Overview
TokenLogic, in close collaboration with Aave Labs, has developed the first LiquidityBufferController for Aave V4. It enables a portion of the idle USDC held in Aave V4 USDC reserves to be deployed to Circle Gateway to generate yield. The initial deployment is proposed for the Ethereum Core Hub.
This publication explains how the LiquidityBufferController works, details its configuration and presents our recommended initial parameters for the Ethereum Core Hub USDC reserve.
Motivation
Aave borrowers pay interest only on the USDC they borrow. When utilization is low, a significant portion of the USDC supplied to the Hub remains idle and generates no yield.
Aave V4 addresses this inefficiency through its liquidity buffer feature, which allows a LiquidityBufferController to sweep a portion of idle liquidity into an external strategy. Importantly, swept USDC continues to count as supplied liquidity, meaning that utilization and borrowing rates remain unchanged.
However, swept USDC is not immediately available for withdrawals or borrowing until it is returned to the Hub. Therefore, the underlying strategy must offer exceptionally high liquidity and support near atomic redemptions.
This is the primary reason for selecting Circle Gateway. USDC deposited into Gateway can be redeemed 1:1 for USDC almost instantly, without slippage, making it well suited to Aave’s liquidity requirements. Withdrawals from Gateway are processed one at a time, with each one settling between Circle and the chain before the next is sent, so returning large amounts to the Hub happens in stages rather than all at once. The controller’s parameters are sized with this pacing in mind, as described in the following sections.
Circle Gateway also keeps the credit exposure for suppliers unchanged. USDC supplied to the Hub already depends on Circle as the issuer, and a Gateway position is a claim on USDC from the same issuer. Sweeping into Gateway therefore introduces no new counterparty credit risk. The additional risks are operational, covering Circle’s attestation service, the keeper and the Gateway contracts, together with the timing of liquidity returning to the Hub. The parameters in this proposal are designed to manage these risks.
Liquidity Buffer Controllers in Aave V4
Aave V4 lets each Hub asset have a LiquidityBufferController. Governance sets it in the asset’s configuration.
The controller is the only address that can take idle liquidity out of the Hub and return it, through two Hub functions:
sweep(assetId, amount)moves idle liquidity from the Hub to the controller.reclaim(assetId, amount)returns it.
The Hub records the amount moved out in a separate swept balance. When the controller calls sweep, the amount moves from the Hub’s idle liquidity to swept, and reclaim moves it back. Utilization is calculated as
so swept funds still count as supplied liquidity in the denominator. Moving funds between liquidity and swept therefore leaves utilization and borrow rates unchanged.
The difference is what can actually be used. Only liquidity can be withdrawn or borrowed, so swept USDC is unavailable to users until the controller reclaims it. This is why the controller has to keep enough USDC idle on the Hub, and recall funds quickly when that idle balance runs low.
LiquidityBufferController
The LiquidityBufferController sits between the Ethereum Core Hub and Circle Gateway. An off-chain keeper bot decides when to move USDC. The contract limits how much can be moved, and USDC can only follow one route: Hub → Gateway → Hub. The keeper can’t send funds anywhere else.
Sweep Excess
Once a day, if idle USDC on the Hub is above the liquid buffer (liquidBufferBps) by at least MIN_SWEEP_AMOUNT, the keeper calls sweepExcess. Before moving any funds, the controller checks three limits:
- Liquid buffer (
liquidBufferBps): idle USDC can’t fall below this share of supply after the sweep. This keeps USDC on the Hub for withdrawals and borrows. - Exposure caps (
exposureCapAbsandexposureCapBps): the total held in Circle Gateway can’t exceed a fixed amount or a share of supply, whichever is lower. - Minimum delay (
sweepMinDelay): This limits how often funds can move.
If all three checks pass, the controller sweeps the excess from the Hub and deposits it into Circle Gateway in the same transaction. Idle USDC is brought back down to the liquid buffer (liquidBufferBps).
Replenish Buffer
The keeper checks the Hub every 5 minutes. When idle USDC falls below the buffer floor (BUFFER_FLOOR_BPS), it starts recalling funds from Circle Gateway. Each recall goes through Circle’s attestation service:
- The keeper signs a burn intent for the amount to recall.
