HIP-5: Staker Governed Assistance Fund 2

Abstract

This Hyperliquid Improvement Proposal (HIP-5) introduces a Staker Governed Assistance Fund 2, allocating 5% of protocol trading fees (1% to start with, growing to 5% over time) to buy back tokens from Hyperliquid's Strict List (e.g., HYPE, PURR, HFUN).

It evolves the AF-1 framework through the allocation to eligible tokens in real time via governance votes from HYPE stakers, creating a dynamic "gauge voting" system for fee distribution. This extends the existing Assistance Fund mechanism, which currently directs 99% of fees to HYPE buybacks, by diverting a modest portion to support trusted ecosystem projects without undermining HYPE's deflationary pressure.

As Hyperliquid's ecosystem expands with projects like Unit ($930M TVL) and Kinetiq ($2.3B TVL), this fund fosters network effects, incentivizes builders, and enhances HYPE's long-term utility as a governance token. At recent fee levels ($91M in September 2025), a 5% allocation yields $55M annually for targeted buybacks, while 95% of fees ($86.5M monthly) continue fueling HYPE purchases. Implementation consists of continuous automated buy executions mirroring AF1’s HYPE mechanism and transparent on-chain governance to manage allocations. This balances short-term HYPE support with sustainable ecosystem growth.

Motivation

Hyperliquid's explosive growth has been driven by a robust tokenomic model where the vast majority of trading fees (99%) are automatically funneled into HYPE buybacks via the Assistance Fund. This creates sustained buy-pressure, reducing circulating supply and aligning protocol success with token holders.

However, as the ecosystem matures beyond the core exchange with protocols building on HyperCore and HyperEVM (e.g., lending, liquid staking or asset onboarding), Hyperliquid may not efficiently support projects that strengthen its own ecosystem. These include Strict List tokens like HYPE, PURR or HFUN, launched via builder auctions and vetted for quality. Many protocols contribute to the overall success of Hypercore, but may not generate significant revenues themselves.

Supporting them via controlled fee sharing can:

1. Bootstrap Ecosystem Projects: Provide buy-pressure to existing strict spot Hypercore tokens, giving teams already staking HYPE more incentive to deploy their tokens in Hypercore. At the same time, new projects launching on Hypercore are encouraged to stake HYPE (or incentivize stakers) in order to compete for votes and buybacks, gaining both liquidity and governance exposure that reinforce long-term builder alignment and position HYPE staking as an important component supporting the Hypercore ecosystem. This makes Hyperliquid a stronger place to build by helping long-term aligned projects bootstrap liquidity while tying together network security, community participation, and builder incentives through HYPE staking and Hypercore integration.
2. Enhance HYPE Resilience: Diversify value accrual beyond pure exchange revenue, adding governance utility and a treasury of appreciating assets that feed back to HYPE holders.
3. Boost Community Engagement: Empower HYPE stakers to direct allocations, mirroring successful DeFi gauge systems (e.g., Curve) while maintaining HYPE's core incentives. The governance UI/UX may be used for other governance decisions in the future.
4. Preserve Competitiveness: A 5% diversion is negligible (~$150K daily reduction in HYPE buys) compared to the ~$3M baseline, yet this small trade-off still provides millions monthly to support growth that’s better aligned with the community than competitors’ approaches.

Without this implementation, Hyperliquid’s native spot tokens risk facing thin liquidity and reduced trading activity after deployment on HyperCore. With it, fees reinforce a virtuous cycle of deeper liquidity, higher volumes, and create another HYPE sink.

Specification

Fee Allocation and Fund Mechanics

Allocation Range: In the medium term, 5% of total trading fees will be directed to the Assistance Fund 2, while 94% will continue to flow to the original Assistance Fund for HYPE buybacks. The targeted 5% of buybacks will be progressive, starting at 1% and subject to community/validator ongoing review.

HYPE stakers will be able to vote to allocate funds toward Strict List tokens or, if they prefer, to HYPE itself, ensuring flexibility if the community believes a higher share should remain allocated to HYPE.

Eligibility: Limited to Strict List tokens and native Hyperliquid Assets  (trusted assets like HYPE, PURR, HFUN).

