Yield Farming APY (Auto-Compounded)

Because the per-period rate shrinks as the number of periods grows, and the two effects balance out at a finite limit. As n increases without bound, (1 + APR/n)ⁿ converges to e^APR, the exponential function. So the maximum APY for a given APR is e^APR − 1, and nothing can exceed it. At 12% APR that ceiling is 12.7497%; at 100% APR it is 171.83%; at 1,000% APR it is about 2,617%. The convergence is very fast — hourly compounding is already within a hundredth of a percentage point of the limit at typical rates. This is why farms that compound every block are effectively at the theoretical maximum already, and why further increases in frequency add essentially nothing.

Several reasons. First, farms often quote APY computed on a short window — the last hour, the last day — and annualise it, so the number reflects a temporary emission rate rather than a sustainable one. Second, many farms include the value of their own token emissions in the yield, and if that token is falling in price, the dollar-denominated return is lower than the token-denominated APY suggests. Third, the APR itself changes as the pool's liquidity and the emissions schedule change, sometimes hourly. Fourth, some farms report the rate per block multiplied by the number of blocks in a day and then annualised without compounding, which understates the figure; others compound it, which overstates it relative to a short sample. Always check what window the quoted APY covers and whether the reward token is the same asset you deposited.

Rarely, and never for long. If a farm genuinely paid a durable 1,000% annual return, capital would rush in until the yield compressed to ordinary levels — this is the same force that drives arbitrage in every market. A four-digit APY almost always means one of five things: the emissions are inflationary and the reward token is losing value, the pool is tiny and cannot absorb meaningful capital without collapsing the rate, the APY is a snapshot of a brief spike, the reward token is not the same as the deposit token and is falling faster than the yield accrues, or the contracts carry an elevated risk of exploit. Sometimes a high APY is genuine and early, and those windows can be extremely profitable for the people who find them. But the base rate is that the yield is not what it appears, and the burden of proof is on the buyer.

Impermanent loss is the opportunity cost of providing liquidity in a two-asset pool versus simply holding the two assets. When the relative price of the two assets changes, the pool rebalances — selling the appreciating asset and buying the depreciating one — so you end up holding more of the loser and less of the winner. The loss is "impermanent" only in the sense that it reverses if prices return to their original ratio; if you withdraw after the ratio has changed, it becomes permanent. In a pool where one asset doubles relative to the other, the IL is about 5.7%. For a 4× move it is about 20%. For a 10× move it is about 42%. Those losses come out of the yield, which is why a 20% APY farm with a large expected price divergence can net less than simply holding.

Auto-compounding almost always wins, unless the gas cost of harvesting exceeds the value of the harvest. The farm compounds at every block or on a short schedule, which puts it at or near the continuous ceiling; manual compounding at a weekly or monthly cadence captures most but not all of that. The exception is on chains with very high gas costs and small positions, where a manual harvest transaction might cost more than the yield it compounds — in that case, waiting and harvesting less frequently is correct. As a rule: if the harvest value is less than roughly twenty times the gas cost, do not bother, and if the farm offers an auto-compound vault, use it.

It depends on the risk you are willing to take and where we are in the market cycle. Stablecoin pools on established protocols have historically paid 3% to 12% APY, and those are the closest thing to a benchmark rate. Blue-chip volatile pairs on major venues have paid 5% to 30%, with the yield varying with trading activity. Newer pools and smaller protocols can pay 50% to several hundred percent, but with correspondingly higher risk of contract failure, token depreciation, and liquidity evaporation. Anything above 100% APY should be treated as compensation for a specific risk you can name — if you cannot articulate what you are being paid to bear, you probably should not be bearing it. The correct comparison is always against the risk-free rate plus an honest estimate of what the additional risk is worth.