/magi

Purpose: Use FORESIGHT after strategy work to track forecast quality, recalibrate heuristics, and share reusable patterns without overreacting to thin data.

Contents

• Workflow
• Accuracy and bracket thresholds
• Calibration rules
• Default assumption library
• Propagation format

Strategic Calibration System (FORESIGHT)

FORESIGHT = TRACK -> VALIDATE -> CALIBRATE -> PROPAGATE

Use this after a simulation, quarterly review, or any point where outcomes can be compared with predictions.

Workflow

TRACK

Record each engagement in a compact structure:

Simulation: [simulation-id]
Type: [SWOT | PESTLE | Porter | BCG | BSC | Ansoff | Full Strategy | KPI Forecast | M&A Evaluation | Crisis Response]
Horizon: [SHORT(0-1yr) | MID(1-3yr) | LONG(3-10yr)]
Frameworks_Applied: [list]
Scenarios_Generated: [count]
Key_Predictions:
  - prediction: [description]
    metric: [measurable KPI]
    baseline_value: [current]
    predicted_value: [target]
    confidence: [High/Medium/Low]
    timeframe: [validation date]
Assumptions_Used:
  - assumption: [description]
    source: [data/industry_default/estimate]
    sensitivity: [High/Medium/Low]
Data_Completeness: [Tier 1 only | Tier 1+2 | Tier 1+2+3 | Full]
Downstream_Handoff: [Magi/Scribe/Sherpa/Canvas/None]

Track at minimum:

VALIDATE

Accuracy Thresholds

Accuracy = predictions within ±15% of actual / total validated predictions

> 0.75      strong
0.50-0.75   moderate
< 0.50      weak

Scenario Bracket Thresholds

Bracket Rate = actuals inside optimistic-pessimistic range / total scenarios

> 0.85      well-calibrated
0.70-0.85   acceptable
< 0.70      too narrow or biased

Validation Triggers

Validation Snapshot

### Strategic Validation

| Metric | Value | Trend |
|--------|-------|-------|
| Simulations completed | 8 | — |
| Predictions made | 15 | — |
| Predictions validated | 10 | — |
| Accuracy rate (±15%) | 70% (7/10) | ↑ |
| Scenario bracket rate | 80% (8/10) | — |
| Framework combinations used | 5 | — |
| Downstream utilization | 88% (7/8) | — |

CALIBRATE

Calibration Rules

1. Require 3+ simulations before changing framework effectiveness weights.
2. Cap any single adjustment at ±0.15.
3. Apply 10% decay per quarter back toward defaults.
4. User-stated framework preferences always override calibrated defaults.
5. If there are fewer than 3 validated predictions, record the result but do not change weights.

Framework Effectiveness

Use effectiveness scores by context, then tune cautiously:

swot_effectiveness:
  overall_assessment: 0.85
  startup_strategy: 0.80
  m_and_a: 0.75
pestle_effectiveness:
  market_entry: 0.90
  long_term_planning: 0.85
  crisis_response: 0.70
porter_effectiveness:
  industry_analysis: 0.90
  competitive_positioning: 0.85
  pricing_strategy: 0.75
bcg_effectiveness:
  portfolio_management: 0.90
  investment_allocation: 0.85
  product_strategy: 0.80

Example:

bsc_effectiveness:
  startup_strategy: 0.65 -> 0.80

Scenario Parameter Defaults

Default Assumption Library (2026 refresh)

PROPAGATE

Journal Format

Write reusable findings to .agents/helm.md:

## YYYY-MM-DD - FORESIGHT: [Simulation Type]

**Simulations assessed**: N
**Overall accuracy**: X%
**Key insight**: [description]
**Calibration adjustment**: [framework/parameter: old -> new]
**Apply when**: [future scenario]
**reusable**: true

<!-- EVOLUTION_SIGNAL
type: PATTERN
source: Helm
date: YYYY-MM-DD
summary: [strategic insight]
affects: [Helm, Magi]
priority: MEDIUM
reusable: true
-->

Pattern Library

Quick FORESIGHT

Use this when evidence is too thin to recalibrate:

## Quick FORESIGHT

**Simulations**: 1 completed
**Predictions**: 2 (too few to calibrate)
**Note**: PESTLE -> SWOT produced clear strategic direction
**Action**: No weight change (insufficient data)