Strict List Governance: The Strict List defines which tokens are eligible for AF-2 participation. It exists to uphold ecosystem integrity by ensuring that only trusted, verified, and mission-aligned projects can receive support. Currently, the Strict List is managed by the Foundation, serving as a safeguard against misuse and low-quality proposals. However, this governance structure is not fixed, proposals may be introduced to delegate management of the Strict List to validators, enabling a more decentralized and transparent process for determining eligibility (incl. tokenomics, team vestings, audits, etc) over time.

The Strict List should be carefully revised each time the AF-2 allocation percentage is adjusted and, once the full 5% allocation is active, it should be formally reviewed on a quarterly basis.

New launches may be added directly to the list at the Foundation’s discretion, provided they meet the established criteria and maintain alignment with Hyperliquid’s long-term ecosystem objectives.

Execution: Fees accrue in USDC (or equivalent) in an on-chain AF-2 address, and HYPE stakers vote on how to allocate them. The collected fees are then split proportionally and executed on a continuous basis to minimize market impact, with votes remaining active until updated by the voter. There are no lockups, allowing agile adaptation to new projects.

Governance

Voting Power

AF-2 allocations will be determined through direct on-chain voting by individual HYPE stakers. Each staker’s voting power is proportional to the amount of HYPE they have staked, ensuring that governance outcomes directly reflect the preferences of those economically aligned with the protocol. This model maximizes decentralization, transparency, and accountability compared to validator-mediated systems.

Voting Process

Stakers can allocate their voting weight across any eligible Strict List tokens (e.g., HYPE, PURR, HFUN) in real time, adjusting their preferences at any moment. There are no fixed epochs, allowing the distribution of AF-2 funds to continuously adapt to changing ecosystem conditions.

Only explicitly cast votes are counted in the final allocation; there are no default votes.

Upon deployment, there will be a 7-day initialization window during which users can submit their votes before AF-2 begins executing any fee redirections. This ensures that the first allocation cycle accurately represents community intent from the outset.

LSD Participation

Liquid Staking Derivative (LSD) protocols are excluded from voting by default to prevent centralized influence and strategic advantage over other ecosystem projects. LSD eligibility requirements and safeguards are described in detail in the Risks and Mitigants section, outlining how only protocols with transparent, holder-based governance can participate in AF-2 allocations.

Simulation Methodology

1. Data Collection and Aggregation

Raw trade-level data and mid-price snapshots were extracted from Hyperliquid’s spot and perpetual markets. Each trade record included timestamp, notional value, and trade direction (buy/sell). Prices were sampled at high frequency and aggregated into 30-minute buckets to form consistent time intervals for analysis.

2. Flow and Return Construction

For each bucket, the net notional flow was computed as the signed sum of all traded volume (positive for net buys, negative for net sells). The mid-price return per bucket was calculated as the percentage change in mid-price relative to the previous bucket. This produced a time series linking net order flow to contemporaneous price changes.

3. Outlier Filtering and Normalization

To reduce noise, extreme flow observations were filtered using z-scores (|z| < 3), and all variables were standardized. This step ensures that a few unusually large trades do not distort the estimated impact coefficient.

4. Buy-Adjusted Price Modeling

We then estimated a linear impact model of the form:

where rt​ is the log-return of the mid-price (in %), ft​ is the net USD flow in bucket t, and β represents the instantaneous price-impact coefficient, that is how much the price moves per USD of net buying. The model fit was evaluated using R2 and correlation between flow and returns.

5. Programmatic Buyback Simulation

Using the estimated coefficient β, we simulated a continuous buy program equivalent to redirecting ~3.3% of protocol fees (≈$100K per day) into AF-2 purchases over 30 days. This figure is based on recent protocol activity, with September revenue reaching approximately $91M (according to DefiLlama), or about $3M per day-making the simulated flow representative of current operating scale.

These simulated buy orders were distributed evenly across buckets (TWAP-style) to emulate gradual execution. The cumulative predicted log-returns were then converted into an adjusted price path showing the incremental price effect of AF-2 activity.

6. Visualization and Scenario Analysis

We generated side-by-side charts comparing the observed price trajectory (“before model”) and the simulated buy-adjusted trajectory (“after model”) over one week. The difference between both paths quantifies the model-implied uplift in price due to continuous AF-2 buy pressure.

This approach offers a transparent, data-driven estimate of how a modest reallocation of trading fees could reinforce liquidity and long-term token appreciation without requiring speculative assumptions about order-book elasticity or external hedging behavior.

