[FIP-4XX] Recalibrate the sfrxUSD / frxUSD Fraxlend IRM

[xm3van]

Authors:

xm3van

---

Summary

This proposal assigns a dedicated interest-rate model (IRM) calculator to the Ethereum sfrxUSD / frxUSD Fraxlend pair and changes two parameters:

Parameter	Current	Proposed
Zero-utilization APR	3.36%	1.50%
Vertex-rate percentage	10%	approximately 8.64% initially

The 1.50% zero rate is a compromise between restoring viable carry and protecting fee income. The initial vertex percentage places borrow APY at the 87.5% vertex, approximately 50 basis points below a conservative sfrxUSD yield reference. These settings are proposed as a 60-day market experiment, with a 30-day checkpoint rather than as a final calibration.

All liquidity-protection parameters remain unchanged: the 75–85% target-utilization band, 87.5% vertex, two-day half-life, approximately 25% minimum full-utilization rate and approximately 10,000% maximum full-utilization rate.

Approximately $2 million of liquidity entered this market in February 2026, but borrowing fell. During the subsequent $2.5 million liquidity period, 89% of assets remained unborrowed and Frax earned less than $1,000 of annualized protocol fees. Liquidity was available; viable borrower economics were not.

Evidence

Observation	Total assets	Borrowed	Utilization	Gross carry
6 February, before migration	$502k	$424k	84.5%	2.22%
$2.5m period average, 15 Feb–31 May	$2.504m	$268k	10.7%	0.52% median
27 September	$625k	$103k	16.4%	1.00%

The evidence table uses contemporaneous realized sfrxUSD yield. The curve scenarios below instead use the lower, conservative 4.088% reference yield, so their carry figures are intentionally lower.

The 7 February migration raised the zero-utilization floor from approximately 0.50% to 3.36%. Although the later liquidity injection reduced utilization sharply, the new floor kept borrow APY around 3.6–3.8%. As sfrxUSD yield approached 4%, gross carry compressed to a level unlikely to compensate for execution, yield, liquidation and unwind risks.

This timing and mechanism strongly support inadequate carry as the principal constraint, but do not prove it was the only cause. Borrower concentration, incentives, tooling and competing venues may also have affected demand.

Market history

The chart should be read as evidence of timing and mechanism rather than a controlled causal test: supplied liquidity increased sharply, but borrowing did not follow once the higher rate floor compressed carry.

Parameter rationale

Zero-utilization rate: 1.50%

The former 0.50% setting was compatible with high utilization, but it would now create approximately 3% carry at current utilization and require a larger demand response to recover lost fee income. A 1.50% floor creates approximately 2.19% carry at current utilization against the conservative reference yield. This is consistent with both the pair’s own pre-migration history—2.22% carry at 84.5% utilization—and the 1.25–2.00% range observed in comparable markets, while reducing the revenue break-even from approximately 35.7% to 28.6% utilization.

This is an experimental calibration, not a claim that 1.50% is optimal. Further revision may be justified if borrowing responds but the 75–85% target range is not approached. If borrowing does not respond at all, governance should first investigate non-rate constraints rather than automatically lower the floor again.

Vertex-rate percentage: approximately 8.64%

The vertex is calibrated from a target rate, not selected independently:

1. Conservative reference yield: the lower of the trailing 30-day and 90-day medians, currently 4.088%.
2. Desired residual carry at the vertex: 0.50%.
3. Target vertex borrow APY: 3.588%.
4. Convert that APY to Fraxlend’s per-second rate convention.
5. Solve: vertex percentage = (target vertex rate − zero rate) / (minimum full-utilization rate − zero rate)

This produces an initial vertex percentage of approximately 8.636%, rounded to 8.64% for discussion. Before the governance vote begins, the reference snapshot, resulting raw integer and affected tables must be updated together and frozen in the approved payload.

At publication, the Frax interface displayed approximately 5.1% sfrxUSD APY, above the 4.088% trailing reference; if trailing realized sfrxUSD APY remains at least 75 basis points above that reference for 14 consecutive days, the vertex may be increased at the day-30 checkpoint toward the 50-basis-point carry target, capped at a 50-basis-point APY increase, while the 1.50% zero rate remains unchanged for the full 60-day experiment.

