6 min Research

Why token buybacks fail, and what treasuries should do instead

First published on X, August 2026.

The crypto industry spent about $18.8 billion on buybacks and burns since January 2025. Tokenomist Research ran the numbers in July 2026 across 27 tokens, then stress-tested the eleven programs with clean on-chain data. Two of the eleven actually shrink supply once scheduled unlocks are netted out. Thirty days after announcing, only two of ten programs beat Bitcoin performance. The playbook is now standard. The results are not.

The logic sounds airtight. Revenue comes in, the protocol buys its token on the open market, supply shrinks, price rises. Equity investors have watched this work for forty years. So why does the token version keep disappointing?

Where the money actually goes

A buyback executes on a DEX, which means every dollar pays a toll before it touches supply. Slippage takes a cut. LP fees take a cut. MEV bots take the largest cut of all, because a scheduled, public, recurring buy program is the most legible order flow in crypto. Front-running it is a business model.

Then the purchased supply meets the market. Price ticks up on the buy, and passive holders and mercenary farmers sell into the strength. Protocol revenue, generated by active users, gets socialized to the people on their way out. The announcement pop fades within a month for eight of the ten programs Tokenomist measured.

Timing makes it worse. Fee revenue peaks in bull markets, so protocols buy most aggressively at the top. CoinGecko noted that one major launchpad’s 2025 buyback purchases sat underwater within weeks of the October crash. Buying high when capacity is greatest, with nothing left to defend the price when support matters most, is the structural signature of a revenue-funded buyback. It is pro-cyclical by construction.

The variants fare no better. Buyback-and-hold parks tokens in a treasury they can leave again. Buyback-and-redistribute hands them straight back to stakers and cuts no supply at all. The largest fee-funded program in the category runs at full speed while its token’s supply still grows roughly 47% a year, because scheduled unlocks outrun the burn. A fee-funded buyback also stops the moment the fees do, which is exactly when holders need it.

And a deeper problem sits under all of this. A stock buyback concentrates a legal claim on earnings across fewer shares. Most tokens carry no such claim. Burning float improves a ratio and confers nothing.

A burn leaves nothing behind by design. A hold leaves the protocol’s own token on the books, which is the one asset a treasury cannot use: selling it undoes the buyback, and equity accounting has always treated self-held shares as a subtraction from capital rather than an asset. Either way, external capital left the treasury and a temporary chart pattern came back.

What a treasury is for

A treasury has one job: fund the mission for as long as the mission takes. That means runway, downside protection, and the ability to act when others cannot. Every buyback dollar shortens the runway to purchase a price effect the market has already learned to fade.

The better question for an L1 foundation or a protocol treasury: is there a way to convert treasury into permanent, structural demand for the token, without spending it?

Back the token instead of buying it

Mayflower is a protocol on Solana and EVM built around the Assured Value Machine, or AVM: a deterministic pricing function that is the market itself. No order book, no human liquidity, no external market maker. Buyers create the reserve, the reserve is the counterparty to every sell, and the reserve equals the area under the price curve at all times. Solvency holds as a mathematical identity, with nothing to monitor or defend.

Every AVM asset carries a floor: a guaranteed minimum redemption price backed by the reserve. The floor can rise. It cannot fall. When it rises, the gain locks in permanently for every holder.

Mayflower opens that machinery as a platform. Any team can create its own AVM market, set the fee schedule, tune the market’s sensitivity, define the floor-raising policy, and choose the reserve asset. The reserve can be any token, and live AVM markets already run with reserves denominated in BTC, ETH, SOL, and ZEC. For an L1 or any project with a native token, the move is obvious: make the native token the reserve.

That produces a new asset, fully backed by the native token, with a floor denominated in the native token. Anyone who wants floored exposure to the chain buys the asset, and their tokens flow into the reserve. Supply leaves the open market and sits behind a floor that only rises. The reserve becomes a gravity well for circulating supply: it can absorb tokens indefinitely, back every issued unit with a rising minimum value, and release most of that capital back into circulation through cash advances.

What the treasury does differently

Instead of routing revenue into open-market buys, the treasury deploys a portion of itself into its own market as the founding buyer. Two things change on day one.

First, the allocation stays on the balance sheet. The treasury now holds a floored position in its own economy, with a known worst case that can only improve. A buyback books an expense. This books an asset with a ratcheting minimum value, and the value is denominated in the thing the treasury cares about most.

Second, the treasury earns instead of spends. Operators on the platform keep the revenue their market generates: fees on issuance, redemption, cash advances, and option execution flow to an escrow the operator controls. A buyback program is a cost center with a decay curve. An AVM market inverts that, paying the treasury on every unit of volume. The incentive structure improves with it, since the operator’s return comes from fees on activity rather than from trading against its own holders.

Liquidity stops requiring sales, too. The cash advance facility lets any position holder, the treasury included, draw up to the floor value of a position at zero interest with no repayment schedule. The advance can never exceed the floor, and the floor can never decrease, so the position is solvent by construction at every moment. Forced close-out is structurally impossible. A treasury can fund operations against its position without ever putting sell pressure on its own token.

What the token gets

Compare the demand each approach creates. A buyback generates episodic demand that leaks to MEV on the way in and exits with the next seller. An AVM market generates locked, cumulative demand: every buyer of the reserve-backed asset is a buyer of the native token, and that capital stays in the reserve behind the floor. This is the demand a buyback is trying to simulate, made structural.

The community also gets a venue where a drawdown cannot become a death spiral. In a conventional market, a crash is a race for the exit, and the last one out loses everything. In an AVM market, price stops at the floor and every holder redeems at the same number. There is no advantage to panicking first, so the panic never compounds.

Skeptics have heard floor promises before, and the graveyard of broken pegs earns the skepticism. The difference is that an AVM floor requires no defense. Redemption at floor is funded before it is promised, because the reserve equals the area under the curve by definition. In December 2025 the protocol ran a public bounty inviting anyone to break the floor. It held.

The comparison, plainly

A buyback dollar leaks on entry, buys decaying price impact, subsidizes exiting holders, and disappears from the treasury. The same dollar deployed into the chain’s own AVM market becomes reserve behind a rising floor, earns fees on every subsequent trade, and remains accessible through advances, while every new participant locks more of the native token away.

Buybacks ask the market to believe in a floor. An AVM market computes one, denominated in your own token.

The mechanics are documented in Get Started.

Sources

All posts