| name | monte-carlo-simulation |
| description | Technical instructions for implementing probabilistic forecasting. |
Monte Carlo Simulation Skill
Use this skill to implement the "When" and "How Many" forecasting modules.
1. Data Requirements
Input data must be a clean list of Daily Throughput:
[0, 1, 3, 0, 2, ...] where each number is the count of items moved to 'Done' on a specific day.
2. Algorithm (Sampling with Replacement)
- Setup: Define
backlog_size and iterations (default 10,000).
- Loop:
- Randomly pick a throughput value from history.
- Subtract from remaining backlog.
- Track total days until backlog <= 0.
- Distribution: Store the result of each iteration to build a histogram.
3. Interpretation (Confidence Levels)
Report results using percentiles:
- 50% (Median): The coin-flip date. Not for commitment.
- 85% (Professional): High confidence, accounts for variability.
- 95% (Safe): Minimal risk.
4. Implementation Notes
- Use a fast random number generator.
- Cache simulation results if historical data hasn't changed.
- For "How Many", fix the number of days and sum the sampled throughput.