Every time I look at a new protocol with a big TVL number, the first thing I do is try to figure out how much of it is being paid to stay there. Not because the number is fake, but because a large chunk of it is usually a paid audience that will leave the moment the check stops clearing. TVL is the metric everyone quotes and the one that reacts most violently to incentive changes, and that gap between how much attention it gets and how fragile it is has cost a lot of people money.
The mechanism is simple enough. A protocol wants deep liquidity or a lot of deposits, so it prints its own token and pays it out to anyone who supplies capital. If the emissions are generous, the yield looks great, and capital shows up. The problem is that this capital has no loyalty to the protocol. It is there for the yield, and the yield is denominated in a token whose price is partly propped up by the same deposits it is trying to attract. When emissions get cut, the yield drops, and the mercenary capital rotates out to wherever the next generous program is. The TVL falls, the token's own liquidity thins out because a lot of the deposits were in trading pairs, and the headline metric that made the protocol look healthy evaporates in a few days.
Splitting organic deposits from rented ones
The goal is to estimate what fraction of TVL would still be there if the token rewards went to zero tomorrow. You will never get this exactly, but you can get close enough to make a decision. A few things I look at.
- Reward APR versus base APR. Separate the yield a depositor earns from the protocol's actual activity, so trading fees, borrow interest, whatever the thing genuinely produces, from the yield that comes from token emissions. If a pool advertises a big number and almost all of it is emissions, assume most of that capital is renting. If the base yield alone is competitive, more of it is likely to stay.
- Deposit concentration. Pull the top depositor addresses and see how much of TVL sits in a handful of wallets. Mercenary capital tends to be large and professional, so a few farming wallets holding most of the deposits is a warning. Organic usage is usually more spread out and messier.
- What the deposits actually are. A stablecoin pair that people use to swap is stickier than a pool that only exists because it prints the governance token. Ask whether anyone would keep that position on for a reason other than the reward.
- Age of the deposits. Capital that arrived the same week the incentive program launched behaves differently from capital that was there before. If TVL and the emissions program have the exact same birthday, that is not a coincidence.
None of this gives you a clean percentage. What it gives you is a defensible guess, something like "maybe a third of this is organic and the rest is renting," which is far more useful than treating the whole number as real.
Reading the cohorts after past reward cuts
The best evidence for how sticky a protocol's TVL is comes from what happened the last time rewards were reduced, because most incentive programs get cut or tapered eventually. If the protocol has been around for a while, that history is on-chain and it is honest in a way the current APR never is.
The workflow I use is roughly this. Find the dates when emissions were cut or a reward program ended. For each of those events, look at TVL right before and then track it over the following weeks. What you want is the retention curve, so what share of deposits stuck around one week later, one month later, and so on. A protocol where TVL drops by more than half within days of every reward cut has trained its users to be mercenaries, and there is no reason to expect the next cut to go differently. A protocol where TVL dips a bit and then stabilizes has an organic base underneath the rented layer, and that base is the part worth paying for.
If you can track individual wallets, go one level deeper. Take the cohort that was farming before a cut and see how many were still deposited a month after. The addresses that leave the instant yield drops are the rented share made visible. The ones that stay through cuts, market drawdowns, and boring stretches are the closest thing to real users a DeFi protocol has.
Pricing the token as if the rented share is gone
Once you have a rough organic fraction, the valuation step is to redo the token's fundamentals on the smaller, stickier number and see whether it still holds up. This is the part most people skip because the full TVL number makes everything look cheap.
Take the fees or revenue the protocol earns and ask how much of it depends on the rented capital. If half the TVL is mercenary and it produces a proportional share of the fees, then a realistic post-emissions revenue is closer to what the organic half generates, not the headline total. Now compare the token's market cap or fully diluted value against that lower revenue. A token that looks like it trades at a reasonable multiple of headline fees often looks expensive once you strip out the fees that only exist because the protocol is paying for them with its own supply.
The other side of this is dilution. The emissions that rent the TVL are new token supply hitting the market, usually sold by the same farmers earning them. So you are often paying a premium multiple on revenue that is partly fake while the token supply grows to fund the very deposits inflating that revenue. When you write it out that way, a lot of "cheap" tokens stop looking cheap.
A quick sanity check I keep coming back to. If this protocol turned off token rewards entirely tomorrow, would people still use it, and would the token still be worth roughly what it is now? If the honest answer is that TVL would collapse and the fees would follow, then the fundamentals you are underwriting are a paid audience, and paid audiences leave.
What to actually do with this
I do not treat rented TVL as an automatic disqualifier. Bootstrapping liquidity with incentives is a legitimate strategy, and plenty of durable protocols started exactly this way. What matters is whether there is anything underneath the incentives, and whether you are paying for the durable part or the rented part.
So before I underwrite a token on its TVL, I try to answer three questions. What fraction of this survives a reward cut, based on how past cuts actually went. What is the token worth on that surviving fraction alone, after the dilution that funds the rest. And is the current price closer to the headline number or to the honest one. When the price only makes sense at the headline number, I pass, because the metric holding the whole thing up is the easiest one in the space to rent and the fastest to leave.