Hi,
Aragon is happy to present our findings to the CoW community for discussion regarding the above RFP, commissioned by CoW Swap for the research and design of a value accrual mechanism for COW. Listed below are the full set of deliverables, along with an executive summary of our recommendations.
Full Deliverables
Summary of Findings
Make the existing value flow legible
At CoW Protocol’s current revenue scale, no change to the value accrual mechanism can have a meaningful impact. Buybacks funded solely by free cash generated by the protocol, under any allocation policy, are too small relative to circulating supply and structural emissions to materially shift the supply curve. The cash flow is too thin to generate a competitive staking APR under distribution models. The constraint CoW Protocol faces today is scale, not design.
While scale is a challenge, CoW is generating free cash, running consistent buybacks, and has materially reduced structural sell pressure (legacy vesting completed in February 2026, replaced by a much smaller allocation as per CIP-83). This fundamental strength is not presented in an easily digestible, widely accessible format.
Formalising and surfacing the revenue waterfall is the low-hanging fruit
Gross protocol fees → variable costs (solver rewards, gas reimbursement, integration fees) → fixed costs (OpEx) → capped buyback.
This is already what happens, but across disjointed wallets and processes. For example, excess COW buybacks beyond the solver incentive budget are funded by cash flow net of OpEx. As OpEx is separately funded by the treasury, this flow is not immediately obvious to external audiences. Extending the existing reporting or providing dashboard data that clarifies cash flows would deliver more market signal than any change to the mechanism at the current scale.
We recommend this be done in two parts:
-
Aragon will propose a set of best practices for communicating value accrual, based on best-in-class examples, as part of the communication plan deliverable
-
We recommend segmenting the value-accrual cash flows into a separate set of contracts, in which additional mechanics can be activated at the DAO’s discretion. Details are provided in the implementation deliverable.
Adjustments in buyback policy and evaluation of buyback-and-distribute
Based on our research findings, we recommend a formal reassessment triggered when annualised free cash flow consistently reaches ~$9M/year, which we project will occur in approximately 18 months under conservative growth assumptions. Two distinct changes should be evaluated at that point:
-
Shift from emissions-linked to surplus-linked buybacks.* Change the current 1.2x solver emissions target by delineating solver emissions replacement buying from the value accrual portion. Use a fixed share (e.g., 80%) of FCF after OpEx as a buyback target, scaling value accrual with revenue growth rather than capping it at solver costs.
-
Consider implementing buyback-and-distribute.* Routing a portion of buybacks to stakers as a displayable APR. At $9M/year, FCF buybacks, net of solver rewards, exceed structural emissions. Furthermore, a staking locker could provide leverage on the amounts mechanically taken out of circulation by the buybacks alone
Reassessment is relatively straightforward, as the simulation model has already been developed as part of the research presented here and requires minimal adjustments.
Evidence Base
Benchmarking: 11 protocols, three consistent patterns
We evaluated 1inch, Paraswap, Jupiter, Hyperliquid, Sky, Ether.fi, Aerodrome, GMX, Aave, Uniswap, and Lido against criteria derived from CoW community sentiment: revenue-funded, visible, simple, solver-neutral.
Escrow mechanisms fail on aggregator margins. 1inch (8% stake rate, resolver-rev-share) and Paraswap (mechanism abandoned) are CoW’s closest structural analogues and clear failure cases. VE and social escrow cannot be sustainable within DEX aggregator economics without sizeable inflation emissions or a degradation of solver competitiveness.
Buybacks do not generate a sufficient signal at a moderate scale. Jupiter spent $70M in buybacks against $1.2B in unlocks, covering 6% of new supply. Aave is slightly deflationary but is still underperforming ETH. Buyback-and-hold preserves optionality but produces no holder-visible value story.
Distribute models are not completely circular thanks to the staking of rewards. Sky (72% restake) and GMX (70% restake) show that only a fraction of bought-back tokens being distributed to stakers are reintroduced into the circulating token supply. These effects were observed at double-digit staking APRs.
Model findings
A weekly Monte Carlo simulation over a 104-week horizon was run against three stochastic inputs (COW price, weekly net revenue, solver share of revenue) and deterministic emissions schedules (CIP-83 core team vesting at ~1M COW/month, CIP-82 Grants DAO’s grants at ~125K COW/week). Each scenario was run across multiple sample paths to produce median, 50% and 90% outcome ranges.
Stochastic input ranges were calibrated from observed data: COW price modelled as geometric Brownian motion using historical weekly volatility; weekly net revenue sampled uniformly across $300K–$700K range (based on a post December 2025 fee structure change); solver share sampled at 40–60% of net revenue (centred on the ~50% historical mean). Policy levers tested included buyback target (current 1.2x emissions vs. surplus-linked at 50–80% of FCF), distribution split (0% vs 100% of buybacks routed to stakers), and revenue trend (0% vs +40% CAGR).
The discretionary buyback programme has limited impact on circulating supply at current revenue levels. Cumulative value-accrual spend has a median of $2.5M at 0% CAGR and $5.5M at 40% CAGR over two years. In the most aggressive scenario modelled (40% CAGR, Buyback and Hold), median circulating supply ends close to its starting level. Vesting and grants add about 3.5% inflation per year, so the supply trajectory is broadly stable.
Staking yields funded by buybacks require both revenue growth and moderate participation to reach competitive levels. Under 40% CAGR, the median APR crosses 10% at week 16 with 10% participation, week 52 with 20%, week 80 with 30%, and does not cross within the two-year horizon at 50%.
In all cases, the treasury remains well funded across the modelled scenarios and capable of supporting value accrual, provided that investment in growth translates into a higher revenue CAGR. Of course, if operating costs rise, the free-cash-flow threshold required for meaningful value accrual rises with them - and this will push the time needed before protocol revenue growth will be sufficient to support either model.