Code over hype. That is what I tell every student in the first session of my education platform, before they have learned a single line of Solidity, before they understand what a hash function does. It is a simple instruction, and almost everyone nods, and almost no one follows it, because hype is easier to read than code.
Last week a single sentence crossed my desk: an investment firm had posted that a project called Papertrade was generating daily revenue second only to Tether, and that it returns roughly ninety percent of its rake to losers. That was the entire substance. No document, no data methodology, no time interval, no mechanism description, no cited source. One sentence, two claims, and a gravitational pull that pulled thousands of readers into believing they had learned something.
They had not. They had learned a headline. And in this market, in this particular bear market we are grinding through, the distance between a headline and a fact is where retail money goes to die.
I spent eleven years as an economic analyst before I ever translated a whitepaper, and the habit that saved me repeatedly was a boring one: when a number arrives without its definition, treat the number as a rumor. So let me do the thing that almost nobody did with that sentence. Let me define the terms, and watch the claim collapse into something far more ordinary, and far more interesting.
Start with the word revenue. In the gambling world, which is what this is regardless of how the press release frames it, revenue has at least four distinct meanings. There is handle, the total amount wagered. There is gross gaming revenue, the amount retained after payouts. There is gross rake, the fee skimmed from the pool before payouts. And there is net income, what survives after every cost including the rebates. These four numbers can differ by more than an order of magnitude on the same product, on the same day, with no dishonesty involved. When a headline says revenue without specifying which, the headline is not communicating a figure. It is communicating a feeling.
Now introduce the second claim. Ninety percent of rake returned to losers. If that is true, then net income equals gross rake multiplied by one tenth. So if gross rake is one hundred, net income is ten. That is the arithmetic. It is not complicated, and it is not negotiable. And it means that to be ranked second in the industry on net income, a single application would need a wagering volume comparable to the entire global online gambling sector. On one dApp. In one bear market.
That does not happen. And here is the part that matters: it does not need to happen for the headline to be technically defensible. Because if the word revenue in that sentence actually means gross rake, then the real net figure is one tenth of what readers absorbed, and the rank collapses from second place to somewhere in the noise. The claim was never false. It was just undefined, and the ambiguity pointed in exactly one direction, the direction that flattered the project.
A number without a denominator is not data. It is advertising wearing data's clothes.
I have audited enough of these mechanisms now to know that the interesting question is never the percentage. The interesting question is what the percentage is calculated on. Ninety percent of rake returned is a rebate. Ninety percent of user losses returned is something else entirely, and that something else is a printing press.
Follow the second version through. If a platform returns ninety percent of a user's net loss, then a user who bets both sides of a market at equal size loses almost nothing on the losing side and keeps the winning side. Repeat at scale. The platform bleeds roughly ninety percent of every hedged position, and hedged positions can be generated faster than any team can patch. No mechanism survives that for a week. So I can exclude it with high confidence, not because I have seen the contract, but because I have seen what happens to contracts that try. If the ninety percent were calculated on losses rather than rake, the product would have been drained before the headline was written.
Which leaves the rebate interpretation, and the rebate interpretation is far less generous than the headline implies. Returning ninety percent of the rake means returning ninety percent of the ten percent, not ninety percent of what you lost. The user-facing generosity is real but modest. The headline-facing generosity is enormous. That gap is the product being marketed.
This is where I want to slow down, because there is a second-order consequence that almost nobody is pricing. If net margin is ten percent of rake, then the business has brutal operating leverage. Wagering volume falls by half, net revenue falls by half, but the team, the compliance burden, the node infrastructure, the audits, the legal review, all of those stay roughly fixed. A ten percent margin business is a beautiful machine in an upcycle and a folding chair in a downcycle. We are in a downcycle. Readers asking whether their funds are safe should understand that the answer depends less on the rake percentage than on the burn rate, and the burn rate is the one number nobody publishes.
Then there is the comparison itself, and this is where I have to be blunt. Tether earns from interest on reserve assets. That income is driven by macroeconomic rate policy. It is, for practical purposes, non-competitive. Papertrade, if it earns anything, earns from users losing to each other, which is maximally competitive and fully exposed to the next clone that copies the parameter.
Putting those two income streams on one leaderboard is a category error. It is like ranking a landlord against a casino because both collected money last Tuesday. The figure may be accurate and the comparison remains meaningless, and the comparison is the only reason the sentence was written. When two numbers share a leaderboard but not a category, the leaderboard is the argument, not the data.
