, competitive fees, while arbitrageurs help align prices but extract value from LPs.
Static fees work in calm markets, but break down as volatility rises. Algebra’s Dynamic Fee adapts fees to market volatility—keeping fees low to attract organic flow and raising them when arbitrage risk increases. In simulations using real ETH–USDT data, Dynamic Fee generated ~3–3.5% more cumulative fees than a static fee with the same average level.
Algebra also introduces a Sliding Fee mechanism that targets arbitrage (“toxic”) flow by increasing fees on repeated same-direction trades, while preserving attractive rates for natural users—further improving LP outcomes.
Bottom line: static fees are simple, but adaptive fees are smarter. Algebra’s fee mechanisms improve LP profitability without sacrificing execution quality or competitiveness.
Fees are one of the most powerful levers in a liquidity pool. They directly affect LP revenue, liquidity depth, and overall DEX performance.
In practice, swap flow comes from two sources:
- Organic (noise) traders – everyday users who care about execution quality and low fees
- Arbitrageurs – traders who exploit price differences between the pool and external markets
Noise traders generate real revenue for LPs.
Noise traders are market participants who trade for non-systematic reasons, such as portfolio rebalancing, exiting a position, transfers, and other everyday use cases. These trades are typically considered unpredictable and hard to model. However, with the rise and widespread adoption of aggregators, the situation is changing: noise traders increasingly route their trades to venues offering the best execution.
Noise trader flow is the primary source of pool revenue. The fees they pay are essentially payment for the exchange service, which is the core function of a DEX. Therefore, attracting as much of this flow as possible – primarily by offering competitive market fees – is a key priority for any protocol.
Arbitrage, while necessary for price alignment, extracts value from LPs and only partially compensates them via fees.
Arbitrage on a DEX is the practice of buying a token in one venue and selling it at a higher price elsewhere, either atomically within a single block or across venues (CEX–DEX, cross-chain).
It only exists when there is an external reference market for the pair. When prices diverge, arbitrageurs trade to realign the pool price with the market. In practice, arbitrage uses LP capital to execute buy-low, sell-high strategies, and the resulting profit represents LPs’ missed value, with fees only partially compensating for this loss.
From the pool’s economic perspective, arbitrage fees are only a partial compensation for LP losses. The total LP loss is described using the loss-versus-rebalancing (LVR) metric: the best-case profit that arbitrageurs could extract given a particular trajectory of the external price. Maximal arbitrage profit equals the LP’s loss relative to implementing the above trading strategy directly.
What is the optimal fee?
Optimal fee should simultaneously attract noise traders under current market conditions and extract the maximum compensation from arbitrage. The figure below illustrates this trade-off.
The figure shows how fee revenue varies with the pool’s fee level, split between arbitrage compensation and fees paid by noise traders. Since arbitrage volume is largely driven by predictable price divergences from external markets, most automatic fee mechanisms focus on maximizing arbitrage compensation, effectively reducing LP losses (LVR).
Dynamic Fee
Dynamic fees adapt to volatility, a key signal for arbitrage risk. When volatility rises, price divergences grow, increasing arbitrage losses for LPs—so fees increase to compensate. When volatility is low, fees stay competitive to attract organic flow. Parameters like baseFee, alpha1, and alpha2 define how fees adjust across low, medium, and high volatility regimes.
The trade-off: during sharp volatility spikes, fees may lag, and higher fees can also reduce noise trader activity.
Algebra was one of the pioneers of this feature — we introduced a dynamic fee mechanism back in 2021, when it was first introduced with Uniswap V3. Since then, we’ve continued to develop and refine additional approaches.
Sliding Fee: Targeting Toxic Flow Without Penalizing Organic Trades
Alongside Dynamic Fees, Algebra offers a Sliding Fee mechanism that distinguishes arbitrage-driven (“toxic”) flow from organic trades. Arbitrageurs often execute consecutive buys or sells during price divergence, and the Sliding Fee exploits this pattern by gradually increasing fees on repeated same-direction trades while lowering fees on the opposite side. This enables higher arbitrage capture while keeping fees competitive for natural users, improving LP profitability (up to ~15% in certain regimes) without hurting execution quality. Implemented as a modular Algebra plugin, Sliding Fee can be combined with Dynamic Fees for more precise, market-aware fee control.
Learn more about it here.
The problem with static fees
Getting back to the topic. Static fees remain effective when arbitrage activity is low. In such cases—like stable pools or early-stage tokens—a competitive fixed fee can be the simplest and most reliable option.
A fixed fee is simple and predictable — and in low-arbitrage environments, it works well.
But markets aren’t static.
When volatility rises:
- Price divergences grow
- Arbitrage losses for LPs increase
- A static fee can no longer protect liquidity efficiently
Algebra’s Dynamic Fee
Algebra introduces volatility-aware dynamic fees:
📉 Low volatility → low fees to stay competitive and attract organic flow
📈 Medium volatility → balanced fees to capture more arbitrage compensation
🔥 High volatility → higher fees to protect LPs from large price dislocations
Fees automatically adapt to market conditions — without manual intervention.
Does it actually work?
We simulated arbitrage in an ETH–USDT concentrated liquidity pool using Binance prices over 2025. The pool had a single wide liquidity position (±200%), with blocks every 2 seconds and fixed transaction costs. Arbitrage was executed only when profitable; noise traders were excluded to isolate arbitrage-driven fees.
We first tested a static fee baseline across a range of fee levels to see how fees impact arbitrage revenue. While very high static fees (25–40%) maximized arbitrage fees, they would make the pool unusable for organic traders and are not practical in real DEX environments.
To fairly compare fee mechanisms, we used a matched static baseline: for each dynamic setup, we compared it against a static fee with the same average level. This isolates the value of fee adaptation over time, rather than higher fees.
Using this approach, we evaluated Algebra’s Dynamic Fee mechanism under fixed parameters.
Under our assumptions, arbitrage fee revenue is maximized at very high static fees (≈25–40%) due to the nonlinear relationship between price divergence and arbitrage profit, where rare large dislocations dominate returns. However, such fees make pools uncompetitive for organic traders and are impractical in real DEX settings.
Moreover, this “optimal” static fee is unstable over time and varies with market regimes. To fairly compare fee mechanisms, we use a matched static baseline: each dynamic setup is compared against a static fee with the same time-weighted average. This isolates the benefit of fee adaptation, rather than higher fees.
Using this approach, we evaluate the Dynamic Fee mechanism under fixed parameters.
Results
Dynamic Fee applied an average fee of 0.286% and achieved 17.4% fee APR, outperforming a static fee at the same level (16.9% APR) and generating ~3% more cumulative fees.
After optimizing parameters (baseFee, alpha1, alpha2), Dynamic Fee reached 18.0% fee APR, or ~3.5% higher fees than the matched static baseline. Performance clustered in a narrow parameter range, indicating robustness rather than reliance on outliers.
The takeaway
Static fees are simple. Adaptive fees are smarter.
Algebra’s Dynamic Fee doesn’t overcharge users — it adjusts when it matters, improving LP outcomes while keeping pools competitive.
