Establish Parameter Committee

Temperature Check
,
1

Summary

This temperature check introduces a framework to establish a Parameter Committee responsible for actively managing Liqwid Finance’s interest rate model parameters across markets.

The committee will be responsible for setting and updating:

  • Base rate
  • Utilization multiplier (pre-optimal slope)
  • Optimal utilization point
  • Post-optimal utilization multiplier (post-optimal slope)

These parameters will be adjusted according to a clearly defined hierarchy of objectives:

  1. Solvency / bad debt prevention
  2. Liquidity availability
  3. Target utilization (capital efficiency)
  4. Borrow demand stability

Motivation

Liqwid operates a dynamic interest rate money market, similar in design to protocols like Aave.

In such systems, interest rate parameters are not static risk settings - they are active control levers that directly influence:

  • Liquidity availability
  • Borrower behavior
  • Liquidation dynamics
  • Protocol health

Dynamic Rates Are Necessary - But Not Sufficient

Dynamic interest rates are designed to self-balance supply and demand.

However, in practice, even leading protocols like Aave - which operate under the same principles - still:

This suggests:

Base dynamic rate mechanisms alone are not sufficient to maintain optimal market conditions

More recently, Aave has gone further by introducing automated systems (Risk Agents), moving toward:

Algorithmic parameter management (parameters that evolve continuously based on data)
(AIP-432, AIP-455)

Current Market Conditions

Across Liqwid markets - particularly stablecoin markets (where efficiency matters most) - we currently observe:

  • Persistently low utilization
  • Large amounts of idle capital
  • Utilization significantly below optimal levels

Current stablecoin market utilization ratios (as of Friday, May 8, 2026, 15:37 UTC):

Market Utilization Optimal utilization
DJED 60.53% 90%
wanUSDT 51.96% 90%
wanUSDC 44.97% 90%
iUSD 37.81% 90%
USDM 38.32% 90%
USDA 31.06% 90%
USDCx 9.92% 90%

*Only markets with over $100k liquidity were included above.

Interpretation

This indicates:

  • Underutilized liquidity
  • Reduced capital efficiency
  • Suboptimal borrower incentives

Recent data suggests:

Borrowers are no longer willing to pay the higher rates that were sustainable during bull market conditions

This creates an opportunity:

Lower rates → higher utilization → tighter spreads → improved overall revenue

Preventing Negative Feedback Loops

A critical dynamic in lending markets is the supply-demand feedback loop:

  • If utilization is low → yields for lenders decrease
  • If yields are insufficient → lenders migrate elsewhere
  • Reduced supply → lower liquidity → worsened borrower conditions
  • Borrow demand declines further → reinforcing the cycle

This proposal explicitly aims to:

Increase borrower demand while ensuring lender yields remain competitive

Why This Matters

If interest rate parameters are not actively managed:

  • Liquidity may remain idle
  • Borrower demand may stagnate
  • Lenders may leave due to uncompetitive yields
  • Markets may become uncompetitive
  • Risk conditions may drift out of alignment

At the same time:

  • Large, infrequent changes → risk of overshooting
  • Infrequent updates → market inefficiency

This leads to:

Smaller, more frequent, data-driven adjustments are likely superior to large, infrequent ones

This approach mirrors how Aave has historically operated through incremental updates rather than large parameter shifts
(AIP-350 - Interest rate updates)

The Lending Market Trilemma

Setting interest rate parameters is fundamentally a multi-objective optimization problem.

There is no single “optimal” configuration - only trade-offs.

The Core Trade-Off

Capital Efficiency (Utilization)
vs
Liquidity Buffer (Withdrawability)
vs
System Safety (No Bad Debt)

You cannot maximize all three simultaneously.

Parameter Trade-Offs Explained

1. Optimal Utilization

  • Higher (e.g. 90%)

    • ↑ Capital efficiency
    • ↓ Liquidity buffer
    • ↑ Risk in stress scenarios

  • Lower (e.g. 80%)

    • ↑ Safety & withdrawal reliability
    • ↓ Efficiency

2. Pre-Optimal Slope (Utilization Multiplier)

  • Higher slope

    • ↑ Borrowing costs earlier
    • ↓ Utilization
    • ↑ Liquidity buffer

  • Lower slope

    • ↑ Borrowing demand
    • ↑ Utilization
    • ↓ Buffer

3. Post-Optimal Slope

  • Steeper slope

    • Aggressively protects liquidity
    • Forces deleveraging
    • May cause rate spikes

  • Flatter slope

    • Smoother borrower experience
    • Weaker protection in stress

4. Base Rate

  • Higher base rate

    • Ensures minimum yield for suppliers
    • Discourages low-value borrowing

  • Lower base rate

    • Increases borrower accessibility
    • May reduce lender incentives

Key Insight

Improving one dimension often comes at the expense of another

Therefore, parameter setting must follow a clear priority hierarchy.