- Circle checks the intent against the controller. The signer must hold the keeper role, and the amount plus the maximum fee can’t exceed what the Hub has swept. Circle then returns an attestation.
- The keeper calls
replenishBufferwith the attestation. The controller mints the USDC from the Gateway and returns it to the Hub through reclaim.
Two settings shape each recall:
- Maximum recall size (
MAX_REPLENISH_AMOUNT): each recall is capped at 10M USDC. The keeper recalls whatever is needed to bring idle back to the liquid buffer (liquidBufferBps), up to this cap. - Recall pacing (
MIN_REPLENISH_INTERVAL): there is a delay of about 30 minutes between two consecutive recalls. After each recall lands, the keeper waits for Circle’s records and the chain to agree before it sends the next one. This is a safety check against Circle API edge cases. The keeper keeps recalling, one recall every 30 minutes, until idle USDC is back at the liquid buffer (liquidBufferBps).
The keeper also recalls if the Gateway balance is ever above the exposure caps (exposureCapAbs and exposureCapBps), for example after supply has fallen.
The buffer floor (BUFFER_FLOOR_BPS) and the liquid buffer (liquidBufferBps) create a dead band. Recalling all the way back up to the liquid buffer, rather than just above the floor, batches withdrawals into a few larger recalls.
With a single threshold, every small withdrawal right after a refill would trigger another recall. Each of those recalls would also start a new MIN_REPLENISH_INTERVAL wait, so a burst of withdrawals would be served by a slow chain of small recalls, each about 30 minutes after the last. Recalling up to liquidBufferBps leaves headroom for those withdrawals without waiting for the next recall.
If Circle’s attestation service is unavailable, recalls can’t run. In that case an admin can pause the controller and use the on-chain withdrawal, which doesn’t need Circle but takes about 7 days. Pausing stops new sweeps and recalls, while the withdrawal path stays open.
Parameters and Risk Management
The controller’s parameters cover four distinct risk surfaces, and they sit on two layers.
On-chain parameters are stored in the LiquidityBufferController contract and enforced by it. Changing them requires the admin role, so every change goes through governance.
Off-chain parameters are part of the keeper bot’s configuration and decide when the keeper acts and how much it moves. The contract doesn’t enforce them, but every action they trigger is still bounded by the on-chain limits.
The four parameters (liquidBufferBps, BUFFER_FLOOR_BPS, exposureCapAbs, exposureCapBps) sized in this ARFC depend on how USDC moves on the Hub, and the analysis below sets their values. The rest are fixed by the controller’s design.
- Liquidity before the first recall lands
The keeper checks the Hub every 5 minutes. Consequently, when idle USDC falls below the buffer floor, the first recall is initiated within 5 minutes, with an average detection time of 2.5 minutes.
For sizing, the analysis assumes the worst case on both counts: idle liquidity starts exactly at the buffer floor, and the full 5 minutes pass before the first recall. Until that recall lands, withdrawals and borrows must be fulfilled using only the USDC remaining on the Hub.
BUFFER_FLOOR_BPS (off-chain): determines the minimum level of idle USDC available on the Hub before the first recall is initiated. It must be sufficient to absorb outflows during the worst case detection and recall period.
- Liquidity while a recall is in progress
A large exit can take several recalls of up to 10M USDC each, about 30 minutes apart. Outflows continue in between, and the USDC refilled by earlier recalls has to absorb them.
liquidBufferBps(on-chain): sets how far each recall refills, and so how much idle USDC carries the Hub through the 30-minute wait between recalls. It is also the floor that sweeps can’t go below.MAX_REPLENISH_AMOUNT(off chain, 10M USDC): constrained by the issuer and determines the maximum amount that can be recalled in a single transaction.MIN_REPLENISH_INTERVAL(off chain, 30 minutes): fixed by design and determines the minimum interval between consecutive recalls.
Because the maximum recall amount does not scale with supply, each recall covers a smaller share of total supply as the Hub grows. The liquidity parameters should therefore be reviewed at relevant supply milestones.
