Alpha Homora V2 fees and realistic APY

Why headline APY is never what you earn

Every leveraged farming interface shows a large number, and every one of those numbers is a projection built on assumptions that expire quickly. The advertised APY of a pool typically combines trading-fee income with reward-token emissions valued at the current price, annualised from a recent window. Neither component is guaranteed to persist for a day, let alone a year.

On top of that projection, a leveraged farmer subtracts borrowing interest on the debt, any protocol fee on rewards or on borrowing, gas costs for opening, harvesting and closing, and the expected impermanent loss of the pair. What remains is the realised return, and the distance between it and the headline number grows with leverage because the subtractions scale too.

The components, one by one

Borrowing interest is the largest and most volatile cost. Rates follow utilisation: when many farmers want to borrow the same asset, the rate rises sharply, and it can do so within hours. A position modelled at a 5 percent borrow rate can be paying double that in a busy week, which is why stress-testing at elevated rates is more useful than optimising the entry number.

Protocol fees are smaller but structural — historically a share of harvested rewards or a spread on interest, directed to the treasury. Gas is a fixed cost that matters enormously for small positions and is nearly irrelevant for large ones; on Ethereum mainnet, a strategy requiring frequent harvesting can be uneconomic below a certain size.

Impermanent loss is the component most often ignored because it is not a line item anywhere. It is real nonetheless: it shows up as a smaller collateral value at exit than the price change alone would imply, and at leverage it hits equity by the multiplier. The mechanics and worked numbers are in leveraged yield farming explained.

A calculation template

The useful formula is straightforward. Net return on equity equals leverage times the gross farm yield, minus leverage minus one times the borrow rate, minus fees, minus expected impermanent loss times leverage. Concretely, at 3x with a 12 percent gross yield and a 7 percent borrow rate: 36 percent minus 14 percent equals 22 percent before fees and divergence loss. Subtract 2 percent of fees and an expected 4 percent of amplified impermanent loss and roughly 16 percent remains.

Now run the same template with pessimistic inputs, which is the step most people skip. Gross yield at 8 percent because emissions cooled, borrow at 12 percent because utilisation spiked, and 8 percent amplified divergence loss: 24 percent minus 24 percent minus fees minus 8 percent leaves a clearly negative result. Both scenarios are plausible in the same quarter, and the position that survives is the one sized for the second.

Reward tokens and the decay problem

A large share of advertised yield in farming is paid in an emitted token. Its price is under continuous sell pressure from farmers realising income, so the value of the reward stream tends to decline over the life of a position unless demand grows to match emissions. Two farmers with identical positions can have very different outcomes purely because one harvested and sold weekly while the other let rewards accumulate.

The practical implication is that yield in emitted tokens should be discounted, not taken at face value, and harvested on a schedule that matches gas costs. Durable yield — trading fees on a genuinely active pool — deserves more weight in the decision than a headline number inflated by temporary emissions, a theme continued in the strategies article.