Aave V3 Market Analysis: Borrowing Patterns and Institutional Fund Flows

[WalleDAO]

Dear Aave community,

RESEARCH FINDINGS: ANALYSIS OF SAME-BLOCK BORROWING PATTERNS IN AAVE V3

Our analysis has identified an interesting protocol design pattern: a single MEV bot has processed $7.3 billion in borrowing activity on Aave V3 utilizing the block-level interest calculation mechanism. This represents a notable difference in yield generation compared to other approaches and may offer insights for protocol optimization. The detailed findings below highlight both this pattern and broader market trends that can inform Aave’s strategic direction.

Two Key Findings

Scale, Efficiency, and the Value of Long-term Borrowing

• Aave leads in scale (user count and loan volume 7x that of Spark), while Spark shows stronger protocol-level borrowing yield performance (0.41% vs 0.19%)
• Long-term borrowing (>30 days) is the main revenue source for both platforms (Aave 92.54%, Spark 94.98%)
• 44.25% of Aave’s borrow funds are used for long-term borrowing, compared to 74.44% for Spark
• Interest rate strategies influence user behavior: stable rates attract long-term users, while dynamic rates suit short-term traders
• Market Insight: Optimizing long-term borrowing experience is key to improving platform revenue while maintaining Aave’s scale advantage

Same-Block Borrowing Pattern Analysis

• 103 wallets executed 24,319 same-block borrowing transactions on Aave V3 ($25.176B total), with largest MEV bot contributing $7.3B utilizing Aave’s end-of-block interest calculation mechanism
• These transactions operate within the normal design parameters of the protocol, which calculates interest based on time blocks rather than individual transactions
• Largest MEV bot borrower primarily used USDC ($3.11B) and USDT ($2.55B), with funds flowing to DODO (76.96%) and Uniswap (13.3%) for arbitrage opportunities
• Key DeFi design consideration: interest calculation mechanisms significantly influence user behaviors and capital flow patterns in lending protocols

Three Market Trends

DeFi Professional Specialization

• Aave positions as trading infrastructure: 13% of funds (approx. $11.394 billion) flow to Uniswap, 9.17% (approx. $8.033 billion) deployed to DODO trading protocol
• Spark focuses on capital optimization: 16.45% of funds (approx. $2.221 billion) flow to MakerDAO ecosystem, 14.40% (approx. $1.944 billion) through aggregator protocols
• The two platforms complement each other: 13.06% (approx. $1.763 billion) of Spark funds bridge to Aave platform
• Trend Prediction: DeFi will continue to evolve toward specialization, with platform cooperation exceeding direct competition

Institutional Funds Dominating the Market

• A single institutional wallet borrowed over $5 billion across protocols (Spark $2.64 billion, Aave $2.45 billion), 100% in WETH
• Ultra-long holding periods: up to 545.4 days on Spark, 296.1 days on Aave
• Complex fund cycling: 50.66% of Spark borrowings flow to Aave, 41.21% of Aave borrowings flow back to Spark
• Trend Prediction: Institutional investors view DeFi as a channel for long-term capital allocation, no longer just pursuing short-term profits

Stablecoin Strategy Differentiation

• Institutional wallets borrowed $5.3 billion (Aave $3.96 billion, Spark $1.34 billion) demonstrating precise stablecoin strategies
• Time strategy differentiation: short-term holding on Aave (average 4.2 days) vs medium-term holding on Spark (average 13.8 days)
• Asset strategy differentiation: primarily borrowing USDT/USDC on Aave, focusing on DAI on Spark
• Trend Prediction: DAI is becoming the preferred asset for institutional investors’ medium to long-term DeFi strategies, while USDC/USDT are more commonly used for short-term arbitrage

Recommendations for Aave

Optimize Long-term Borrowing Experience - Consider introducing incentive mechanisms for long-term borrowers, such as interest rate discounts for borrowing periods exceeding specific thresholds. Explore long-term fixed-rate products to attract stability-oriented institutional users.

Strengthen Ecosystem Integration - Deepen integration with major DeFi protocols, especially seamless connections with trading platforms.

Protocol Design Considerations - Evaluate interest calculation mechanisms and their impact on different user behaviors, including potential optimizations for same-block transaction patterns.

Institutional User Strategy - Research institutional users’ borrowing fund flows and needs, develop more features targeting institutional requirements.

Data Source: Dune Analytics Dashboard
Dashboard covers lending events on Ethereum from Aave V3 protocol since January 27, 2023 and Spark Protocol since March 29, 2023 until May 7, 2025 at 23:59:59

[WalleDAO]

I’m particularly interested in hearing the community’s thoughts on the interest accrual mechanism and how different users interact with it. Has anyone else observed interesting usage patterns or have thoughts on how various interest calculation approaches might affect different user segments?

[GovernAvid]

Which interest-calculation approach: per-transaction accrual, a small flat flash-loan fee, or a hybrid rebate model, best balances preventing zero-cost borrowing with gas efficiency, especially when comparing mainnet to L2?

[WalleDAO]

Some considerations for discussion:

• Different lending protocols use various approaches to interest calculation - some use time-based models like Aave, while others implement per-transaction fees
• Each model creates different incentives for users and affects protocol revenue in unique ways
• For example, a proportional transaction fee might change the economics for flash loan-like operations
• The gas efficiency vs. revenue trade-off is particularly interesting when comparing L1 vs L2 implementations

[WalleDAO]

Update: I’ve submitted these findings through the official security channels as recommended by the team (Immunefi submission ID: 45460). I’m now working with the AAVE team to properly evaluate the impact of these block-level interest calculation patterns on protocol revenue.

[bgdlabs]

Hello @WalleDAO. As we commented to you in Immunefi, this is intended behaviour on the Aave v3 protocol: the system doesn’t have any type of borrow origination fee (akin to the one existing on Aave v1) as it was creating important negative dynamics, and borrow interest in the protocol accrues on a per-second basis, which means that during the same block, doesn’t accrue as time technically doesn’t “pass” in the VM.

Instant liquidity by borrowing at the same time is tending to be free more and more in DeFi, so given the complexity and probably negative product consequences of changing this behaviour to always have a fee on borrowings, it is more probable to align all mechanisms to zero fee than otherwise.

---

As a general reminder, even if not applicable in this case, not being a vulnerability, a public disclosure of what is believed by the submitter as a problem followed by a submission to the bug bounty program immediately disqualifies for bounty and it is non responsible security-wise.
The correct flow is always submit via the bug bounty program everything believed to be an issue to start with, and if happens to not be, follow with any public analysis if wanted.

[WalleDAO]

Thank you @bgdlabs for the clarification. I appreciate your explanation about the intended behavior of Aave v3’s interest accrual mechanism and understand this is by design rather than a vulnerability.

My research was intended to highlight interesting protocol usage patterns and potential economic considerations rather than suggest a security issue. I’ve learned a valuable lesson about the proper channels for different types of discussions.

I’m genuinely interested in protocol economics and optimization. Would you be open to discussing whether there might be any merit in exploring alternative interest models that could potentially enhance yield for liquidity providers while maintaining the protocol’s competitive position in the market? Or perhaps this current model is indeed optimal for Aave’s strategic positioning?

Thanks again for engaging with my analysis.