- Counterparty and exit risk (Circle)
Every USDC in Circle Gateway depends on Circle. In the worst case, Circle’s attestation service is unavailable and the only exit is the on-chain withdrawal, which takes about 7 days.
exposureCapAbs(on-chain): the most that can be in Circle Gateway, as a fixed amount. It caps the delay the Hub could face.exposureCapBps(on-chain): the same cap as a share of supply. It shrinks the exposure automatically if supply falls, and the keeper recalls any excess.
- Operational and keeper risk
The keeper is an off-chain bot. The contract limits what a faulty or compromised keeper can do: funds can only move Hub → Gateway → Hub.
sweepMinDelay(on-chain, 1 day): limits how often funds can be swept, so a faulty keeper can’t keep moving funds back and forth.MIN_SWEEP_AMOUNT(off-chain, 100,000 USDC): stops small sweeps whose gas and fees would outweigh the yield.MIN_REPLENISH_AMOUNT(off-chain, 10,000 USDC): the smallest recall the keeper sends to bring the Gateway balance back under the exposure caps, which avoids recalling dust. Recalls triggered by low idle liquidity are always sent in full, up to the Gateway balance, regardless of size.- Pause and emergency exit (on-chain): the pauser or the admin can halt all sweeps and recalls at any time. While paused, the admin can still pull all funds back from Circle Gateway through the Gateway Wallet’s on-chain withdrawal. This needs no attestation from Circle’s API, but it is subject to a withdrawal delay set by Circle, currently 7 days (50,400 blocks).
Interest Rate Response Under Delayed Redemptions
Because swept USDC continues to count as supplied liquidity, reported utilization reflects the Hub’s total economic supply rather than the liquidity physically available on the Hub. As a result, if Gateway holds a portion of total supply and all remaining idle USDC is consumed, reported utilization can remain below 100%. For example, with 10% of supply held in Gateway, the Hub could have no idle USDC remaining while reported utilization stands at only 90%. This is below the Core Hub USDC reserve’s 92% uoptimal. Consequently, borrow rates would not reach the steep slope above the kink, weakening the usual interest rate signal that incentivises repayments and additional supply.
Under normal conditions, this is unlikely to be material. Gateway redemptions are near atomic, and the keeper initiates recalls as soon as idle USDC falls below BUFFER_FLOOR_BPS of supply.
The risk becomes relevant if recalls are delayed or Circle’s attestation service becomes unavailable. In that scenario, Gateway funds can only be recovered through the onchain withdrawal process, which takes approximately 7 days.
If recalls prove materially slower than expected, or Circle’s attestation service becomes unavailable, the IRM may need to be adjusted to account for the reduced liquidity available on the Hub. One option would be to lower uOptimal by the share of total supply held in Gateway, so that borrow rates better reflect the liquidity effectively unavailable to users.
Ethereum Core Hub USDC: Liquidity Profile
Supply and idle liquidity
USDC supply on the Ethereum Core Hub increased from $9.0M to $59.3M over the past three months, reaching a peak of $65.1M on 29 October. During this period, idle USDC averaged 9.2% of total supply, equivalent to approximately $2.3M. Idle USDC was above 10% of supply in 29% of hours.
Supplier and Borrower Concentration
The 10 largest suppliers hold 32% of USDC supplied, and the 10 largest borrowers hold 44% of USDC borrowed. The largest supplier alone holds 13% ($6.9M), more than the Hub usually keeps idle.
Concentration has stayed in a stable range since August, apart from two short spikes. Each was a single address building a USDC leverage loop: depositing USDC, borrowing USDC against it and depositing the borrowed USDC again over several rounds. Because a loop adds to supply and borrows at the same time, only the address’s own capital (about $2–3M) actually leaves the Hub when it unwinds. The unwind still produces sharp outflows, though: each withdrawal leaves before the next repay comes back. On 17 September a single $2.07M withdrawal took 11.9% of supply.
Worst-Case 30-Minute Liquidity Drawdowns
The controller’s job is to keep enough USDC idle on the Hub to cover withdrawals and borrows until recalled funds arrive. The keeper takes up to 5 minutes to notice that a recall is needed, and consecutive recalls are about 30 minutes apart. The key question is therefore how much USDC can leave the Hub over a short period.