The 50-basis-point residual balances estimation error against liquidity risk: 25 basis points offers little buffer for normal yield drift, while 100 basis points would subsidize borrowing too heavily near the vertex. If realized yield exceeds the conservative reference, carry widens safely; if yield declines, the reference rule follows it downward more quickly.

Parameters not changed

The half-life and full-utilization bounds address liquidity protection, not the underlying-yield mismatch. The adaptive full-utilization rate is already at its approximately 25% minimum, so a faster half-life cannot lower normal borrowing costs further. Above the vertex, the unchanged curve still rises to approximately 25% APR at full utilization.

Effect of the proposed curve

Using the 4.088% yield reference and the current minimum full-utilization rate:

Utilization	Current borrow APY	Current carry vs. reference	Proposed borrow APY	Proposed carry vs. reference
0%	3.42%	0.67%	1.51%	2.58%
16.4%	3.84%	0.25%	1.90%	2.19%
25%	4.06%	0.03%	2.10%	1.99%
50%	4.70%	−0.61%	2.69%	1.40%
75%	5.35%	−1.26%	3.29%	0.80%
84.5%	5.60%	−1.51%	3.52%	0.57%
87.5% vertex	5.68%	−1.59%	3.59%	0.50%
100%	28.34%	−24.25%	28.34%	−24.25%

The proposal creates a wide incentive at low utilization and compresses it as the market fills. It does not guarantee positive carry above the vertex or if sfrxUSD yield falls materially below the reference.

Revenue and lender economics

Protocol fee revenue is:
total assets × utilization × borrow APR × 10% protocol fee

The following scenarios use the $624,966 actually supplied on 27 September. They assume no new deposits and apply the proposed curve.

Scenario	Utilization	Borrowed	Borrow APR	Frax fee/year	Versus current
Current market	16.4%	$103k	3.76%	$386	1.0×
Proposed curve, unchanged demand	16.4%	$103k	1.88%	$193	0.50×
Conservative response	25%	$156k	2.08%	$325	0.84×
Moderate response	50%	$312k	2.66%	$830	2.15×
Strong response	75%	$469k	3.24%	$1,517	3.93×
Previous utilization restored	84.5%	$528k	3.46%	$1,825	4.73×
Vertex	87.5%	$547k	3.53%	$1,928	4.99×

The proposed curve exceeds the current protocol-fee run rate at approximately 28.6% utilization, or about $179,000 borrowed at current supply. Until that point, both protocol fees and lender yield are lower. At 84.5% utilization, estimated lender APY rises to approximately 2.67%, compared with approximately 0.57% today.

These are scenarios, not forecasts. The proposal is a deliberate demand experiment: a lower price must produce enough additional borrowing to offset the lower rate.

External benchmark

Comparable Aave analyses place viable yield-bearing stablecoin loop spreads in a similar range. LlamaRisk reported 192 bps of carry for a profitable PT-sUSDe/USDe position, while another assessment showed 126–177 bps across PT-sUSDe stablecoin loops. These support positive carry as a demand driver but are not directly equivalent because PT yield is fixed and sfrxUSD yield floats. See the Aave Plasma assessment.

Implementation

The current V3 calculator is shared by six Fraxlend pairs. Its setters must not be used for a market-specific change. Deploy a dedicated VariableInterestRateSettable calculator with:

Constructor field	Value
`_vertexUtilization`	`87500`
`_vertexRatePercentOfDelta`	`86362008189293168` at the stated reference yield
`_minUtil`	`75000`
`_maxUtil`	`85000`
`_zeroUtilizationRate`	`475646880`
`_minFullUtilizationRate`	`7912352300`
`_maxFullUtilizationRate`	`3164940920000`
`_rateHalfLife`	`172800`

Execution steps:

1. Recalculate, publish and freeze the exact vertex input before the governance vote begins; the approved payload must not depend on a later execution-date calculation.
2. Deploy and verify the dedicated calculator.
3. Assign its administrative roles to Frax governance-controlled addresses.
4. Accrue pair interest immediately before migration.
5. Call setRateContract(newCalculator) on pair 0xc045a53936d793839bfca146058976ef4285161e.
6. Verify the resulting previewed rate and confirm no other pair uses the calculator.

