Every so often someone shows me a farm paying some healthy number, maybe twenty or thirty percent APY, and asks whether they should put a few hundred dollars in. The APY is real. The problem is that APY is quoted as a rate on your capital and says nothing about the flat cost of getting in and out, and on a small position that flat cost is the whole story. You can be earning a genuine twenty percent and still finish the year down, because the yield is a percentage of your stake while gas is a fixed number of dollars that does not care how much you deposited.
The way I think about it is that a yield position has a size-independent cost stapled to it. You pay to enter, you pay every time you compound or harvest, and you pay to exit. Those are dollar amounts, not percentages. So the effective yield on a small ticket is the headline rate minus a fixed drag, and that drag shrinks as your position grows. There is a position size below which the drag eats the whole thing, and knowing where that line sits is most of the decision.
The fixed cost you are actually paying
Start by counting the transactions a full round trip requires, not the ones the interface shows you up front. A simple deposit-and-hold farm is usually two or three signed transactions on the way in. If the vault pays in a reward token you have to sell, you might need an approval, a claim, and a swap for every harvest. Then you pay again to withdraw. On a lending position it can be as light as one approval and one supply. On a leveraged or looped strategy it is many more, because each loop is its own borrow and re-supply.
My rule of thumb for the transaction count on common strategies looks like this:
- Single-asset lending, deposit and later withdraw: roughly 3 transactions total, counting one approval.
- LP position with a reward token: roughly 4 to 6, since providing liquidity, staking the LP, and later unstaking and removing all cost gas, plus each harvest.
- Looped or leveraged farming: often 8 or more, and every rebalance adds to the pile.
Now attach a dollar figure to each transaction. On mainnet Ethereum a routine DeFi interaction historically runs anywhere from a couple of dollars in quiet periods to well over fifty dollars when the network is busy, and a swap or an add-liquidity call costs more than a plain transfer because it touches more contract storage. On the major L2s the same interaction is typically a small fraction of that, often cents to low single-digit dollars, because they post compressed data back to mainnet and split that cost across everyone in the batch. That gap between chains is the entire reason this math matters.
The break-even formula
Here is the piece worth writing on a sticky note. Let G be the total gas you expect to pay for a full round trip, entry plus all harvests or compounds plus exit. Let R be the strategy's net APY as a decimal, so twenty percent is 0.20. Let T be how long you plan to hold, as a fraction of a year. Your gross yield in dollars is your position size P times R times T. You break even when that yield equals your gas:
P times R times T equals G. Solve for P and you get P equals G divided by (R times T).
That is the minimum position size where the strategy stops costing you money. Below it you are paying to participate. A quick worked feel for it: if a round trip costs you 40 dollars in gas, the net APY is 0.15, and you hold for half a year, then your break-even size is 40 divided by (0.15 times 0.5), which is 40 divided by 0.075, roughly 533 dollars. And break-even is not the goal. To actually keep, say, three quarters of the headline yield rather than hand it to gas, you want to be several times above that line, because at exactly break-even your net return is zero.
Two things fall out of this immediately. Shorter holds raise the break-even size, because T shrinks the denominator, which is why quick in-and-out farming on mainnet almost never works for small money. And a lower net APY raises it too, so the safe, boring 4 percent lending positions need a much larger ticket to justify the same gas than a spicier farm does.
Compounding is where small positions quietly bleed
The formula above treats gas as a single lump, but the sneaky version is compounding. Auto-compounders and manual harvest-and-restake loops sell your reward, swap it, and re-deposit, and every one of those cycles is more signed transactions. On a large position, compounding weekly is clearly worth it because the extra yield from reinvesting dwarfs the gas. On a small position, each compound can cost more in gas than the reward it is reinvesting, so you are literally paying to shrink your stack.
The honest test for compound frequency is to compare the reward accrued since the last compound against the gas cost of one compound. If the pending reward is smaller than the gas, do not compound yet. This is why a lot of small farmers are better off letting rewards accrue and harvesting rarely, or picking vaults that socialize compounding across many depositors so the per-user gas is diluted, which is exactly the kind of vault that earns its keep on a small ticket.
A quick decision workflow
When someone asks me whether a farm is worth it at their size, I run the same short check:
- Count the round-trip transactions honestly, including approvals and every harvest you actually intend to do.
- Estimate G by multiplying that count by a realistic per-transaction gas cost for the chain you are on, using a busy-day number rather than the best case, because you do not control when you will need to exit.
- Pick your real net APY, after fees and after any reward-token slippage, not the headline.
- Estimate your hold length T as a fraction of a year.
- Compute P equals G divided by (R times T). If your intended position is not comfortably a few times that number, the farm is too small for this chain.
Most of the time the fix is not to abandon the strategy but to move it. The exact same farm that is a losing proposition for a few hundred dollars on mainnet is perfectly fine on an L2 where G drops by an order of magnitude and the break-even size falls with it. So the practical output of the math is usually a chain choice, not a yes-or-no. Small ticket, cheap chain. Large ticket, and mainnet's deeper liquidity and lower slippage can start to pay for its own gas.
One failure mode worth naming, because I have watched it happen: people size the entry correctly and forget the exit. Gas spikes are correlated with exactly the moments you want out, when the market is moving and everyone is transacting at once. If your position is only marginally above break-even in calm conditions, a volatile exit can push the round-trip gas past your entire yield and you leave with less than you came in with. Budget the exit at a stressed gas price, not a quiet one.
None of this requires fancy tooling. It is one division you can do on your phone before you sign the first approval. When I am comparing where a given strategy should live across chains, I lean on Blockcircle to see the yield and the execution costs side by side, but the arithmetic is the part that keeps small positions honest. Gas is a fixed cost, yield is a rate, and the whole question is whether your size is big enough to make the rate win.