How to Price Variable Risk

Our reference solver implements a sophisticated multi-factor rate computation pipeline (IRISSolverRateEngine) combining stochastic interest rate models, volatility forecasting, risk premia, and competitive bidding adjustments.

By studying this open-source reference pipeline, your team can construct your ideal pricing mechanisms:

1. CIR Rate Forecast (Cox-Ingersoll-Ross mean reversion)
2. Jump-Diffusion Component (Adding Poisson processes for sudden rate spikes)
3. GARCH(1,1) Volatility Estimation (Capturing conditional variance clustering)
4. Trend Analysis Module (MA crossover, momentum, high/low regime detection)
5. Self-Impact Adjustment (Slippage calculations for high-volume positions)
6. Duration Risk Premium
7. Negative Carry Premium (Black-76 option pricing variations)
8. Position Sizing (Half-Kelly Criterion calculations based on Bond Capital)

Example: CIR Forecast

The CIR model captures mean-reverting interest rate dynamics critical for calculating long-duration risk on pools like Aave.

Expected rate: r_cir = theta + (r_current - theta) * exp(-kappa * T)

Where:

• r_current: Current variable rate from the venue
• theta: Long-term mean rate (calibrated parameter, default 0.045 = 4.5%)
• kappa: Mean reversion speed (default 6.0 -- high speed = faster mean reversion)
• T: Loan duration in years