However, this methodology should be viewed as an initial empirical approximation rather than a definitive model. The linear relationship between net flow and returns across time buckets is clear and intuitive, but the absolute magnitudes such as monthly price changes exceeding 100% under simulated $100K/day buy programs suggest the model may overstate sensitivity to flow due to simplifications in liquidity depth, execution granularity, and market feedback effects. Future refinements could include non-linear impact modeling, dynamic liquidity adjustments, or integration of real order-book depth data.

Overall, this represents a solid first step toward quantifying AF-2’s potential market impact and provides a foundation for more precise simulations as richer data and execution models become available.

Simulation Outcomes

PURR

The simulation for PURR compares the observed price path against a modeled trajectory incorporating a steady $100K/day (around 3.3% of daily revenue when the data was collected) buy program. As shown in the chart, the model-implied price path (orange) exhibits a clear upward shift relative to the observed data, reflecting consistent buy-side support. This directional consistency confirms the linear relationship between cumulative net flow and log returns previously observed.

The simulation results in a ~30% monthly price increase. While the model effectively captures the direction of flow-driven price movement, it likely underrepresents market depth and offsetting liquidity provider behavior.

HYPE

For HYPE, the simulation shows that reducing buyback flow by approximately $100K/day to redirect it toward AF-2 has minimal observable effect on price dynamics. The simulated path remains nearly identical to the observed one, indicating that HYPE’s market depth effectively absorbs this scale of change without meaningful deviation. This reinforces that a 2–5% fee reallocation would not materially weaken HYPE’s buyback pressure, while still generating significant capital for ecosystem growth.

These simulations provide a simplified, illustrative view of how continuous AF-2 buybacks could influence long-term price trajectories under steady conditions. In practice, short-term market behavior would likely diverge, as participants reprice the ecosystem to reflect a new competitive dynamic where multiple Strict List assets actively compete for a share of future Hyperliquid revenue.

Risks and Mitigants

HYPE Dilution

The primary concern is a potential reduction in buyback volume for HYPE due to the <5% fee reallocation. However, simulations indicate that a ~$100K/day reduction in buy flow produces negligible impact on HYPE’s price trajectory, as its market depth efficiently absorbs this adjustment. Moreover, HYPE stakers retain the option to direct AF-2 votes toward HYPE itself, ensuring that governance can dynamically rebalance allocations if market conditions warrant additional support.

Fund Misallocation

AF-2 is restricted to Strict List tokens (assets that have already undergone Hyperliquid’s vetting and builder auction process) minimizing the risk of low-quality or exploitative projects receiving buybacks. On-chain governance by HYPE stakers provides transparent oversight, and optional incentive mechanisms such as bribes can further align voter interests with long-term ecosystem value creation.

Low Participation

A potential risk is insufficient voter engagement in AF-2 governance. To address this, votes are persistent: once submitted, they remain active indefinitely until updated, eliminating the need for epochs or fixed voting periods. Even if only 10% of token holders vote, the system still runs at 100% efficiency, ensuring full fund utilization regardless of participation levels. To further encourage engagement, third parties may introduce optional incentives such as bribes.

Market Impact and Volatility

AF-2 purchases are executed via continuous TWAP-style algorithms, distributing flow evenly through time to minimize slippage and volatility. Real-time allocation updates further ensure that liquidity conditions and token weights adjust smoothly, preventing abrupt market distortions or excessive short-term price swings.

LSD Governance Concentration

A potential risk of AF-2 governance is the disproportionate influence of LSD protocols, which hold significant amounts of delegated staked HYPE and could redirect buybacks toward their own preferences. To prevent this, AF-2 implements strict participation requirements:

• Default Exclusion: All LSD addresses are blacklisted from voting unless they have implemented a verifiable, holder-based governance framework that allows individual LSD holders (not token holders or operators) to determine how their underlying staked HYPE votes are cast.
• Conditional Reinstatement: Once such a governance system is live and transparent, the LSD may regain voting eligibility after the approval of validators.
• Rationale: This enforcement ensures that influence over AF-2 allocations remains distributed among genuine, independent stakers, preventing recursive self-funding loops or vote capture by intermediaries controlling large pools of stake.

While the current framework prevents direct governance influence by LSD protocols lacking verified holder voting, indirect influence through vote incentives, points systems or bribe-style mechanisms remains a potential vector.

