Skip to content

scenario-forecast.md

Latest commit

 

History

History
381 lines (271 loc) · 18.2 KB

scenario-forecast.md

File metadata and controls

381 lines (271 loc) · 18.2 KB
Raw
long-horizon-mode false

🔮 Scenario Forecast Template — Probability-Weighted Futures

📌 Template Instructions: Copy to analysis/daily/{date}/{article-type}-run{N}/intelligence/scenario-forecast.md. Develop 3-5 forward scenarios (standard mode) or ≥6 scenarios (long-horizon mode) with probability weights, early-warning indicators, and trigger events. See methodologies/per-artifact-methodologies.md §scenario-forecast.

🎯 Purpose: Forward-looking analysis mapping plausible futures for the next 7/30/90 days (standard) or 12m/term-end/EP-election (long-horizon). Each scenario includes probability, narrative, early-warning indicators, trigger events, and stakeholder impact.

0️⃣ Long-Horizon Mode

long-horizon-mode is an optional frontmatter flag that marks a run as long-horizon in the template. Set it to true when writing a term-outlook or election-cycle analysis run. However, for validator-enforced long-horizon article types — term-outlook and election-cycle — Stage C still applies the long-horizon rules from manifest.articleType even if this flag is unset or false. Standard behaviour (3–5 scenarios, 7/30/90-day horizon) is preserved only for article types outside that enforced set.

For long-horizon runs — whether triggered by long-horizon-mode: true or by validator-enforced manifest.articleType — the following requirements override the standard rules:

Requirement Standard mode Long-horizon mode
Minimum scenario count ≥ 3 ≥ 6
EP-election outcome branch Optional Mandatory (centre-right / centre-left / fragmented coalition)
Per-scenario WEP confidence band Required Required — drawn from the decay table in forward-projection-methodology.md §3
Regime-change branch When any tripwire fires Mandatory (at least 1 dedicated branch)
Wildcard/black-swan branches Recommended ≥ 2 mandatory
Anti-pattern "5+ scenarios" Applies Overridden — cap lifted to ≥ 6 main + residual

Activation signals (checked by Stage-C validator)

The validator enforces the ≥ 6 scenario floor when the manifest's articleType is one of the configured types: term-outlook, election-cycle. Registry fields such as electoralOverlay === true and scenarioMaxHorizonMonths >= 36 explain why those types are included, but are not the fields the validator directly inspects.

EP-election outcome branches (mandatory in long-horizon mode)

The three mandatory EP-election branches are:

Branch Label Probability range
Centre-right majority EPP + ECR + ID ≥ 360 seats Derive from seat-projection.md
Centre-left majority S&D + RE + Greens/EFA ≥ 360 seats Derive from seat-projection.md
Fragmented coalition No bloc ≥ 360 — EP-wide grand coalition needed Often highest probability in long horizon

Each EP-election branch must carry:

  • A WEP band drawn from the T+EP-election ±6m row of the decay table (Highly Unlikely → Almost No Chance; see forward-projection-methodology.md §3).
  • A reference class from historical-baseline.md §Electoral baselines (≥ 3 EP-term analogues).
  • ≥ 3 early-warning indicators with ISO-date thresholds.

📋 Document Metadata

Field Value
Report ID [REQUIRED: SF-YYYY-MM-DD-runNN]
Analysis Date [REQUIRED: YYYY-MM-DD]
Horizon [REQUIRED: 7 days / 30 days / 90 days / 12m / term-end / EP-election]
Scenarios Developed [REQUIRED: count 3-5 (standard) or ≥6 (long-horizon)]
Long-Horizon Mode [REQUIRED for term-outlook and election-cycle articleTypes: true / false]
Confidence [REQUIRED: 🟢/🟡/🔴]

1️⃣ Horizon Statement

Timeframe covered: [REQUIRED: e.g. "Next 30 days (2026-05-15 to 2026-06-15)"]

Forecast window rationale: [REQUIRED: ≥60 words explaining why this horizon was chosen — e.g. proximity to next plenary session, pending Commission decision, Council deadline, etc.]

2️⃣ Baseline Assumption

Current-state claim from which scenarios branch:

[REQUIRED: ≥150 words stating the single most important current-state political fact or condition that all scenarios share as their starting point. Example: "EPP-S&D coalition currently holds 55% of votes but faces internal fracture on Green Deal implementation." Cite evidence.]