To measure this, we treat every withdrawal and borrow since 1 August as a possible start of a liquidity squeeze. There are 6,949 of these start points. From each one, we follow every flow in and out of the Hub for the next 30 minutes and track the running net outflow. We record its deepest point, as a share of supply at the start.
The maximum cumulative liquidity shortfall is used rather than the net outflow over the 30 minute window, as idle USDC must be sufficient to cover the worst point in time. A repayment received later cannot offset a withdrawal that has already failed due to insufficient liquidity.
In the example above, although the Hub ends the 30 minute window with a net outflow of $1.0M, the cumulative outflow reaches $1.5M at its peak. Therefore, 3.0% of supply is the value recorded.
Most outflows are quickly offset: three quarters of start points end the 30 minutes with more USDC back on the Hub than left it. The risk sits in the tail. At the 99.5th percentile, the deepest net outflow reaches 6.1% of supply. The largest on record was 15.8%, a single $1.7M withdrawal on 1 August, when supply was $10.8M.
Over the first 5 minutes alone, the window before the keeper can start a recall, the 99.5th percentile is 4.2%. These two horizons, the first 5 minutes and the 30 minutes between recalls, are the basis for sizing BUFFER_FLOOR_BPS and liquidBufferBps in the next section.
Sizing the Liquidity Buffers
The Crossing Test
To determine appropriate values for BUFFER_FLOOR_BPS and liquidBufferBps, the controller is tested against actual Hub flows, assuming the least favourable initial liquidity conditions.
The analysis reuses the 6,949 starting points from the previous section, covering every withdrawal and borrow since 1 August. Immediately before each starting point, idle USDC is assumed to be exactly at the buffer floor. This represents the worst case: all liquidity above the floor is already swept, so the outflow at the starting point immediately pushes idle USDC below the threshold.
From that point onward, every subsequent flow is replayed alongside the keeper’s actions:
- The keeper checks the Hub every 5 minutes. For a conservative assumption, the first check takes place a full 5 minutes after the starting point.
- At each check, if idle USDC is below the floor, the keeper recalls enough USDC from Circle Gateway to restore idle liquidity to the liquid buffer (
liquidBufferBps), subject toMAX_REPLENISH_AMOUNTper recall. - Following each recall, the keeper must wait
MIN_REPLENISH_INTERVAL(30 minutes) before initiating another one.
The replay has no fixed duration. It continues until idle USDC is back above the floor and no recall is pending, ensuring that exits requiring multiple recalls are followed until fully resolved.
A starting point is considered a failure if idle USDC runs out at any point before the replay is complete. In this scenario, the Hub cannot fulfil the required outflow, causing the user’s withdrawal or borrow transaction to revert.
The figure below illustrates two possible replay outcomes. In the passing example, the recalls replenish idle USDC to the liquid buffer, which is sufficient to absorb subsequent outflows during the 30 minute cooldown. Idle USDC remains above zero throughout the replay. In the failing example, subsequent outflows exceed the available idle liquidity before another recall can be initiated, causing idle USDC to fall below zero and a withdrawal or borrow transaction to revert.
Most starting points resolve quickly. Either subsequent inflows bring idle USDC back above the floor before the first keeper check, or a single recall is sufficient. Only the largest exits require a second recall.
Failures occur at one of two distinct stages, each corresponding to a different parameter:
- Before the first recall arrives: The Hub runs out of idle USDC while waiting for the initial recall, typically because a single withdrawal or borrow exceeds the available liquidity at the floor.
BUFFER_FLOOR_BPSmust provide sufficient liquidity to absorb these outflows.
These failures fall steeply as the floor rises.
- During the 30 minute cooldown after a recall: The Hub runs out of liquidity before another recall can be initiated. The liquid buffer must be large enough to absorb outflows during this waiting period. This is what
liquidBufferBpsis designed to cover, as long as a single recall of up toMAX_REPLENISH_AMOUNTcan restore idle USDC to the buffer, which holds at current supply.
The Crossing Test is run across every combination of floor (BUFFER_FLOOR_BPS) and buffer (liquidBufferBps) values. For each combination, the proportion of starting points that fail indicates how frequently the Hub would have been unable to fulfil a withdrawal or borrow due to insufficient idle liquidity. This failure rate serves as the measure of liquidity risk used to compare parameter combinations in the next section.