Proposed Framework

The committee will manage parameters according to the following descending order of importance:

1. Solvency / Bad Debt Prevention

Ensure:

  • Liquidations can be processed under stress
  • Sufficient liquidity exists during market shocks
  • Bad debt risk is minimized

2. Liquidity Availability

Maintain:

  • Sufficient buffer for withdrawals
  • Healthy market functioning during normal conditions

3. Target Utilization (Capital Efficiency)

Optimize:

  • Proportion of supplied capital actively generating yield
  • Balance between idle liquidity and active borrowing

4. Borrow Demand Stability

Ensure:

  • Predictable rate environments
  • Minimal parameter volatility
  • Borrower confidence and retention

Committee Structure

Initial Composition

  • Core team members
  • Same signers as the admin multisig

Governance Oversight

Committee authority is explicitly subordinate to governance.

  • Any changes to the committee members/admin multisig (including adding, removing, or replacing signers) must be approved through a governance vote
  • The committee will operate under an initial 6-month mandate, after which governance will reassess its performance, structure, and scope
  • Following this review, the DAO may choose to renew, modify, extend (e.g., to a 12-month term), or revoke the committee’s authority based on outcomes and community sentiment (revocation does not require a separate action - unless governance explicitly approves an extension, the committee’s authority expires at the end of its term)
  • Ensures accountability, transparency, and continued community control through defined review cycles

Future Expansion
The committee may expand to include:

  • Non-core contributors
  • Risk analysts
  • Governance participants

Selection based on:

  • Contribution
  • Analytical rigor
  • Alignment
  • Membership composition may also be reviewed and adjusted at the end of each term, supporting rotation and broader decentralization over time
  • All changes remain subject to governance approval

Operating Cadence

Instead of rigid intervals, the framework will follow a flexible, data-driven cadence:

  • Parameters reviewed continuously
  • Adjustments made incrementally and as needed

Guiding Principles

  • Favor small, iterative adjustments over large changes
  • Avoid parameter “shock” to borrowers
  • Prioritize data-driven decision making
  • Maintain consistency with observed market behavior

Reference Approach: Aave Model

A key benchmark for this framework is the methodology used by Aave:

  • Frequent, incremental parameter updates
  • Data-driven adjustments based on utilization and market conditions
  • Reduced risk of overshooting
  • Progressive evolution toward automation (Risk Agents)

This proposal explicitly considers:

Adopting Aave’s model as-is as a starting point, given its maturity and proven track record.

Communication & Transparency

The committee will ensure:

  • Public communication for every parameter update
  • Clear rationale and data backing each decision
  • Posts documenting:
    • Changes made
    • Expected outcomes
    • Observed results over time

Data Accessibility & Independent Analysis

The current closed-source nature of Liqwid contracts introduces friction, particularly for:

  • Data reproducibility
  • DAO decision-making
  • Third-party analytics

Improving data accessibility is therefore critical to the success of this framework.

To encourage decentralization and external contributions:

A method for accessing market data will be provided, including:

  • Historical utilization
  • Lender / borrower / protocol rates
  • Total supply
  • Total liquidity
  • Total debt

Path Forward

This framework is intended as an iterative progression:

Manual adjustments → structured frequent updates → fully algorithmic parameter management

This mirrors the evolution seen in Aave:

  • Manual governance updates
  • Risk steward committees (AGRS)
  • Automated agents (Risk Agents)
    (AIP-432)

Expected Impact

  • Increased utilization
  • Improved lender yields
  • Stronger borrower demand
  • Reduced risk of liquidity fragmentation
  • Prevention of negative feedback loops

Risks & Considerations

1. Centralization (Short-Term)

Mitigated by governance oversight, transparency, and communication.

