Dynamic Allocation Engine
Overview
Birch Hill Labs employs a quantitative, systematic framework to manage institutional-grade DeFi lending vaults. Our methodology moves beyond simple yield seeking, utilizing a proprietary allocation engine that solves for capital preservation and liquidity efficiency.
Our approach has three primary objectives:
Qualitative Assessment: A framework analyzing market microstructure.
Quantitative Limits: A CVaR (Collateral Value at Risk) model to set hard supply caps.
Dynamic Execution: A dual-layer rebalancing protocol to manage duration and utilization risk.
Before capital is allocated, every target market undergoes a rigorous liquidity stress test. We analyze the quality of that liquidity through the followers layers:
Liquidity Depth vs. Slippage: measuring the raw liquidity available for the collateral asset against potential liquidation volume. The available market liquidity must exceed outstanding debt by a safety multiple, ensuring that even if slippage spikes, the liquidation bonus provides enough buffer to incentivize third-party liquidators.
Concentration Risk (Supplier & Borrower): analyzing the Herfindahl-Hirschman Index (HHI) of the pool. We penalize markets where liquidity is provided by a small cohort of LPs or where borrowing is dominated by a few large whales.
Liquidation Clustering : mapping the liquidation prices of all active loans. If a large percentage of loans share the same liquidation price, a specific price dip can trigger a "death spiral" of cascading liquidations.
Oracle Fidelity & Spread: assessing the "freshness" of on-chain pricing relative to major CEX volume. We monitor the latency between off-chain price movements and on-chain updates to prevent toxic flow arbitrage.
Liquidity Stability: analyzing historical LP behavior during historical stress events. Does liquidity flee when volatility spikes (mercenary capital), or does it remain sticky? We prioritize markets where LPs have historically demonstrated commitment during drawdowns.
Market Maker SLAs & Backstops: evaluation of the presence of designated market makers. Markets with contractual obligations or signed SLAs for liquidation performance in "black swan" events receive higher allocation weights.
Lender Activity Trends: tracking the rate of change in collateralization. A market where lenders are actively topping up collateral or deleveraging voluntarily is viewed more favorably than a passive market relying solely on liquidation mechanisms.
Dynamic Rebalancing
Our allocation engine employs a dual-layer rebalancing strategy to manage utilization rates and respond to crisis events.
Systematic Rebalancing (Cron-Based)
Frequency: Daily We run a daily optimization looking for utilization & market inefficiencies, such as:
Utilization Smoothing: If BTC markets are at 100% utilization and ETH markets are at 20%, we reallocate to balance yield and liquidity access.
Cap Enforcement: If a market exceeds its optimal utilization, we bring it back to target
Internal Proprietary Optimizations: Birch Hill custom risk modeling inputs
Trigger-Based Rebalancing (Event-Driven)
Frequency: Continuous Monitoring (Block/Intraday) We utilize automated triggers that override the daily schedule when critical thresholds are breached, such as:
Utilization Spikes: If a pool creates a "liquidity crunch" (utilization spikes to near 100% rapidly), reallocate.
Liquidity Flight: If DEX liquidity drops by >25% intraday, the CVaR model instantly recalculates a lower supply cap, triggering a withdrawal.
Crisis Management
In moments of systemic stress (e.g., depegs, solvency scares), the engine shifts to a capital preservation mode:
Price Source Dislocation: We monitor the delta between CEX and DEX prices. If spreads widen significantly (indicating market maker failure), we pause new allocations.
Duration Risk Assessment: We actively de-allocate from "soft" markets (high concentration, low MM activity) and consolidate into "hard" assets (BTC/ETH/USDC) where we are comfortable holding duration risk.
Market Maker Communication: We verify MM activity across chains (e.g., Solana vs. Mainnet). If MMs pull bids (as seen in previous black swan events), we immediately contract our supply caps to zero for affected markets.
Last updated