Review and success criteria

Observe the market after 30 days and conduct the parameter review after 60 days. Absolute borrowed frxUSD is the primary measure; utilization can also rise because suppliers withdraw. Make no routine parameter change at day 30, preserving a consistent 60-day test, unless a material yield decline or liquidity-risk event requires emergency action.

Success requires:

• Borrowed balance above approximately $179,000, the modeled revenue break-even at current supply.
• Median gross carry above 1% under ordinary utilization.
• No persistent utilization above 85%, bad-debt increase or material withdrawal impairment.

At the 60-day review:

• If borrowing grows but remains materially below target, consider lowering the 1.50% zero rate further or widening target carry.
• If borrowing does not respond, investigate distribution, incentives, leverage tooling and competing venues before lowering rates again.
• If utilization persistently exceeds 85%, increase rates or tighten the curve to protect liquidity.
• If suppliers leave the market, temporarily lower reserve factor.

Recalibration instruction

At day 60 and monthly thereafter, the team should:
1. Set the reference yield to the lower of the trailing 30-day and 90-day medians of realized sfrxUSD APY.
2. Set target vertex APY to `reference yield − 0.50%`, convert it to Fraxlend’s per-second rate, and solve the vertex percentage using the formula above.
3. Propose an update only if target vertex APY differs from the live setting by at least 0.25%; cap each monthly change at 0.50% APY and publish the inputs and result.

Keep the zero rate at 1.50% for the first 60 days. Reconsider it only at the 60-day review, based on borrowed balance and utilization—not merely a change in sfrxUSD yield. Every recalibration remains an explicit governance-controlled transaction.

The underlying yield series is already a trailing 30-day realized APY. Its 30-day median is responsive, while its 90-day median is steadier but slower to recognize sustained declines. Taking the lower of the two is deliberately asymmetric: it follows falling yield faster but requires a sustained increase before raising borrower rates. The available history is too short for a reliable forecasting claim, so this rule should be treated as a conservative operating convention rather than a validated predictor.

Risks

• Demand may not respond: at unchanged utilization, protocol fees fall approximately 50%.
• Yield may fall: a static monthly calibration cannot guarantee carry between reviews.
• Supply may leave: lender returns initially decline unless utilization rises.
• Leverage concentration may grow: profitable looping can produce correlated, rapidly unwinding positions.
• Higher utilization reduces liquidity: the unchanged vertex and adaptive full-utilization controller remain the main protections.
• Historical comparison is confounded: the pre-migration period does not isolate the effect of rates.

Long-term direction

A static curve will eventually drift away from sfrxUSD yield. The stronger design is a Fraxlend-compatible yield-indexed IRM that targets:
borrow rate at target utilization ≈ EMA(sfrxUSD yield) − governance carry target\
Curve governance has already deployed an analogous EMAMonetaryPolicy for its sfrxUSD-long LlamaLend market. It smooths the underlying sfrxUSD reward rate and uses it as the curve reference. See the executed Curve proposal.

The proposed static calibration is a bridge toward a yield-indexed Fraxlend IRM. A minimal design would:

• Read the scheduled sfrxUSD reward rate and smooth it with an EMA.
• Target `EMA yield − governance carry target` at normal utilization.
• Retain the existing adaptive full-utilization controller and steep post-vertex curve.
• Bound rates and rate changes, and use the last valid observation if the reference becomes stale.
• Operate through a dedicated, audited calculator for this pair.

This should require a separate technical review and governance vote. It is not safely achieved by reading an instantaneous yield inside the current view-only V3 calculator.

Vote

• For: deploy a dedicated calculator with a 1.50% zero rate and an initial approximately 8.64% vertex-rate percentage, with all other parameters unchanged.
• Against: retain the current 3.36% zero rate and 10% vertex-rate percentage.
• Abstain

Methodology

The analysis uses 315 daily Ethereum archive-state observations from 17 November 2025 through 27 September 2026. Fraxlend state is reconstructed with previewAddInterest(); sfrxUSD yield is the trailing realized ERC-4626 share-price APY; and borrower APY compounds the verified per-second rate for 365 days.