Evaluation
AuraScore 81/100

Quantitative Algorithmic Model Derivation Inspection Checklist

Inspect and stress-test autonomous quantitative reasoning agents generating stochastic financial models and numerical risk simulations.

Use this template when evaluating agents that independently formulate mathematical financial models, derive closed-form solutions, or design algorithmic risk engines. It validates numerical stability, asymptotic consistency, and financial logic integrity.

Template

Role: Senior Quantitative AI Risk and Validation Architect specializing in stochastic calculus, numerical methods, and algorithmic derivatives modeling.

Context

  • Model specification framework: {{model_specification_framework}}
  • Simulation scale and methodology: {{stochastic_simulation_scale}}
  • Parameter boundary conditions: {{parameter_boundary_conditions}}
  • Numerical stability threshold: {{numerical_stability_threshold}}
  • Backtesting benchmark suite: {{backtesting_benchmark_suite}}
  • Algorithmic explainability standard: {{explainability_standard}}

Task

Deliver an end-to-end technical evaluation checklist to inspect, stress-test, and validate an autonomous quantitative agent's mathematical modeling derivations, parameter calibrations, and numerical stability before production sandbox execution.

Method

  1. Establish structural verification gates for theoretical model assumptions within {{model_specification_framework}}.
  2. Detail explicit checks for stochastic calculus operations (e.g., Itô's lemma applications, martingale properties, drift-diffusion terms).
  3. Define boundary and limit condition stress-tests reflecting {{parameter_boundary_conditions}} (e.g., zero volatility, infinite horizon, negative interest rates).
  4. Formulate numerical stability validation checks under {{stochastic_simulation_scale}} to identify discretization errors and explosive gradients.
  5. Construct tests to verify convergence rates against {{numerical_stability_threshold}}.
  6. Design sanity checks comparing agent-derived closed-form solutions against standard baseline implementations in {{backtesting_benchmark_suite}}.
  7. Detail logic tests to identify arbitrage vulnerabilities, lookahead bias, or non-causal features in generated strategies.
  8. Establish compliance checks aligning the agent's step-by-step rationales with {{explainability_standard}}.

Constraints

  • MUST evaluate mathematical formulations for dimensional consistency across all parameters.
  • MUST NOT allow unconstrained optimization without explicit penalty terms or bounded domains.
  • Checklist items MUST isolate numerical discretization issues from theoretical model misspecifications.
  • Must enforce hard failure gates for any derivation violating no-arbitrage or conservation principles.

Output format

A detailed markdown verification checklist structured as:

  • Section 1: Theoretical Derivation & Stochastic Rigor (5-6 checklist items)
  • Section 2: Boundary Value & Asymptotic Limits Validation (4-5 checklist items)
  • Section 3: Numerical Stability & Simulation Precision (4-5 checklist items)
  • Section 4: Quantitative Integrity & Bias Prevention (4-5 checklist items)
  • Section 5: Regulatory Explainability & Audit Trail (3-4 checklist items)
  • Pre-deployment sign-off gate table with mandatory blocking criteria.

Self-review

  • Ensure checks distinguish between analytical solutions and numerical approximations.
  • Confirm that {{parameter_boundary_conditions}} and {{numerical_stability_threshold}} are explicitly tied to pass/fail items.
  • Check that financial logic checks prevent common modeling traps like lookahead bias and survivorship bias.
AuraScore breakdown
81/100Provisional
Instruction clarity15/15 · Strong

Explicit role, a named task, and discrete steps the model can follow.

Context architecture12/12 · Strong

Background, inputs and variables the model needs before it starts.

Constraint engineering10/12 · Adequate

Hard boundaries — what the model must and must not do.

Output specification6/14 · Thin

A named, field-level shape for the response.

Reasoning structure10/10 · Strong

Ordered work items that force analysis before an answer.

Model compatibility10/10 · Strong

Length and structure that travel across frontier models.

Token efficiency5/10 · Thin

Signal density — instruction weight without padding.

Reusability7/7 · Strong

Documented variables so the scaffold adapts to new inputs.

Robustness5/5 · Strong

Quality bar, assumptions and behaviour when inputs are thin.

Observed performance1/5 · Thin

How much real usage the template has behind it.

ai-agents
agents-evaluation
complex-reasoning-analysis-math
quantitative-finance
stochastic-calculus
model-risk