Forecasting Techniques

Forecasting Techniques

Metadata

• Name: forecasting-techniques
• Description: Multiple methods for projecting future values
• Triggers: forecasting, projections, growth rate, CAGR, market prediction

Instructions

Apply forecasting techniques to project $ARGUMENTS into the future.

Choose appropriate method based on data availability and context.

Framework

Three Main Approaches

| Method | Data Required | Time Horizon | Precision | Best For | |----------|----------------|--------------|------------| | Time Series Extrapolation | 5-10 years of historical | Short-medium | High | Stable environments | | Derived Demand | Proxy variables, cross-correlation | Short-medium | Medium | Related markets | | Expert Opinion | Structured surveys | Any | Low | New products |

1. Time Series Extrapolation

Trend Analysis

• Simple growth rate: Compound annual growth (CAGR)
• Linear regression: Straight line fit to historical data
• Moving average: Smooths volatility, lags trends
• Exponential smoothing: Recent trends weighted more heavily

Steps:

1. Gather historical data (3+ years preferred)
2. Analyze patterns (cycles, seasonality, trends)
3. Choose model (CAGR, regression, etc.)
4. Apply to future periods
5. Validate against expert opinion

Example Output:

Year | Historical | Projected | Growth Rate |
|------|------------|------------|-------------|
| 2023 | $100 M | - | - |
| 2024 | $115 M | +15% | CAGR = 15% |
| 2025 | $132 M | +15% | CAGR = 15% |
| 2026 | $152 M | +15% | CAGR = 15% |
| 2027 | $175 M | +15% | CAGR = 15% |

2. Derived Demand

Proxy Methodology

• Identify proxy variable that correlates with demand
• Use readily available data with reliable trend
• Apply correlation coefficient
• Adjust for unique factors

Examples:

• GDP growth as proxy for consumer spending
• Housing starts as proxy for home goods
• Demographics for category-specific demand

Steps:

1. Identify correlation (r² should be > 0.5)
2. Gather proxy data
3. Apply coefficient
4. Adjust for local factors
5. Add confidence intervals

3. Expert Opinion

Structured Survey Method

• Multiple expert interviews
• Weighted by expertise or track record
• Delphi technique (iterative rounds)
• Scenario-based questioning

Advantages:

• Captures qualitative insights
• Accounts for disruptive changes
• Incorporates expert judgment

Process:

1. Define forecasting questions
2. Select experts (diverse backgrounds)
3. Conduct interviews (structured format)
4. Aggregate with weighting
5. Present scenarios (base, optimistic, pessimistic)
6. Review and iterate if needed

Output Process

1. Define scope - What's being forecasted?
2. Select method - Based on data and time horizon
3. Gather inputs - Historical data, drivers, expert inputs
4. Apply technique - Run the chosen method
5. Calculate projections - For each year/period
6. Validate - Cross-check with other methods
7. Add scenarios - Best, base, worst case
8. Document assumptions - Clearly state all key inputs

Output Format

## Forecasting Analysis: [Subject]

### Forecast Methodology

**Method Used:** [Time Series/Derived Demand/Expert Opinion]
**Time Horizon:** [Years]
**Base Year:** [Year]
**Data Quality:** [High/Medium/Low]

---

### Projections

| Metric | 2024 | 2025 | 2026 | 2027 | 2028 | CAGR |
|--------|--------|--------|--------|--------|--------|------|
| Revenue | $X M | $Y M | $Z M | $W M | $V M | % |
| Growth | X% | Y% | Z% | W% | % |

---

### Key Drivers

| Driver | Impact | Uncertainty | Scenario Impact |
|--------|---------|-----------------|--------------|
| [Driver 1] | High | Medium | [Description] |
| [Driver 2] | Medium | Low | [Description] |
| [Driver 3] | Low | High | [Description] |

---

### Scenarios

| Scenario | 2028 Revenue | Probability | Key Assumptions |
|----------|----------------|------------------|----------------|
| **Base** | $X M | 50% | [Assumptions] |
| **Optimistic** | $Y M | 30% | [Assumptions] |
| **Pessimistic** | $W M | 70% | [Assumptions] |

---

### Confidence Intervals

| Metric | Low | Base | High | Confidence |
|--------|------|------|------|------|----------|
| 2028 Revenue | $X ± Y% | $Z M | $W M | 80% |

Tips

• Triangulate methods when possible
• Use multiple methods for cross-validation
• Be explicit about assumptions - don't hide them
• Present confidence intervals for transparency
• Consider mean reversion - growth rates tend toward averages
• Validate with real outcomes when available
• Document track record of forecasts - improve over time

References

• Makridakis, Spyros. Business Forecasting. 1998.
• Armstrong, J. Scott. Principles of Forecasting. 2001.
• Wikipedia. "Forecasting - Methods and Applications" (multiple sources)

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: linuszz
• Source: linuszz/business-strategy-planning-skills
• License: MIT

Install and usage instructions live in the source repository linked above.