Two sources, very different economics
A protocol flashes 40% APY and the first thing I want to know is where that yield is actually coming from. There are really only two answers. Real yield comes from economic activity that already happened: trading fees users paid, interest borrowers paid, liquidation penalties the protocol collected. Inflationary yield comes from the protocol minting fresh tokens and handing them to you. On a dashboard they look identical. Underneath they could not be more different.
Real yield works like a dividend. Somebody paid for a service and a slice of that payment lands in your wallet. The yield exists because value moved. Inflationary yield is closer to a stock split with a small cash sweetener. You end up holding more tokens, but the total supply grew too, so each token is a thinner slice of the same pie. Unless the price climbs enough to cover the dilution, you made nothing in real terms.
How to spot the source
Most protocols mix both, which is exactly why it's hard to see what you're really earning. A liquidity mining program might advertise 50% APY where 45% is token emissions and 5% is trading fees. The 5% holds up as long as volume does. The 45% depends on the token price not falling, and it almost always falls, because the people earning those rewards are selling them.
To pull the two apart, go to the protocol's docs or analytics. Take the total fees over a recent window, divide by total value locked, and you've got the fee-based yield. Subtract that from the headline number and whatever's left is coming from the printer. If a pool shows 30% APY and the fee yield is 3%, the other 27% is emissions.
Token Terminal, DeFiLlama, and a lot of protocol dashboards break this down for you. The pattern almost never changes. The flashiest APYs are dominated by inflationary yield, because that's the whole point of high emissions: buy liquidity into a new pool or bootstrap some network effect before organic usage exists.
The death spiral
Inflationary yield tends to run the same loop every time. A protocol launches with fat emissions to pull in liquidity. Early farmers earn big in the native token. They sell to lock in gains. Selling drags the price down. As the price drops, the dollar APY falls even if the token-denominated APY looks the same. Lower APY pulls in less new money. Existing farmers, now earning less, yank their liquidity. TVL drops, the protocol gets less useful, organic fees shrink, and the real yield gets even thinner.
This ran hundreds of times through DeFi summer in 2020 and the yield-farming stretch of 2021. Protocols would launch with a jaw-dropping APY, vacuum up billions in TVL over a few days, then slowly bleed out as the price slid and mercenary capital rotated to the next thing. The survivors were the ones generating enough real usage, and enough real yield, to keep liquidity around after emissions cooled off.
What a realistic number looks like
So what counts as believable real yield? It depends on the asset and the risk. A few rough benchmarks I keep in my head:
- Stablecoin lending on established venues like Aave or Compound: usually 2-8%, moving with borrowing demand.
- Volatile pairs on DEXes: fee yield of 5-30% on high-volume pairs, though impermanent loss quietly eats a chunk of that.
- Liquid staking: anchored to the underlying reward, roughly 3-5% for ETH, plus whatever extra you earn using the staking token elsewhere.
Anything meaningfully above those ranges should make you stop and ask what you're missing. A stablecoin pool paying 25% is either taking on risk you can't see (smart contract, bridge, counterparty) or paying you in a token that's inflating away. There's no risk-free high yield in DeFi, same as there isn't one in traditional finance.
Real yield as a quality signal
Protocols that throw off real yield tend to be better long-term holdings, and the reason is boring: they have product-market fit. If people are paying fees to use the thing, the thing is doing something valuable. Fees are proof of demand. Inflationary yield only proves the treasury is spending tokens to rent attention.
Comparing real fees against fully diluted valuation gives you a quick quality read. A protocol pulling $50 million in annual real fees at a $500 million FDV is running a 10% fee yield, which is genuinely attractive. Same $500 million FDV but only $5 million in real fees is 1%, which means the market is pricing in a lot of growth that may never show up.
When I look at a farm, I strip out the inflationary part and check what the position would earn if emissions stopped tomorrow. If that number still looks good, there's a real foundation under it. If it's basically nothing, then the whole trade is you betting you can exit the inflating token before the crowd does. That's a timing bet, so size it like one and don't confuse it with an investment.