2. Misconfiguration or “Overshooting” Risk

Mitigated by:

  • Smaller, incremental updates
  • Higher adjustment frequency
  • Avoiding large parameter swings

3. Borrower Instability

Mitigated by:

  • Predictable, gradual parameter changes
  • Transparent communication

Next Steps

  1. Gather feedback
  2. Iterate on framework
  3. Conduct temperature check vote
  4. Proceed to governance if consensus emerges

Poll

  • I do not support this framework
  • I support establishing the committee and framework
  • I support the framework with all updates going through governance proposals
  • I support adopting the AGRS parameter management model
0 voters
2

  1. I am surprised. I assumed this was already fully automated, run periodically by the team, and parameters changes proposed via governance when necessary after analysis.
    (Actually, I was looking a few days ago at open-sourcing a dashboard of per token risk metrics mostly according to Liqwid risk framework. Because for all Cardano lending protocols, users have no visibility on actual calculated risks and parameters choice. And if you look, you would be quite surprised at how divergent the parameters can be for a given token across protocols, and even worried about the parameters of some.)

  2. So, from where I arrive, I don’t see a difference with a committee for now made mostly of core contributors, and what is supposed to already be in place.

  3. As mentioned, it’s fully data driven. So what is the point of a committee if clear calculations and a framework is in place ?

  4. 4-6 weeks parameters changes is WAY TOO OFTEN. I’ll aim for not more often than every 3 months. Borrowers and lenders need some level of predictability on rates. Remember, some say dynamic rates are not that great because of rates predictability issues, and changing the parameters often make it even more so. I don’t want to have to check the parameters of a market I am in and how its utilization and rates react every 4-6 weeks. Yes I already monitor utilization and rates which are dynamic, but their movements are rather natural and kind of predictable based on the market or events. Changing the parameters very often breaks that predictability of rates changes.

  5. Saying it again. To me that is one of the main task of the team under ‘protocol maintenance/monitoring’ or ‘risk control’ or ‘markets monitoring’, and I am surprise you come now with this. Best in my opinion is to make it very much fully data driven, and open-source the calculations/metrics. All the needed data is public anyway.

  6. I would be more cautious about interpreting levels of utilization as effects of suboptimal parameters. We had plenty of different utilization regimes (some with high utilization) without changing parameters, and mostly driven by other factors.

  7. the real solution to the problem you’re stating is something like ‘algorithmic dynamic rates’
    → dynamic rates are supposed to be the solution to this balance of supply/borrow and create the self-correcting healthy markets.
    → you’re saying it’s not enough (and actually the problem description is really what dynamic rates are supposed to solve)
    → so the real solution is going even further than dynamic rates: ‘algorithmic dynamic rates’ = the parameters are not static numbers but slowly algorithmically evolving numbers based on some metrics

1 Like
3

I fully agree with the last point (7) of Gil. Here is also my personal take.

As I have mentioned several times previously, the Cardano DeFi ecosystem does not currently operate as a fully efficient market where information flows freely and capital allocation naturally converges toward optimal outcomes.

There is no universal or “magic” formula for interest rates, and a reduction in borrowing rates does not automatically translate into increased borrowing demand up to a theoretical optimal inflection point. The recent USDCx market launch, which introduced lower interest rates without generating materially higher borrowing activity relative to other stablecoin markets, illustrates this dynamic well.

In practice, DeFi lending markets are influenced by a much broader set of factors, including but not limited to:

• CNT collateral quality
• ADA market outlook and volatility expectations
• Broader Cardano and blockchain macro environment
• Available stablecoin liquidity
• Stablecoin infrastructure maturity
• RWA / TradFi integrations
• Application UI/UX
• Ecosystem integrations and composability
• Perceived security, reliability, and ease of use
• Reputation, brand perception, and trust

While I believe the idea behind establishing such a committee is constructive and well intentioned, I do not believe that the creation of the committee alone is sufficient to fully understand or efficiently manage these market dynamics.

For these reasons, I would vote No on the proposal in its current form.

4

Hey @gil and @FlorianVolery - thanks for the thoughtful feedback. I’ll respond following Gil’s structure for clarity.

1) Automation / data-driven approach

I largely agree with the principle: this should be fully data-driven and ultimately automated, with transparent, open metrics.

Where I see the current gap is not in the availability of data, but in how accessible it is and how consistently and systematically it’s being applied. Historically, we’ve relied on a proprietary model at launch, but:

  • it hasn’t been iterated on frequently enough, and
  • it diverges from frameworks used by more mature protocols like Aave

That divergence mattered less in bullish conditions, where borrower demand was relatively inelastic. But recent lower utilization suggests that, in current market conditions, borrowers are more rate-sensitive.

