Where a token’s value comes from, and how you see whether it gets there

Why is bitcoin the simplest case to read?

The comparison with ordinary currencies sits entirely there. A central bank can decide to print, and a meeting decides it — discretionary monetary policy. Bitcoin has a monetary policy on a schedule, which no meeting can change. Issuance went from 50 bitcoin a block in 2009 to 3.125 in 2024, and will keep halving. Calling it digital gold isn’t marketing. It’s a description of the rule.

Two warnings, though, and the first is a big one. Bitcoin holds as a store of value for as long as the market treats it as one. It is an agreement about what something is worth, exactly as gold is, and an agreement can come apart. The second is about the practical comparison. It swings far more than physical gold, but it sells at any moment and in any amount, it divides down to the eighth decimal, and it costs nothing to keep in a vault, because there is no vault.

What does it mean for a token to capture value?

This is the difference that separates a token from a share. A share carries a right to the profits by law, and if the company earns, that right is worth something even when nobody exercises it. A token gets what the protocol’s code gives it, nothing more. If the code provides for no passage of value, there is no court to ask for it.

There are four possible roads, and the next four sections take one at a time: burning the tokens used to pay, paying whoever stakes the token, handing out a share of the revenue, or granting nothing but a voting right. The first three move value. The fourth, on its own, moves only the hope that one of the first three arrives one day.

How does burning the fees work?

The big example is Ethereum after the 2021 change that introduced the base fee — the mandatory part of every fee gets burned. When the network is used heavily, Ethereum burns more than it issues and the amount in circulation falls; when it is used lightly, it grows again. The mechanism ties demand for the network directly to the scarcity of the token, with nobody having to decide it.

The warning sign here is theater. A project that announces it is burning tokens while it goes on issuing them with no cap is doing a subtraction with two wrong numbers. Burn a hundred and issue a thousand and the scarcity is a press release. The figure to look at is always the net one — issued minus burned — and it wants looking at over months, not on the day of the announcement.

Where do the staking rewards come from?

A staking reward that comes from users’ fees rises and falls with how much the protocol is used and dilutes nobody. One that comes from new issuance pays whoever stakes by diluting everyone else, and holds only while the price rises.

If it comes from the fees, the circle is a healthy one. The more the protocol is used, the more fees come in, the more there is to hand out — and if use falls, the payment falls. That is uncomfortable, and it is the sign that it is real. If it comes from new issuance, what happens is that everyone else is diluted to pay whoever stakes. It works while the price rises, and stops exactly when it is needed.

Ethereum, after the move to proof of stake, sits in between in an interesting way. It issues very little — around half a percentage point a year — and burns the base fees at the same time. The result is a system close to zero-sum, which in periods of heavy use turns negative. So the validator’s yield comes largely from the people using the network, not from the people holding it.

Is revenue sharing a dividend?

GMX, where perpetuals trade on Arbitrum, passes 30% of its trading fees to whoever stakes its token. The more volume goes through, the more fees come in, the more arrives. The link is direct and can be followed onchain, fee by fee. The advantage isn’t only the transparency, it’s the alignment — whoever holds the token wants the protocol used, not the price rising on its own, and for once the two coincide.

The warning sign is the mirror image, and simple to check. A protocol that promises to share its revenue but has no users is sharing zero, to a great many decimal places. Before you look at the share, look at the volume — and look at it over a long stretch, because volume bought with rewards leaves along with the rewards.

Is a voting right, on its own, worth anything?

Put bluntly: governance tokens are democratic in theory, and in practice whoever holds the most wins — a vote weighted by wealth.

In theory controlling a protocol that moves billions is worth something, and in theory that value should sit in the token’s price. In practice, if the vote is tied to no flow of money, what you are buying is the chance that one day the vote decides to give you one — which is a wait. UNI’s price, over those years, did worse than the tokens that already had a road written for them.

Then the wait ended. In November 2025 the governance turned the fee switch on, and from there a part of the fees starts coming back to the token. That confirms the rule rather than breaking it — the value arrived when a mechanism was written, not when one was promised.

What are the signs that the tokenomics are broken?

The five signs of broken tokenomics: endless issuance, which is continuous dilution; three-figure yields, which are paid by printing; invented utility, which is forced demand; unlocks close by, which are sales coming; capital inflated by rewards, which is adoption on rent.

The second is the three-figure yield. Almost always it means they are paying you by printing, and it holds while new people keep arriving. The third is invented utility — the token you “need in order to use the platform” when the platform would work perfectly well with ether or a stablecoin. It is an internal currency made compulsory to create demand where there isn’t any.

The fourth sits in the calendar, and it wants reading in full. If three months from now millions of tokens held by the team and the investors unlock, that date is more important information than any chart. The fifth is capital inflated with rewards — it looks like adoption and it is rent. As soon as the rewards come down, the capital moves elsewhere the same week.

Put bluntly: a token that exists only to use the site of whoever issued it is Netflix making you buy NetflixCoin to watch the shows.

How do you judge a token, in practice?

Two: follow that value’s road all the way to the token. Is it burned, handed out, paid to whoever stakes, or does it not arrive? Three: check that it holds without incentives. Take the rewards away and ask whether anyone would still use that protocol. If the answer is no, you are not looking at an economic activity. You are looking at a promotion.

Four: look at the supply numbers — how much is issued, how much is burned, how much is already in whose hands, and when they can sell. Five: ask whether the product solves a problem somebody actually has. Five out of five is rare. Four out of five is worth an argument. Three out of five is a no dressed up as a maybe.

Where are the tokenomics that work?

The four families of tokenomics that hold: yield from real fees, recognizable because it follows the volume; tokens burned by use, where scarcity grows with no announcements; declared revenue sharing, with a fixed and verifiable share; governance with locked tokens, where whoever votes lives with the decision.

The second is tokens that burn when the protocol is used. If use grows, scarcity grows by itself, with no announcements. The third is declared revenue sharing — a fixed share of the revenue to whoever holds the token, written down and verifiable. Simple, and hard to fake for exactly that reason.

The fourth is governance with something at stake. Voting means locking up the tokens, so whoever decides lives with the consequences of what they decide. It isn’t value capture yet, but it is the difference between a vote and a poll. Everything outside these four — unsustainable yields, uncapped issuance, promises with no users — isn’t a new family. It’s the same thing under another name.

What is left to ask, every time?

The market is full of projects with excellent marketing and badly built tokenomics, and telling them apart isn’t a talent. It is a check you run in half an hour with the protocol’s public data. Whoever runs it buys a mechanism. Whoever doesn’t buys the wait for somebody to buy after them, which is a different trade, and a far more crowded one.

The uncomfortable part is that this check rules out almost everything. Most tokens don’t get past the first step, and the right answer in those cases is not to buy — which isn’t a missed opportunity. It’s the same thing the seller is doing.