// Research

Which Token Value Accrual Model Works?

We mapped 159 tokens across six value accrual mechanisms and tested which ones translate into shareholder returns. The answer is not what the crypto narrative insists.

Apr 2026 159 protocols Data powered by Artemis
Summary

The alpha is not in the mechanism

Three findings

1. Active value accrual (direct fee, buyback-burn, buyback-hold, ve-model) outperforms governance-only by 10 percentage points on average across the dataset.

2. Revenue scale, not mechanism design, drives returns. The protocols returning capital while also generating real revenue outperform everything else by a wide margin.

3. Governance-only is a dead model. 48 tokens scored, median return -67%, only 1 positive in the cohort. Institutional allocators actively avoid this structure.

Protocols assessed
159
Across 6 accrual models
With empirical data
135
1Y performance + revenue
Positive 1Y return
5
3.7% of protocols with data
Median 1Y return
-66%
Unweighted across models
Framework

Six value accrual models

Every liquid token that accrues value does so through one of six mechanisms. Each has a TradFi analog. Each creates a different investor profile.

Model 01

Direct fee distribution

Protocol fees flow to token stakers in cash, stables, or native assets. The most legible model for allocators because it maps to a dividend.
Sample
13
Avg 1Y
-61%
Median
-71%
Examples: GMX, dYdX, Jito, Rocket Pool, Gains Network
Model 02

Buyback and burn

Protocol revenue funds buybacks that are destroyed. Reduces supply over time. The top-performing model in the dataset, carried almost entirely by Hyperliquid.
Sample
12
Avg 1Y
-35%
Median
-66%
Examples: Hyperliquid, Jupiter, Raydium, Orca, PancakeSwap, Kamino, Ethena, Pendle
Model 03

Buyback and hold

Protocol revenue funds buybacks held in treasury, staked, or redistributed. Creates a protocol-owned float and optionality without burning supply.
Sample
10
Avg 1Y
-51%
Median
-41%
Examples: Meteora, Aave, Sky, Lido, Fluid, Maple
Model 04

Vote-escrow (ve-model)

Lock tokens for time-weighted share of fees, emissions, and bribes. Creates active management overhead but can flywheel meaningfully.
Sample
14
Avg 1Y
-67%
Median
-77%
Examples: Curve, Aerodrome, Velodrome, Balancer, Convex
Model 05

Governance only

No fee switch, no buyback, no economic accrual. Token holders vote and nothing else. $0 flows to holders despite protocol revenue.
Sample
48
Avg 1Y
-65%
Median
-67%
Examples: Uniswap, Arbitrum, Optimism, EigenLayer, Compound, Morpho
Model 06

Other / hybrid

Points programs, RWA-backed, LRT, memecoins, stablecoins. Heterogeneous by design. Investor legibility varies wildly.
Sample
62
Avg 1Y
-71%
Median
-77%
Examples: EtherFi, Virtuals, Pepe, Ondo, Liquity, Worldcoin
Chart 01

Average return by accrual model

Every model in the dataset printed negative 1Y returns on average. The spread between the best and worst is 14 percentage points.

1-year return by value accrual model (%)
Market-cap unweighted average. Apr 2025 – Apr 2026.
The punchline
Strip out Hyperliquid, and buyback-burn collapses from -35% to -56%. One outlier carried the model. Mechanism design is table stakes, not alpha.
Chart 02

Revenue scale is the real signal

Sort the 50 protocols with clean Artemis revenue data by daily revenue and the pattern is clearer than any mechanism split. The top quintile by revenue averaged +8% returns; the bottom quintile averaged -81%.

1Y return vs daily revenue (log scale)
Each point is a protocol. Color by value accrual model. Bubble size by FDV.
Interpretation
The two tokens with >$500K/day in revenue are Hyperliquid and Polymarket. Both are the standout performers. Accrual model didn't matter for either. Revenue did.
Findings

What the data actually says

01

Any active accrual beats governance-only by 10 points

The 49 protocols across direct-fee, buyback-burn, buyback-hold, and ve-model averaged -55% over the trailing year. The 48 governance-only protocols averaged -65%. The gap widens when restricted to revenue-generating governance-only tokens like Uniswap, Arbitrum, and EigenLayer, where the opportunity cost of not accruing is most visible.