So the goal here isn’t to replace a data-driven system with a committee - it’s to formalize and enforce a data-driven process, with:

  • clearer assumptions
  • consistent iteration
  • and ultimately a path toward automation

I fully agree that open-sourcing risk metrics and calculations would be a strong step in that direction.

2) Role of the committee

The intent is not to introduce subjectivity where a framework already exists.

Right now, the process is effectively:

ad hoc + infrequent updates + implicit assumptions

The committee is meant as a transitional coordination layer to:

  • ensure parameters are actually reviewed and updated regularly
  • apply a consistent framework
  • reduce inertia and decision latency

In other words, it’s less about who decides, and more about ensuring decisions happen in a structured, timely way until we reach a more automated system.

3) Update cadence (4-6 weeks)

I understand the concern around predictability - it’s valid.

That said, I think the key distinction is:

  • large, infrequent changes → disruptive, harder to anticipate
  • small, incremental adjustments → smoother, more predictable evolution

My initial framing leaned toward the former. After reviewing approaches like Aave, I’m increasingly convinced the latter is preferable.

In practice:

  • smaller adjustments, more frequently
  • tightly scoped changes
  • clear communication and transparency

This reduces the risk of overshooting while still allowing parameters to track market conditions.

That said, cadence should be configurable - not dogmatic. A hybrid approach (e.g. frequent reviews, but changes only when thresholds are met) may address both concerns.

4) On utilization and market dynamics (Florian’s point)

I agree that utilization is not solely a function of parameters.

As Florian outlined, lending markets are influenced by a wide set of factors:

  • collateral quality
  • macro conditions
  • liquidity availability
  • ecosystem maturity
  • user trust and UX

And importantly: borrower expectations and risk appetite

So the proposal is not based on the assumption that:

“better parameters alone will fix utilization”

Rather, it’s that:

parameters are one controllable lever, and currently under-optimized relative to available data.

The USDCx example is a good illustration - lowering rates alone didn’t drive demand, which reinforces that parameters must be evaluated in context, not in isolation. I do want to note that a single interest rate reduction relative to other markets at launch doesn’t do the AGRS framework or adjacent structures justice. More rate reductions in the USDCx market according to the framework are more likely to attract new borrowers. Additionally, a reduction in rates shouldn’t scare off lenders, given that lenders are currently earning a sub-optimal yield spanning from a low-utilization in the first place.

5) On “algorithmic dynamic rates”

I think this is the key point where there’s actually strong alignment.

Dynamic rates are meant to self-balance markets - but in practice, even protocols using similar mechanisms (like Aave) still:

  • adjust parameters regularly
  • and are now moving toward automated parameter management systems (e.g. risk agents)

That evolution is essentially what you’re describing:

parameters that evolve algorithmically based on market conditions

The path I’m proposing is:

  1. Standardize a framework (clear metrics, thresholds, methodology)
  2. Apply it consistently via a lightweight governance process
  3. Gradually move toward automation (algorithmic adjustments)

So I don’t see this as an alternative to your suggestion - but rather as a stepping stone toward it.

Summary

  • I agree the end state should be fully data-driven, transparent, and ideally automated
  • The committee is a temporary mechanism to enforce consistency and iteration, not to replace objective frameworks
  • Parameter adjustments should likely be smaller and more incremental, even if reviewed frequently
  • Market outcomes depend on many factors - parameters are just one lever, but an important and currently underutilized one
  • Longer term, this should evolve toward algorithmic parameter tuning, in line with where leading protocols are heading
5

I think the framework is very well-designed, intelligent, and flexible. Of course, it will depend a lot on common sense and empowering the committee to make the right decisions. I think the frequency is good; in fact, I think the committee should have constant market updates. We’re not talking about abrupt changes just fine-tuning and perhaps don’t change at all, in case of a major parameters reset It should respect a longer timeframe and open up the discussion more. In short, Kylix is going in the right direction. There’s no point in reinventing the wheel; AAVE has years of experience behind it. It’s about taking advantage of what’s already there, refining it, and adapting it to our scenario.

1 Like
6

I feel liqwid currently does not offer competitive rates not just in broader crypto space but in cardano world as well. even if the utilization is as low as 40% or less is also resulting in close to 18-20% of borrowing rates which will push away any borrower from the platform.
biggest problem here i see with an empowered committee is Centralization of decision powers. but i feel it is much needed thing to keep the borrowing cost low and competitive with constantly changing dynamics of the defi markets.
but i would suggest that keep powers of this committee time bound where after every 6-12 months the committee will seek those powers of authority. and by default there should not be any powers given to the committee.

