Net Buyback Yield: When "Aggressive Buybacks" Are Just Dilution Theater

Why this matters now

Two of the most-discussed tokens in May 2026 have aggressive buyback programs and are still in active vesting periods. Both projects (correctly) tell their holders that buybacks are running. Both projects (correctly) disclose that more tokens are unlocking. Most market commentary stops at the first half of that sentence.

Hyperliquid (HYPE) hit a fresh all-time high of $62.18 on May 21, 2026 (the day before this report). Its 30-day rally is +45%. The headline narrative is "HYPE buyback machine is back." Underneath: 69.5M HYPE of core-contributor tokens unlock through year-end 2026, worth roughly $4.18B at $60.11. At the current run-rate of ~93% of fees flowing to buybacks, the protocol generates approximately $360M of buybacks through year-end. The ratio of insider unlocks to buybacks is roughly 11.6 to 1.

Pump Fun (PUMP) is the inverse setup. The token is ~80% below its September 2025 ATH ($0.0088 to $0.0018), and is being treated by some market participants as a low-conviction memecoin-infrastructure asset. Underneath: the team has committed to 50% of revenues for buybacks through the next year, the historical run-rate was actually closer to 93%, and 121.2B PUMP unlocks through year-end (cliff July 2026) worth roughly $219M at current price. Projected buybacks are $130-270M depending on which fee assumption you use. The ratio of insider unlocks to buybacks is approximately 1 to 1.

The headline buyback intensities are nearly identical (93%+ of revenue for both). The net buyback yields are not. That is the disclosure asymmetry this report is designed to surface.

Primary-source data verification

Per TokenIntel's Rule Zero, every number below was independently pulled from CoinGecko (price, supply, mcap) and DefiLlama (fees, revenue) on May 22, 2026. The unlock schedules are cited from project tokenomics docs as referenced in the originating DeFi Report comparative note. The fee numbers are live and the buyback intensities are observable, not projected.

*PUMP fees on the "pump" launchpad slug on DefiLlama. The "pump.fun" bonding-curve subset is smaller. Using launchpad-level aggregate to reflect the full protocol fee base. The DeFi Report's 90-day fee figure of $97M for Pump Fun sits between these two reads, suggesting our 30d snapshot is consistent with a slightly slower 90-day average.

The Net Buyback Yield comparison

To compute Net Buyback Yield, we project buybacks and insider unlocks through year-end 2026 (May 22 to December 31, approximately 7 months). For unlocks, we model three sell-through scenarios (25%, 50%, 100%) since unlocked tokens do not all sell immediately. For buybacks, we use current run-rates and disclosed buyback intensity.

Side-by-side

Why headline buyback yield is the wrong metric

Headline buyback yield = annualized buybacks divided by market cap. Reported buyback yields for both tokens look attractive in isolation:

• HYPE headline buyback yield: $620M annualized buybacks / $14.39B mcap = 4.3%. That looks like a high-quality cash-flow asset.
• PUMP headline buyback yield: $360M annualized buybacks (50% of revenue) / $643M mcap = 56.0%. That looks like an extraordinary recapture rate.

Both numbers are technically correct. Both numbers are misleading without the unlock side of the equation. The right question is not "how aggressively is the protocol buying its own token?" The right question is "after accounting for ongoing structural supply, is float actually shrinking?"

For HYPE, the answer at every sell-through level is no. The buyback program is real, sizable, and well-disclosed. It is also small enough relative to the unlock schedule that the protocol's circulating supply is going up, not down, in dollar terms. This is not a criticism of the buyback program. It is a recognition that buybacks at 4.3% of mcap cannot offset unlocks of 29% of mcap through year-end.

For PUMP, the answer at conservative sell-through is yes (net buyback positive). The disparity comes entirely from the unlock side. PUMP's insider allocation is a smaller absolute number in dollar terms because the token has already de-rated. Sophisticated holders who entered before the cliff face a worse arithmetic; the market today is pricing the cliff already.

Caveats and limitations

Not all unlocks sell

The 100% sell-through assumption is conservative. Core-contributor and investor allocations may be held, staked, or pledged as collateral rather than sold. In practice, observed sell-through after major unlock cliffs is in the 20-50% range over the 90 days following the cliff. This is why we present a range of assumptions rather than a single point estimate.

Demand-side dynamics matter

A token in a rally regime can absorb unlocks at scale (HYPE rallied 45% in the 30 days leading into a major unlock window). A token in a drawdown regime cannot. The NBY calculation captures structural supply pressure but does not capture demand elasticity. HYPE's current rally suggests demand is large enough to absorb significant unlocks at current levels. The NBY tells you the structural arithmetic, not the next 30 days of price action.