These dynamics are expected to evolve naturally as AF-2 matures; future proposals may address them through disclosure standards, cap mechanisms, or incentive alignment rules to ensure that vote-based reward structures remain transparent and consistent with long-term ecosystem health.

Other Approaches

Validator Voting

Another viable design is to restrict buyback voting to active validators. This channels governance through the entities that secure the network and ensures there is no idle staked HYPE participating, while introducing distinct trade-offs that are worth highlighting.

A validator-only model would determine buybacks exclusively through active validators, with each validator’s influence proportional to the HYPE delegated to them, including any self-bond. In practice, users don’t vote directly, but express preferences by choosing which validator to back, and validators cast the buyback votes on their behalf.

Implementation

This approach is simpler to implement than individual voting because it leverages the governance surface that already exists around validators rather than introducing per user voting logic.

Participation 

There would be more participation, as every staked HYPE ends up attached to a validator that is more used to voting than an individual.

LSD Dilema

Stake held via LSD protocols should be excluded from voting. Without a holder-level, verifiable passthrough mechanism, LSDs can concentrate influence behind a few operators, making outcomes less representative of end users.

Backroom Deals

This model increases the risk of backroom deals: a project could pay or otherwise incentivize validators to steer buybacks toward that project’s token. With a smaller voter set, collusion and vote-buying become easier to coordinate and harder to disprove.

Freedom

It also reduces user freedom. Instead of expressing preferences directly, users must pick between staking with their preferred validator or delegating to the validator whose buyback votes they like the most, forcing a trade-off between operational trust and economic alignment. It also forces the users most aligned with network security (those who stake the most) to keep monitoring how their validators vote, which can be tiring. Additionally, if a validator changes their vote and the delegator dislikes it, they are forced to wait 1 day until their stake moves to a different validator.

Rising fee redirection

Phased fee redirection can be adopted instead of jumping directly from 0% to 5%. The program would start at 1% at the moment of implementation and then step up in fixed intervals (e.g. every 2 months) until reaching the 5% target. This gradual ramp provides predictability, reduces shock to ecosystem flows, and gives participants time to adjust and keeps the final objective unchanged.

HIP-5 Implications

If this governance proposal passes and future upgrades refine the voting framework it will not only make Hyperliquid a more attractive place to build, but also unlock new business models such as:

Convex-style vote aggregators:

Projects offer bribes to attract votes and the aggregator allocates its consolidated votes to maximize depositor yield, channeling a larger share of the AF2 buyback budget into targeted tokens.

Traditional bribes pay LPs to deepen pools, which often leads recipients to sell the rewards to realize profit, creating sell pressure. In contrast, AF2-directed bribes steer programmatic buy pressure, resulting in direct demand rather than rented liquidity.

In short, as long as AF2 buys more tokens than the project gives away in bribes, the project benefits on a net basis.

Improved LSD Cash-Flow Mechanics under HIP-5

LSDs built on Hyperliquid could unlock a new value stream once HIP-5 is active. Eligible LSDs can participate in AF-2 and align their incentives with the broader ecosystem.

The most powerful mechanism enabled by this structure is the buy-and-burn loop. LSD protocols could use AF-2-driven incentives to buy back and burn their own LSD tokens, simultaneously redeeming and burning the underlying HYPE. This process permanently reduces the circulating supply of both the LSD token and HYPE itself, while amplifying yield and value for LSD stakers.

Additional unlocks

These examples only scratch the surface; HIP-5 can enable new coordination models and value loops across builders, stakers, and validators, laying the groundwork for long-term ecosystem growth.

Conclusion

HIP-5 proposes a prudent evolution to what AF-1 already started: redirecting 5% of protocol fees to a stake governed Assistance Fund 2, preserving HYPE’s buyback engine while seeding tomorrow’s ecosystem growth. With community approval, AF-2 can be deployed swiftly to strengthen builder incentives, deepen liquidity, and expand Hyperliquid’s economic moat.

Disclosure:

From the Strict List, apart from HYPE, the proposers hold PURR primarily, HFUN secondarily, and other Strict List assets residually. No contributors hold material exposure that would constitute a conflict of interest under the current proposal.

Contributors:

This proposal was authored by @Ericonomic (Supercexy), @CalebAndersDev (Flowdesk), @CFrugho, @0xHakai_ (Three Sigma), @FSobrini (Chorus One), and @HansonBirringer (Flowdesk). Additional feedback was gathered from long-time Hyperliquid community members.