Scope & Results

Looking at select pools on ethereum mainnet:

• Are there structures among pools that may indicate different fitness for gas subsidies (fee tiers, MEV bot usage, etc.)

Gas subsidies should avoid arbitrage-heavy niche pools (e.g. "token-stable"). The most popular pools are also the most retail adjacent- further analysis on routers used, time of day, etc. are warranted.

• Are there historical patterns on how gas costs correlate with or possibly reduce LP fees available from swaps?

All pool types suffer when gas is very high. But some niches (e.g., ETH / staked-ETH) offer an opportunity to support LST pegs and target less bot/mev/arbitrage pools - despite these pools typically having smaller LP fees relative to volume.

Why subsidize gas?

Ethereum is well known for its high variance in transaction costs.

Higher transaction costs can dissuade users from even mutually beneficial trades with liquidity providers.

On decentralized exchanges traders face the following costs:

• Direct transaction cost (blockspace on ETH)
• Direct liquidity pool fee (paid to liquidity providers)
• Slippage (acceptable loss due to potential changes in price prior to trade execution)
• Price impact (the resulting change in price from having sold a token)

Liquidity providers (LPs) want more fees, traders want lower fees and better prices (lower bid-ask spread). Transaction costs are a drain on both parties. For decentralized exchanges to compete with centralized exchanges, they must retain and grow liquidity provided by LPs.

But removing transaction cost is not trivial. Also, Many traders may be so large, that the transaction costs are small compared to other costs.

Bots & Power Users

On ethereum, transactions include a nonce which indicates the # of transactions in the user's tx history (this enables transactions to be unique as the nonce goes up by 1 every transaction).

Bots & Power Users submitting thousands of transactions a month, with 100,000s or even millions of transactions in their history are disproportionately engaged in:

• ETH/BTC
• BTC/Stablecoin
• Token/Stablecoin

This may seem counterintuitive as the vast majority of value flow is to ETH-Stablecoin and Stablecoin to Stablecoin pools.

But this should actually make sense. Smaller operations and non-power users coalesce to the top pools, while bots arbitrage these top pools against less popular pools (at high frequencies).

High tx costs can be ignored by whales, to a limit.

The category of "long tail" assets (here bundled as token-stable) seem most affected by gas price. But it is worth noting that low transaction fees are not necessarily associated with large increases in Liquidity Pool fees. A few reasons stand out:

• By definition, lower onchain activity = lower gas fees. So if trading falls, both gas and LP fees fall together.
• lower onchain activity may be correlated with lower USD prices across crypto ("bear market") which would suppress USD denominated fees given the same trades.