Buyback rates are not contractually guaranteed

HYPE's 93% buyback rate is observed (assistance-fund + protocol-buyback flow). PUMP's 50% buyback commitment is forward-looking from the team. Both can change. If HYPE's revenue compresses (e.g. from a perp-volume slowdown), the buyback compresses. If PUMP's revenue compresses, the buyback compresses faster relative to unlocks because unlocks are scheduled while buybacks are revenue-dependent.

Fee categorization matters

DefiLlama's "fees" and "revenue" for Hyperliquid are close in magnitude because most fees flow to the assistance fund. For Pump, the categorization is more sensitive to whether you count the bonding-curve issuance subset only or the full launchpad-plus-DEX aggregate. We used the full aggregate. A more conservative read using only the bonding-curve subset would scale PUMP's projected buybacks down by roughly 60%, putting PUMP at NBY of approximately +4% to +18% at 25-50% sell-through, still positive across both scenarios.

Unlock dollar values reprice with the token

HYPE's $4.18B unlock estimate uses $60.11 (today's price). If HYPE corrects 25% before year-end, the unlock value drops to approximately $3.14B, narrowing the dilution gap. Conversely, if HYPE continues rallying, the unlock value scales up. This is a partial natural hedge: bad-NBY tokens self-correct by drawing down. It does not change the relative ranking versus PUMP, which has already taken its drawdown.

Applying the framework to other tokens

Net Buyback Yield is a generalized concept. It applies to any token with both a buyback program and a vesting schedule. Quick qualitative reads on tokens in TI coverage:

• PENDLE: Team and investor allocations fully vested by September 2024 (per Pendle docs). NBY = headline yield (no unlock subtraction). At ~$20M annualized holder revenue / $325M mcap = ~6.2% NBY, fully captured. Cleanest of the de-rated tokens in coverage. See the PENDLE investment memo.
• ENA: Fee switch inactive (zero buybacks today) and approximately 5.97B ENA still to vest through 2028. NBY = negative by definition until the fee switch activates. See the Ethena investment memo.
• MKR/SKY: Endgame buyback mechanism is active and continuous. Sky's defined conversion roadmap manages the float dynamics deliberately. Worth a dedicated NBY computation. Likely positive but the calculation requires careful read of Endgame allocator framework.
• AAVE: Aavenomics buyback program is active ($1M weekly). 2025 emissions are largely behind. NBY likely modestly positive, deserves a dedicated read.
• AERO: Emissions-heavy model. Buybacks small relative to emissions. NBY likely negative.
• UNI: Buyback proposal in governance. If approved, would create a real NBY computation. Currently zero buybacks → NBY only depends on emissions which are zero (UNI is fully unlocked).

A full sweep of TI coverage at standardized NBY assumptions is a candidate for future expansion of the ERM framework. The point of this report is to introduce the concept and demonstrate that buyback intensity in isolation is a misleading metric whenever a token is still in active vesting.

Investment implications

The NBY framework does not by itself constitute a buy or sell signal. Several other considerations matter equally: demand dynamics, narrative momentum, revenue durability, competitive positioning, regulatory risk. NBY is a single input to a multi-input decision.

That said, the headline number is striking enough to flag for HYPE specifically. At -4.9% to -26.7% NBY through year-end, depending on sell-through assumption, the structural supply pressure is real. The recent rally to all-time highs has absorbed unlocks so far because demand has been strong. If demand softens (e.g. perp-trading volumes normalize from current cycle highs), the unlock pressure becomes visible and the same buyback program that looks impressive today will be revealed as insufficient to maintain the float.

For PUMP, the NBY framing supports an asymmetric long-bias case. At +15.7% to +24.1% NBY in base and conservative scenarios, the protocol is actively shrinking its float. Combined with the cheap P/R multiple (0.7x at the launchpad-level fee aggregate) and the fact that the price has already absorbed the upcoming cliff, the setup is the kind of compressed-multiple-plus-positive-NBY combination that has historically produced strong forward returns in DeFi tokens. This is a framework observation, not a directional recommendation; PUMP-specific risks (revenue cyclicality, bot-inflated user metrics, competitive landscape) are separate considerations beyond NBY.

TI standard going forward

TokenIntel will compute and surface Net Buyback Yield alongside headline buyback yield on any research page where both an active buyback program and an active vesting schedule are present. The first applications are:

• Hyperliquid (HYPE) research page: NBY callout added in the Tokenomics section flagging the unlock-vs-buyback math.
• Other coverage tokens (HYPE, AAVE, AERO, UNI, JUP, MKR/SKY): NBY computation added to each ERM-eligible research page over the next several refreshes.
• Monthly ERM refresh cron: extended to compute and surface NBY alongside ERM. Source: data/erm-config.json + data/erm-state.json.

This is a methodology adoption, not a one-off report. Every token-buyback claim should be evaluated against the unlock side of the equation. The headline figure alone is insufficient.