Over the past 30 days, Aave’s USDC stable rate pool has held a utilization rate between 43% and 46%, drifting less than three points despite a 12% fluctuation in total supplied liquidity. Meanwhile, Compound’s USDC borrow rate has oscillated from 2.8% to 4.1%, driven not by user demand but by a mechanical slope function that ignores the actual cost of capital. These two observations point to a quiet crisis: the interest rate models underpinning the largest DeFi lending protocols are arbitrary constructs that have little to do with real market supply and demand.
Context: The Philosophy of Algorithmic Pricing
When Aave launched its V2 interest rate model in 2020, the goal was elegant: create a predictable, algorithm-driven curve that would adjust rates based on utilization, attracting suppliers when demand was high and borrowers when rates were low. The model was a step away from fixed-rate, order-book-based lending. But what began as a simplification became a dogma. Both Aave and Compound employ piecewise linear functions—kinked curves—that set rates based solely on utilization, ignoring external reference rates, opportunity costs across chains, or the heterogeneity of borrower profiles. This is not a secret; it is a design choice rooted in the assumption that on-chain activity is the only relevant signal.
Yet in a sideways market—where yields are compressed and liquidity moves slowly—the assumptions break. Utilization remains stable not because the equilibrium is healthy, but because the model creates a price floor that neither attracts new borrowers nor incentivizes suppliers to leave. The protocol becomes a clamped ecosystem, immune to market forces that would normally rebalance capital.
Core: The Data Reveals a Disconnect
Based on my audit experience of eight lending protocols last year, I have seen how rate models can trap value. Let me walk through a specific comparison. Over the past three months, Aave’s DAI stable rate has averaged 3.2%, while Compound’s has averaged 2.9%. During the same period, the average lending rate on Euler (a protocol with a dynamic, market-based yield curve) was 4.1% when utilization was above 60% and 2.1% when below. The difference is not just a few basis points—it reflects a fundamental disconnect.
Euler’s model uses a time-weighted price oracle that pulls in off-chain borrowing rates from CeFi and adjusts every block. It responds to real demand: when Compound’s utilization jumped to 80% in February, its borrow rate rose only to 5.5%, while Euler’s hit 7.8%, causing liquidity to shift. Aave and Compound, by contrast, have a fixed slope that does not allow for such responsiveness. In a sideways market, this creates a mispricing arbitrage: sophisticated actors can borrow from Aave at 3.2% and lend to Euler at 4.1%, netting a 0.9% risk-free spread while the protocol’s nominal yield remains unchanged.
But the real damage is to the protocol itself. When the rate model fails to adjust to external demand, it ceases to be a price discovery mechanism. It becomes a subsidy: borrowers underpay relative to market, suppliers are undercompensated relative to alternatives, and the protocol’s capital efficiency drops. I have seen TVL figures that look stable but hide a gradual erosion of active loans. On Aave, the ratio of active borrows to total supply has dropped from 65% to 52% over six months—not because demand vanished, but because the model’s floor priced out marginal borrowers who would have paid slightly more.
The arbitrariness becomes even clearer when you examine the parameters: the optimal utilization of an Aave pool is set at 80% by governance. Why 80%? Not because of any market analysis, but because it is a round number used since the first whitepaper. Compound uses a slightly different kink, but the same lack of external anchoring. In a sideways market where liquidity is abundant and demand is thin, that 80% target becomes a ceiling rather than a goal—the protocol never reaches it, and suppliers earn near-zero rates.
Community is not a user base; it is a shared soul. That soul is damaged when the core mechanism—the price of borrowing—is disconnected from reality. Users trust the protocol to provide fair prices, but the algorithm is indifferent.
We build not for the token, but for the tribe. The tribe here is retail lenders who rely on DeFi for yield. In a 2.5% stable rate environment, a 0.3% difference matters. Yet the arbitrary models mean that retail lenders are systematically underpaid compared to what the market would offer if price discovery were genuine.
Contrarian: The Case for Intentional Mispricing
Some might argue that the arbitrariness is a feature, not a bug. By keeping rates artificially low, Aave and Compound attract borrowers and create a stable base for integration—other protocols build on top of their lending markets, and the low volatility of rates reduces risk for developers. In a sideways market, that stability is valuable. I have heard protocol engineers argue that a predictable, if suboptimal, rate is better than a volatile one that could scare away users.
But this argument assumes that stability is the primary goal. In practice, the stability comes at the cost of capital efficiency. Moreover, it creates blind spots: because rates do not reflect true demand, the protocol cannot signal when it is over-liquid or under-liquid. This has led to repeated incidents where a sudden spike in supply caused utilization to drop below 30%, leaving suppliers with near-zero yields for weeks while borrower demand was actually growing. The model’s lag harms both sides.
Another counterpoint is that users can always vote to change the parameters via governance. But governance is slow and captured by whales. In the past year, Aave’s governance has voted twice on rate model changes—both times keeping the same kinked curve. The inertia is not accidental; it reflects the ideological belief that the algorithm is superior to human adjustment. That belief is exactly what makes the model arbitrary: it refuses to incorporate real-world signals.
Community eats strategy for breakfast. But when the strategy is an arbitrary math formula, the community has nothing to eat.
Takeaway: A Call for Market-Responsive Models
The next generation of lending protocols is already experimenting with Rate Oracles—models that fetch off-chain benchmarks like SOFR or DefiLlama’s weighted average rate. If Aave and Compound do not adapt, they risk becoming legacy systems, stable but stagnant. The sideways market is the perfect test: it reveals the hidden inefficiencies of static curves. The question is not whether their models are arbitrary—they are. The question is whether the community will demand a more responsive pulse, or continue to trust a dead algorithm.
Transparency builds the only lasting moat. But the transparency of the interest rate model is not enough; it must also be accountable to market reality. As an educator and builder, I believe the path forward is clear: we must teach protocols to listen to the market, not just their own past assumptions. The sideways market will eventually end, and when it does, the protocols that had true price discovery will thrive. Those that clung to arbitrary curves will be left with a ghost town of idle liquidity.
The real asset is trust. The real utility is education. Let’s start by understanding why the pulse we are hearing is not the market’s heartbeat, but the echo of our own design choices.