Governance-only is the IR equivalent of a public company paying no dividend and buying back no stock. Eventually allocators stop pretending it's a going concern and start pricing it as an option on management waking up.
02

Hyperliquid is the buyback-and-burn category

On headline numbers, buyback-and-burn won this year (-35% avg) and buyback-and-hold came second (-52%). That reads as a clean win for burns, but the story inverts when you strip out Hyperliquid. Ex-HYPE, buyback-and-burn averaged -56% and buyback-and-hold averaged -52%. One token decided the category.

Meteora is the cleanest buyback-and-hold case. $10M buyback program, 95/100 on Novora's IR Score, treasury accumulation transparently disclosed. Down 22% on the year, less than the cohort median. Holding the bought tokens in a transparent treasury preserves optionality and creates a visible, audited float. Burns destroy optionality in exchange for a marketing headline.

03

The ve-model requires perpetual bribes to function

Aerodrome is the only ve-model token in the dataset with positive 1Y return (+5%), and the mechanism depends on $BASE ecosystem inflows sustaining the bribe market. Velodrome, Curve, Balancer, and every smaller ve-fork printed -54% to -84%. The model creates a flywheel, but the flywheel needs continuous new capital. When that stops, the whole structure unwinds.

04

Direct fee distribution is the most legible but hardest to defend

The dYdX paradox is the tension here. Best-in-class IR, 100% of trading fees to stakers, 75% net revenue buyback, and down 82% on the year because revenue collapsed. The mechanism did what it promised. The business did not. For an allocator, this is the cleanest read: you are buying a share of protocol revenue, and if revenue drops, so does the token.

Hyperliquid is the inverse: buyback-and-burn via the Assistance Fund (99% of fees), zero traditional IR infrastructure, +193% on the year. Revenue dominates mechanism.

05

Everything else (points, RWAs, LRTs, memecoins) averaged -71%

The hybrid bucket is where most 2024–2025 launches ended up: EtherFi, Renzo, Puffer, Usual, Virtuals, AI16Z, the entire LRT cohort, the memecoin cohort. This cohort traded on narrative and TGE airdrops, not on cash-flow mechanics. Once the airdrop unlock printed, there was nothing left to defend the price with.

Framework

Which model for which stage

No single model wins. The right model depends on protocol stage, revenue scale, and institutional legibility needs. Use this as a decision matrix for treasury design.

Model Best for Institutional legibility Primary risk
Direct fee Mature protocols with >$10M annual revenue ★★★★★ Revenue visibility cuts both ways
Buyback-burn Protocols at scale with stable revenue base ★★★★☆ Destroys optionality, no floor
Buyback-hold Growth-stage with accumulating treasury ★★★★☆ Treasury becomes governance target
ve-model DEXs with active bribe economies ★★☆☆☆ Requires ecosystem subsidies to work
Governance-only Never. Retrofit a mechanism. ★☆☆☆☆ Pricing floor erodes over time
Other / hybrid Pre-TGE, points-era, narrative-led launches ★★☆☆☆ No post-airdrop defense
Universe

162 protocols, ranked

Full dataset filtered by accrual model. Performance data is 1-year price change via Artemis. Protocols without clean data are categorized but not ranked.

Model
1Y return
Sort
Search
# Protocol Model Sector 1Y % FDV ($M) Daily Rev ($K)
Closing

The protocol treasury's job

The market will not pay a premium for good mechanism design. It will punish the absence of any mechanism at all.

The cleanest empirical read of 2025 is that value accrual did not generate alpha. Revenue did. But the 48 governance-only protocols in this dataset show the cost of going without a mechanism. When the market has a choice between a token that pays you and a token that does not, it picks the one that pays.

The right question for a treasury is not which mechanism maximizes upside. The data says none of them reliably do. The right question is which mechanism makes this token investable from a fundamental lens of an institutional allocator. That lens rules out governance-only and hybrid categories immediately. It favors buyback-and-hold with transparent treasury disclosure, buyback-and-burn for protocols at scale with conviction about long-term supply reduction (Hyperliquid), direct fee distribution for mature revenue-generating protocols, and, for a narrow set of DEX-native tokens, a ve-model tied to a live bribe market.

For everything else, including most tokens launched in the last 24 months, the honest answer is retrofit a mechanism before your next unlock. Do it while you still have optionality.