Metadata

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

Instructions

Choose appropriate method based on data availability and context.

Framework

Three Main Approaches

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:

• Gather historical data (3+ years preferred)
• Analyze patterns (cycles, seasonality, trends)
• Choose model (CAGR, regression, etc.)
• Apply to future periods
• 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:

• Identify correlation (r² should be > 0.5)
• Gather proxy data
• Apply coefficient
• Adjust for local factors
• 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:

• Define forecasting questions
• Select experts (diverse backgrounds)
• Conduct interviews (structured format)
• Aggregate with weighting
• Present scenarios (base, optimistic, pessimistic)
• Review and iterate if needed

Output Process

• Define scope - What's being forecasted?
• Select method - Based on data and time horizon
• Gather inputs - Historical data, drivers, expert inputs
• Apply technique - Run the chosen method
• Calculate projections - For each year/period
• Validate - Cross-check with other methods
• Add scenarios - Best, base, worst case
• 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)