How Much ETH Should a Treasury Keep for Token Swaps?

A treasury should keep enough ETH to pay for its expected swaps and related transactions during a period of elevated fees, plus a buffer for delays and failed execution. There is no useful universal target in ETH: the right amount depends on how often the treasury trades, which network it uses, and how quickly it can replenish the reserve. Treat ETH as operating inventory, sized from transaction records and a defined service level, rather than as a fixed percentage of the treasury’s token holdings.

Why do token swaps need an ETH reserve?

On Ethereum mainnet, a swap’s transaction fee is paid in ETH even when the asset being exchanged is a stablecoin or another token. The fee depends on the computation required and the network’s demand when the transaction is included. A treasury therefore needs ETH on the same network as the transaction; ETH held elsewhere may not be available to pay that fee without a transfer or bridge.

A token swap can also require a separate approval transaction before the first trade with a particular token and contract. That approval uses gas too. Once permission is in place, later swaps may not need the same setup step, although a new token, contract, or allowance policy can bring it back. For a fuller account of the swap path and its moving parts, see fermi swap. The treasury should count approvals, swaps, and any required follow-up transactions, rather than budgeting for the exchange alone.

Gas is only one part of execution cost. A swap may also involve a venue fee and price impact, while slippage determines how much the received amount can differ from the quoted amount before the transaction reverts. These costs affect how much value the treasury receives, but they are distinct from the ETH needed to submit the transaction. Keeping those categories separate makes the reserve easier to explain and monitor.

How should a treasury calculate its ETH buffer?

Start with the treasury’s actual transaction history or a planned operating schedule. Estimate the gas units for each transaction type, then multiply by a fee assumption that reflects the conditions under which the treasury must still be able to trade. Ethereum’s fee has a base component and a priority fee; congestion can push the base fee higher, and urgency can make a larger tip appropriate. Use recent observations and a deliberately stressed case, not one quiet hour as the budget.

A practical reserve model is:

• List expected swaps by network and transaction type over the replenishment period.
• Include first-use approvals and other setup transactions where they are likely.
• Estimate gas for each type and price it using a stressed fee assumption.
• Add a separate operational buffer for fee spikes, retries, and replenishment delays.

Convert the resulting fee budget into ETH using the price at the time the reserve is set, then define how often the treasury will review and replenish it. The buffer should reflect the consequence of running short. A treasury that can wait for a routine transfer can tolerate a smaller emergency margin than one whose swap is needed to meet a settlement deadline. Make the replenishment trigger explicit: for example, compare the available ETH with the cost of the planned transactions under the stress case, rather than relying on a vague minimum balance.

Keep the estimate tied to the specific network and transaction path. A route through several contracts can use more gas than a direct swap; a rollup may have a different fee profile and native fee token. A dashboard that combines balances across networks can hide this distinction. Track spend and balance per chain, and record failed or retried transactions so the next estimate reflects how operations actually behave.

Should the treasury hold more ETH or use another route?

The alternative to a larger ETH reserve is not free execution; it is a different balance of liquidity, delay, and control. Holding less ETH reduces idle inventory and exposure to ETH price moves, but a shortfall can force the treasury to delay a swap or make a separate transfer at an inconvenient time. Holding more improves readiness while tying up value in an asset that may not match the treasury’s liabilities.

Moving swaps to a lower-cost network can reduce the fee burden, but it adds network selection and liquidity considerations. The treasury must hold the relevant gas token there and ensure the desired trading route can meet its size and timing needs. Bridging assets or replenishing a separate network balance adds steps and possible delay, so a lower quoted transaction fee does not by itself make that route better for every operation. Compare total execution cost and operational reliability, not gas in isolation.

For most treasuries, the sound starting point is a small, ring-fenced ETH operating reserve sized to a stated replenishment window and a stressed transaction budget. Review it as trade frequency, routes, and settlement urgency change. The signals to watch are fee levels during the treasury’s actual trading windows, the share of transactions requiring new approvals, repeated failures or retries, and how long it takes to replenish ETH on each network. Those measures show whether the reserve is too lean, unnecessarily large, or simply held on the wrong chain.