So where does that leave the ninety percent rebate as a competitive position? Here I want to be careful, because I have watched this exact pattern in three cycles now, and I know how it ends.
A high rebate is a price. Prices are copyable. Any competitor with a competent contract engineer can match or exceed a rebate percentage within one development sprint, and several will, because the parameter is the easiest thing in the stack to change. What follows is an arms race in rebate ratios that compresses gross margin across the entire vertical until the only survivors are the platforms with some other moat: liquidity depth that produces better odds, a license that makes the product legally durable, or a community so embedded that switching costs are social rather than financial. A rebate is not a moat. A rebate is a discount, and discounts are how you buy volume when you do not yet have a moat.
I want to say something uncomfortable here, because this industry has trained itself to treat any revenue at all as vindication. It is not. Revenue measures whether money entered the building. It says nothing about whether that money reaches token holders, whether it funds buybacks, whether it sits in a treasury controlled by a multisig nobody has inspected. Revenue is a fact about a business. Token value is a claim about a distribution, and the two are only connected by a document most projects never publish.
Now the part that keeps me up at night, and the reason I am writing this instead of letting the headline pass.
There is a version of this structure that is genuinely elegant and that I would defend. If the rebate is paid in a stablecoin, calculated on rake, settled on a fixed cycle, defended against multi-account farming, and disclosed with its parameters in public documentation, then the design is not a trick. It is an honest growth subsidy, a deliberate choice to trade margin for retention, and I respect it. It is the same bet I made when I left my corporate job in 2017 to teach instead of trade. You give up the fast money for the durable relationship.
But there is another version. If the rebate is paid in a native token, especially a locked or linearly vested one, then the rebate is not a subsidy at all. It is deferred selling pressure wearing a gift ribbon. The daily metrics look magnificent. The token holders absorb the bill over a vesting schedule, slowly, in a way that never appears in the headline. I have watched this exact maneuver destroy more retail portfolios than any exploit, because an exploit announces itself and a rebate schedule does not.
And there is a third version, the one I have seen most often. The rebate mechanism requires adjustable parameters. The rake ratio, the settlement window, the definition of a loser, the eligibility rules. Every one of those is a knob. A protocol with that many knobs requires an admin key. A protocol with an admin key is not a protocol in the sense the word is used in marketing. It is an application with an owner, and the ninety percent figure is a setting, not a property.
A generous number written into an upgradeable contract is a promise from a person, not a guarantee from code.
This is the point where readers usually ask me what I think the project is worth. I will not answer that, and the refusal is the most honest thing in this article. I do not know whether the project has issued a token. I do not know whether the income figures come from a single peak day or a thirty-day average. I do not know whether the rebate settlement is per hand or per cycle, which determines whether the mechanism is farmable. I do not know whether the investment firm publishing the claim holds a position in the project, and that single unknown determines whether the sentence was research or a soft press release with academic framing.
All of those unknowns are resolvable. That is what makes this frustrating rather than mysterious. A block explorer will show the transaction count. A public dashboard will show whether the volume is a spike or a trend. The documentation, if it exists, will state the rebate basis in one sentence. A disclosure page will state the investment relationship. Any single one of these checks takes less time than reading the headline took, and almost nobody ran them.
That is the real story here. Not Papertrade. Not the ninety percent. The story is that a sentence with no denominator, no timeframe, and no methodology moved through the ecosystem as though it were a finding, and the ecosystem mostly did not notice the difference. Information does not stay intact as it travels. It sheds its qualifiers first, then its context, then its source. What arrives at the end of the chain is not the claim. It is the impression the claim was engineered to leave.
Truth decays slowly, and then it decays fast, and the decay is always in one direction: toward the version that feels most exciting. A careful analyst writes that revenue appears to be gross rake on a single peak day, second among applications on one chain, with an investment relationship that has not been disclosed. What survives transmission is that the project is second only to Tether. Those two sentences describe the same event. Only one of them is useful, and it is never the one that spreads.
I do not know where this project ends up. I know where the pattern ends up, because I have watched it three times. The rebate gets matched, the margin compresses, the token unlocks, and the community that arrived for the yield discovers it was paying for its own retention. Build anyway, I tell my students, because the technology is still the most interesting thing we have, and because the people who build through this part of the cycle are the ones who are still standing when it turns.
But build with your eyes open. Hold the line on definitions. When someone hands you a number, ask what it is measured against, over what period, by whose accounting, and who benefits from the way you read it. If the answer is that the number was never defined, then you have not been given information. You have been given a mood.
And in a bear market, moods are the most expensive thing you can buy.