1 Like
7

I fully agree with Schwab’s perspective here - especially on the importance of flexibility, steady iteration, and leveraging proven frameworks like Aave.

Thanks for the thoughtful feedback - this aligns really well with the direction I’m aiming for.

1 Like
8

First of all, thank you for chiming in. I’m really glad to see you back posting on the forum and engaging in what I think is incredibly important governance discussion - it genuinely means a lot given the context, and I appreciate you going out of your way to contribute here.

I also want to say I’m very receptive to your suggestion around introducing a time-bound mandate for the committee. I think this strikes a strong balance: it doesn’t meaningfully add friction compared to routing every parameter update through governance, but it does introduce a healthy checkpoint where the DAO can evaluate the committee’s performance and decide whether to continue, adjust, or fully rethink the framework.

Based on this, I’ll adjust the temperature check to include an initial 6-month term for the committee. From there, we can reflect on performance and outcomes during that first cycle and decide whether extending to a 12-month period - or keeping a shorter cadence - makes the most sense for the DAO.

Beyond that, having a defined term creates a natural moment for reflection and iteration - whether that’s refining the process itself or rotating/adding members to ensure continued decentralization in how fee structures and interest rate parameters are managed within Liqwid.

I also agree with your broader point on competitiveness - keeping borrowing rates attractive in a rapidly shifting DeFi landscape is critical, and that’s ultimately the motivation behind exploring a more responsive model like this.

Appreciate the thoughtful input here - this is exactly the kind of discussion that helps move things in the right direction.

1 Like
9

I also want to clarify - if the current wording has led to any confusion - how I interpret the vote options in this temperature check:

  • I do not support the framework
    Maintain the current system with no changes.
  • I support adopting the AGRS parameter management model
    Establish a committee and follow the same operating approach as AGRS: higher frequency updates, smaller adjustments, and continuous market-driven reevaluation.
  • I support the framework with all updates going through governance proposals
    All updates continue to go through explicit governance proposals and approvals (as is the case today), while signaling support for evolving the parameter model itself - potentially combining elements of AGRS with the original proposal.
  • I support establishing the committee and framework
    Implement the committee structure alongside the initially proposed model: 4–6 week review periods, with flexibility for more significant adjustments when needed, and more frequent/aggressive updates limited to the second slope during periods of elevated market stress.

My goal here is to ensure that each option is clearly understood so that the signal we gather accurately reflects the DAO’s preferences.

1 Like
10

Nice addition here Shiv, agreed

1 Like
11

I feel liqwid currently does not offer competitive rates not just in broader crypto space but in cardano world as well.

You are correct @Shiv Liqwid rates are not competitive even within the Cardano DeFi ecosystem.

At the end of the day there is a tradeoff of competitive interest rates to attract borrowing demand and generating revenue. We all agree Cardano DeFi is inefficient markets yet still we can set parameters such that the interest rates at the optimal utilization point (kinkPoint) are both attractive to borrowers and are generating strong revenues. This is exactly what algorithmic money markets enable via accurate parameter configuration. If the parameters are not correctly configured having some framework in place to adjust them based on recent data is the only viable method to achieve the balance between consistent borrow demand and maximizing protocol revenues. This is what this proposal is aiming to implement.

2 Likes
12

I’d want us to have clearer guidance on how to set those params. Possibly define what a “small” adjustment is. i.e., how much % from the existing value. Essentially, leave less space for interpretation.

1 Like
13

That’s a very fair point, and I agree that reducing ambiguity here is important.

One of the strengths of the Aave AGRS model is precisely that it introduces clearer guidance around both adjustment magnitude and frequency, which helps minimize subjective interpretation while still allowing for responsiveness.

I think we can - and should - apply a similar approach here. Even if we opt for a lower-frequency model (e.g., 4-6 week review cycles), we can define target ranges or bounds for what constitutes a “small” vs. “meaningful” adjustment (e.g., % changes per update), alongside the already explicit expectations around cadence.

This would strike a balance between structure and flexibility: giving the committee clear operating guidelines while still allowing them to react appropriately to market conditions.

Happy to incorporate more explicit parameterization into the framework so expectations are well-defined upfront.

2 Likes