Balancing Liquidity Risk, Revenue and Operations
The Crossing Test serves as the risk filter. A liquidBufferBps and BUFFER_FLOOR_BPS pair is accepted only if no more than 0.5% of starting points fail. This means the configuration successfully handles at least 99.5% of the historical withdrawal and borrow sequences observed since 1 August.
Floors (BUFFER_FLOOR_BPS) below 5% fail this threshold regardless of the buffer, up to 12%. From a 5% floor upward, the pair passes once the buffer (liquidBufferBps) reaches approximately 7% to 8%. The only exception below a 5% floor is a 4.5% floor paired with a buffer of at least 13%.
Within the passing range, increasing the buffer has little effect on the failure rate. The floor is the primary driver of risk reduction.
The same replays also show how often one recall is not enough. A second recall is needed when outflows during the 30-minute MIN_REPLENISH_INTERVAL push idle USDC back below the floor after the first recall has landed. While the floor largely determines the failure share, the buffer determines how often this happens: a larger liquidBufferBps restores more headroom with each recall, so fewer exits need a second one.
The Crossing Test evaluates liquidity risk under the least favourable starting conditions, but does not capture revenue generation or the actual frequency of recalls, as both depend on how idle USDC evolves over time.
To compare the passing parameter combinations, the controller is therefore replayed continuously against the Hub’s actual flows since 1 August. For conservatism, each recall is assumed to take 5 minutes to settle after being initiated by the keeper.
For each parameter pair, the replay records two measures. The first is net annualised revenue at the current supply level of $59.0M, assuming a Gateway annual percentage rate of 2.9%, deducting Circle fees and gas costs, and applying no exposure cap. The second is the number of recalls the keeper would have made, annualised.
A narrower gap between the floor and buffer generates more revenue but requires substantially more recalls. Each recall replenishes idle liquidity only slightly above the floor, so another recall is often needed soon after. For example, a 5.5% floor with a 7% buffer generates approximately $23.1k in annual revenue but requires around 244 recalls per year. Each recall goes through Circle’s attestation service, which adds operational overhead and execution dependencies.
The replay is a single historical path, so it is used to compare revenue and operational load rather than to measure liquidity risk. Whether an individual large outflow would have found the Hub short depends on the exact amount of idle USDC at that moment, which varies between parameter pairs without a consistent pattern. Liquidity risk is therefore measured by the Crossing Test alone.
Selected parameters: BUFFER_FLOOR_BPS = 5% and liquidBufferBps = 9%.
The pair is selected on two criteria from the Crossing Test and two from the replay:
- Risk: 0.46% of starting points fail the Crossing Test (17 distinct events), within the 0.5% acceptance threshold.
- Multiple recalls: 0.47% of starting points needed a second recall in the Crossing Test, 51% fewer than with an 8% buffer (0.96%) and 84% fewer than at the highest-revenue combination (2.96%).
- Operations: Approximately 76 recalls and 80 sweeps per year, roughly 1 to 2 recalls per week. This is 40% fewer recalls than with an 8% buffer (126) and 69% fewer than at the highest-revenue combination (244).
- Revenue: Approximately $13.9k per year at current supply, against $20.7k with an 8% buffer.
As this is the controller’s first deployment, and supply on the Hub is growing rapidly, a more conservative pair is preferred over the higher revenue options. The buffer can and should be reassessed once live operating data is available, and again as supply grows, since each 10M USDC recall covers a smaller share of supply on a larger Hub.
Historical Replay of the Selected Parameters
To show how the controller would have behaved in practice, the selected parameters (BUFFER_FLOOR_BPS = 5%, liquidBufferBps = 9%) are replayed on the Hub’s actual flows since 1 August.
Over the three months, the keeper would have made 21 sweeps and 20 recalls. No withdrawal or borrow would have reverted, and there would have been no downtime.
The bottom panel shows each recall as a stem, with its height representing the time idle USDC remained below the floor before the recall landed. The 5 to 10 minute duration reflects conservative assumptions: up to 5 minutes until the next keeper check, plus a 5 minute delay for Circle’s attestation service. In practice, settlement onchain should take only 1 to 2 blocks, and the average detection time is 2.5 minutes.
