Intelligence Calibration System
Purpose: Use this file when running SHARPEN, validating predictions, tuning source reliability, or emitting reusable competitive patterns.
Contents
- SHARPEN loop
- Tracking schema
- Accuracy bands
- Source calibration
- Confidence by analysis type
- Propagation format
SHARPEN Loop
TRACK -> VALIDATE -> CALIBRATE -> PROPAGATE
Without SHARPEN, analyses decay into one-off opinions. With SHARPEN, Compete learns which sources, claims, and outputs actually matter.
TRACK
Record every material analysis.
Analysis: [analysis-id]
Type: [SWOT | Feature Matrix | Positioning | Battle Card | Alert Response | Market Trends]
Competitors_Analyzed: [list]
Sources_Used: [list with reliability tier]
Key_Predictions:
- prediction: [description]
confidence: [High/Medium/Low]
timeframe: [validation window]
Actionability: [immediately_actionable | needs_further_research | monitoring_only]
Downstream_Handoff: [Spark | Growth | Magi | None]Track at minimum:
| Data Point | Why |
|---|---|
| prediction accuracy | calibrates confidence |
| source reliability | tunes evidence weighting |
| actionability rate | tells whether outputs change decisions |
| downstream utilization | shows whether handoffs are usable |
| alert hit rate | tunes urgency thresholds |
VALIDATE
Accuracy Bands
| Accuracy | Interpretation | Default action |
|---|---|---|
> 0.80 |
strong analyst performance | maintain method |
0.60-0.80 |
acceptable with room to improve | inspect weak spots |
< 0.60 |
weak prediction quality | review sources and framing |
Validation Triggers
| Trigger | Check |
|---|---|
| feature launch | feature prediction accuracy |
| pricing change | price intelligence accuracy |
| trend materialization | market trend accuracy |
| win/loss update | battle card effectiveness |
| quarterly review | overall intelligence quality |
Summary Template
### Validation Report
| Metric | Value | Trend |
|---|---|---|
| Analyses produced | 12 | — |
| Predictions made | 8 | — |
| Predictions validated | 5 | — |
| Accuracy rate | 80% (4/5) | ↑ |
| Actionability rate | 67% (8/12) | — |
| Downstream utilization | 75% (6/8) | — |
Strongest area: [...]
Weakest area: [...]CALIBRATE
Default Reliability Weights
official_sources: 0.90
review_platforms: 0.75
financial_data: 0.85
job_postings: 0.65
community_signals: 0.60Calibration Rules
- require
3+data points before changing a source weight - maximum adjustment per cycle is
+/-0.15 - learned adjustments decay
10%per quarter toward defaults - explicit user source preferences override calibration
Confidence by Analysis Type
| Analysis Type | Typical Accuracy | Default Confidence |
|---|---|---|
| Feature Matrix | 85% |
High |
| Tech Stack | 90% |
High |
| SWOT | 70% |
Medium |
| Market Trends | 60% |
Medium |
| Pricing | 55% |
Low |
PROPAGATE
Journal Entry
## YYYY-MM-DD - SHARPEN: [Analysis Type]
Analyses validated: N
Overall accuracy: X%
Key insight: [...]
Calibration adjustment: [source/type: old -> new]
Apply when: [...]
reusable: true
<!-- EVOLUTION_SIGNAL
type: PATTERN
source: Compete
date: YYYY-MM-DD
summary: [intelligence insight]
affects: [Compete, Magi, Growth]
priority: MEDIUM
reusable: true
-->Pattern Library
| Pattern | Indicators | Typical Timeframe | Reliability |
|---|---|---|---|
| Pricing undercut | price reduction 10%+ |
1-3 months |
High |
| Feature convergence | 3+ competitors ship the same feature |
6-12 months |
Medium |
| Market consolidation | 2+ acquisitions in segment |
12-24 months |
Medium |
| Niche expansion | leader enters adjacent market | 3-6 months |
High |
| Talent drain | key hires move to competitor | 6-12 months |
Low |
Quick SHARPEN
Use this when there are too few predictions to recalibrate.
## Quick SHARPEN
Analyses: 2
Predictions: 1
Note: insufficient data to recalibrate
Action: keep current weights