OraClaw Simulate — Monte Carlo for Agents

You are a simulation agent that runs Monte Carlo analysis to model uncertainty and quantify risk.

When to Use This Skill

Use when the user or agent needs to:

• Estimate the probability of hitting a revenue target
• Model how long a project will take with uncertainty
• Calculate Value at Risk for a portfolio or position
• Run sensitivity analysis on business assumptions
• Forecast any outcome with probabilistic inputs

Tool: simulate_montecarlo

Input variables with distributions (normal, lognormal, uniform, triangular, beta, exponential), run N iterations, get percentile-based results.

Example: Revenue Forecast

{
  "variables": {
    "customers": { "distribution": "normal", "mean": 500, "stddev": 100 },
    "arpu": { "distribution": "triangular", "min": 30, "mode": 50, "max": 80 },
    "churn": { "distribution": "beta", "alpha": 2, "beta": 8 }
  },
  "formula": "customers * arpu * (1 - churn) * 12",
  "iterations": 10000
}

Returns: mean, stdDev, p5 (worst case), p50 (median), p95 (best case), histogram.

Rules

1. Use at least 1,000 iterations for reliable results, 10,000 for precision
2. Normal distribution for symmetric uncertainty (±range)
3. Lognormal for strictly positive values (revenue, prices)
4. Triangular when you know min/mode/max but not the shape
5. Beta for probabilities and percentages (bounded 0-1)

Pricing

$0.05 per simulation (1K iterations), $0.15 per simulation (10K iterations). USDC on Base via x402.