Being below the floor does not mean users had to wait. Throughout these minutes the Hub still held between 0.2% and 5% of supply, and every withdrawal and borrow was paid from it. The longest episode, about 21 minutes on 28 October, came from a second outflow arriving just after a recall, which had to wait for MIN_REPLENISH_INTERVAL before the next recall could be sent.
Reported utilization is unchanged, because swept USDC still counts as supply. The dashed line shows the highest utilization the Hub could reach if all idle USDC were borrowed before a recall.
The controller changes only the upper tail of the idle distribution of USDC reserve. Without it, idle was above 10% of supply 28% of the time. With it, that share falls to 13%, because the excess above the buffer is swept into Circle Gateway. Below the buffer, the distribution is almost unchanged. Idle fell below 5% of supply 3% of the time both with and without the controller. Those dips come from the Hub’s own flows when the Gateway is already empty, not from the controller.
Exposure Caps
Exposure caps limit the amount of USDC held in Circle Gateway, rather than managing day to day liquidity. That is the role of BUFFER_FLOOR_BPS and liquidBufferBps. Instead, the caps limit the Hub’s exposure to Circle, particularly in the event that Circle’s attestation service becomes unavailable.
A fixed cap of 10M USDC is proposed, matching MAX_REPLENISH_AMOUNT, so the entire Gateway position can be recalled in a single transaction. Above this level, a full unwind would require multiple recalls separated by 30 minute cooldowns. Any increase beyond this limit should therefore require a review of the parameters.
To also limit relative exposure during periods of low utilization, a second cap of 25% of total Hub supply is proposed. This ensures that no more than a quarter of supplied USDC can be held in Gateway.
Neither cap would have been reached in the historical replay. Gateway exposure peaked at 9.7M USDC, or 16.9% of supply, during the late October idle liquidity spike. Consequently, neither cap would have constrained sweeps or required additional recalls.
Growth and Review Triggers
The floor, buffer and percentage cap are set as shares of supply, so they adjust automatically as the Hub grows. The recall size does not: each recall is capped at 10M USDC whatever the size of the Hub. For the initial parameterization, exposureCapAbs is therefore set at 10M USDC, so the Gateway never holds more than one recall can return. Every parameter in this proposal has been sized for that regime, in which a single recall always brings the full position back to the Hub.
The 10M USDC cap is a starting point and is expected to be raised over time. As supply grows, idle liquidity available for sweeping above the buffer will increasingly exceed 10M USDC, causing the cap to leave potential yield unearned. Assuming utilization remains unchanged, this is expected to become relevant when supply reaches approximately $200M to $250M.
Increasing the cap would change the risk profile, as a full unwind would require multiple recalls separated by 30 minute cooldowns, requiring the Hub to absorb larger outflows between recalls. Before raising the cap, BUFFER_FLOOR_BPS and liquidBufferBps should be recalibrated to account for multi recall exits. The cap can then be increased incrementally in line with the updated parameters.
Recommendation
This publication recommends deploying the LiquidityBufferController on the Ethereum Core Hub USDC reserve with a 5% buffer floor, a 9% liquid buffer, a 10M USDC fixed exposure cap and a 25% relative exposure cap.
A buffer floor (BUFFER_FLOOR_BPS) of 5% is recommended as one of the lowest levels that keeps the Hub within the 0.5% Crossing Test limit. Even in the least favourable case, with all liquidity above the floor already swept at the moment an outflow arrives, this floor covers 99.5% of the withdrawal and borrow sequences observed since 1 August. The analysis shows that the floor carries most of the liquidity risk. The failures that remain are mostly single transactions larger than the idle USDC available, which only a higher floor can absorb.
A liquid buffer (liquidBufferBps) of 9% is recommended. Narrower buffers also pass the risk filter and earn more, but they require more round-trips through Circle. A 9% buffer cuts recalls to roughly one or two per week, and halves the share of exits that need a second recall. Given that this is the controller’s first deployment and supply is growing rapidly, fewer operations are prioritised over the additional revenue.
A fixed exposure cap (exposureCapAbs) of 10M USDC is recommended, matching the maximum recall size. This ensures the full Gateway position can always be returned in a single recall, the regime in which the floor and buffer were sized. Above this level, a full exit would require several recalls 30 minutes apart, so any future increase should be proposed together with a recalibration of the floor and buffer.
A relative exposure cap (exposureCapBps) of 25% of supply is recommended as a backstop. The fixed cap does not limit the share of the Hub exposed to Circle, and that share can only become large if utilization falls sharply and leaves a significant amount of idle USDC above the buffer.
Replayed on the Hub’s flows since 1 August, this configuration would have made 21 sweeps and 20 recalls. Most of its effect falls on the upper tail of idle liquidity: the share of time idle USDC exceeded 10% of supply falls from 28% to 13%.
At current supply and recent utilization of approximately 91%, the configuration is expected to generate approximately $14k in net annual revenue, assuming a Gateway APR of 2.9%. This is likely to be the lower end of the range. The Hub is new and highly utilized, so idle USDC currently sits mostly below the buffer, leaving little to sweep. Revenue is considerably more sensitive to utilization than to supply. If average utilization settled between 85% and 88%, expected revenue at current supply would rise to approximately $47k to $93k per year. If the exposure cap is raised in line with growth, revenue also scales roughly in proportion to supply. At 88% to 85% utilization, this means approximately $401k to $792k at 500M USDC and $803k to $1.6M at 1B USDC.
These parameters are intended for the initial deployment and should be reviewed after approximately three months of live operation, allowing actual keeper response times, attestation latency, recall frequency, and revenue to be compared against the assumptions in this analysis. A further review should take place once the 10M USDC exposure cap begins to bind regularly, expected at approximately $200M to $250M in supply, or earlier if utilization falls.
Specification
Market
| Network | Hub | Hub Address | Asset | LiquidityBufferController |
|---|---|---|---|---|
| Ethereum | Core Hub | 0xCca852Bc40e560adC3b1Cc58CA5b55638ce826c9 | 0xA0b86991c6218b36c1d19D4a2e9Eb0cE3606eB48 | TBD |
Onchain Configuration (LiquidityBufferController)
| Parameter | Value |
|---|---|
| liquidBufferBps | 900 (9%) |
| exposureCapBps | 2500 (25%) |
| exposureCapAbs | 10,000,000 USDC |
| sweepMinDelay | 86400 (1 day) |
| maxFee | 2 USDC |
Offchain Configuration (keeper)
| Parameter | Value |
|---|---|
| BUFFER_FLOOR_BPS | 500 (5%) |
| MIN_SWEEP_AMOUNT | 100,000 USDC |
| MAX_REPLENISH_AMOUNT | 10,000,000 USDC |
| MIN_REPLENISH_INTERVAL | 1800 (30 minutes) |
| MIN_REPLENISH_AMOUNT | 10,000 USDC |
| CHECK_INTERVAL | 300 (5 minutes) |
Roles
| Role | Holder |
|---|---|
| DEFAULT_ADMIN_ROLE | TBD (executor) |
| PAUSER_ROLE | TBD |
| KEEPER_ROLE | TBD (bot) |
Next Steps
- Gather community feedback on this ARFC.
- If consensus is reached, escalate this proposal to the Snapshot stage.
- If the Snapshot outcome is YAE, escalate this proposal to the AIP stage. The AIP will:
- set the LiquidityBufferController on the Ethereum Core Hub USDC asset;
- configure the on-chain parameters outlined in the Specification.
- Following execution, TokenLogic will monitor the controller and publish its activity, including sweeps, recalls, Gateway exposure, idle liquidity and revenue, on the TokenLogic dashboard.
- Review the parameters after approximately three months of live operation, or earlier if any of the review triggers occurs.
Disclaimer
TokenLogic is an active service provider to the Aave DAO, the beneficiary of stream 100086 and the KPI as outlined in this publication. The scope of this engagement is available via this forum proposal.
TokenLogic supports and maintains an independent delegate voting platform within the Aave community.
TokenLogic and associated entities have no undisclosed material conflicts of interest at the time of submission.
Copyright
Copyright and related rights waived via CC0.
