3️⃣ Scenarios

%%{init: {"theme":"dark","themeVariables":{"primaryColor":"#1565C0","primaryTextColor":"#ffffff","primaryBorderColor":"#0A3F7F","lineColor":"#90CAF9","secondaryColor":"#2E7D32","secondaryTextColor":"#ffffff","tertiaryColor":"#FF9800","tertiaryTextColor":"#000000","mainBkg":"#1565C0","secondBkg":"#2E7D32","tertiaryBkg":"#FF9800","noteBkgColor":"#FFC107","noteTextColor":"#000000","errorBkgColor":"#D32F2F","fontFamily":"Inter, Helvetica, Arial, sans-serif"}}}%%
flowchart TD
    BASELINE[Baseline:<br/>[CURRENT STATE]]
    
    BASELINE -->|[probability %]| S1[Scenario 1:<br/>[NAME]]
    BASELINE -->|[probability %]| S2[Scenario 2:<br/>[NAME]]
    BASELINE -->|[probability %]| S3[Scenario 3:<br/>[NAME]]
    
    S1 -->|indicator| EW1A[Early Warning:<br/>[indicator 1]]
    S1 -->|indicator| EW1B[Early Warning:<br/>[indicator 2]]
    S1 -->|trigger| T1[Trigger:<br/>[event + date]]
    
    S2 -->|indicator| EW2A[Early Warning:<br/>[indicator 1]]
    S2 -->|trigger| T2[Trigger:<br/>[event + date]]
    
    S3 -->|indicator| EW3A[Early Warning:<br/>[indicator 1]]
    S3 -->|trigger| T3[Trigger:<br/>[event + date]]
    
    style BASELINE fill:#1565C0,color:#ffffff
    style S1 fill:#2E7D32,color:#ffffff
    style S2 fill:#FF9800,color:#000000
    style S3 fill:#D32F2F,color:#ffffff
Loading

Scenario 1: [REQUIRED: Name]

Probability: [REQUIRED: %]
Likelihood: [REQUIRED: Most Likely / Plausible / Tail Risk]

Narrative

[REQUIRED: ≥150 words describing how this scenario unfolds. Include named actors, specific procedures, coalitions, and decision points. What changes from the baseline? What sequence of events leads to this outcome? Cite at least one procedure ID or plenary date.]

Early-Warning Indicators

  1. [REQUIRED: specific, measurable indicator with threshold — e.g. "EPP cohesion on ENVI votes drops below 85%"]
  2. [REQUIRED: indicator 2]
  3. [REQUIRED: indicator 3]

≥3 indicators required per scenario.

Trigger Events

  1. [REQUIRED: named event with date bound — e.g. "2026-06-03: ENVI committee vote on regulation XYZ"]
  2. [REQUIRED: trigger event 2 with date]

≥2 date-bounded trigger events required per scenario.

Stakeholder Impact Summary

Stakeholder Impact Explanation
[REQUIRED: actor name] [🟢 Positive / 🟡 Mixed / 🔴 Negative] [REQUIRED: one-line]
[REQUIRED] [...] [REQUIRED]
[REQUIRED] [...] [REQUIRED]

Scenario 2: [REQUIRED: Name]

Probability: [REQUIRED: %]
Likelihood: [REQUIRED: Most Likely / Plausible / Tail Risk]

Narrative

[REQUIRED: ≥150 words]

Early-Warning Indicators

  1. [REQUIRED]
  2. [REQUIRED]
  3. [REQUIRED]

Trigger Events

  1. [REQUIRED: event + date]
  2. [REQUIRED: event + date]

Stakeholder Impact Summary

Stakeholder Impact Explanation
[REQUIRED] [...] [REQUIRED]
[REQUIRED] [...] [REQUIRED]

Scenario 3: [REQUIRED: Name]

Probability: [REQUIRED: %]
Likelihood: [REQUIRED: Most Likely / Plausible / Tail Risk]

Narrative

[REQUIRED: ≥150 words]

Early-Warning Indicators

  1. [REQUIRED]
  2. [REQUIRED]
  3. [REQUIRED]

Trigger Events

  1. [REQUIRED: event + date]
  2. [REQUIRED: event + date]

Stakeholder Impact Summary

Stakeholder Impact Explanation
[REQUIRED] [...] [REQUIRED]
[REQUIRED] [...] [REQUIRED]

4️⃣ Probability Check

Scenario Probability
Scenario 1 [%]
Scenario 2 [%]
Scenario 3 [%]
Total [REQUIRED: must = 100%]

5️⃣ Cross-Scenario Sensitivity

Single variable whose movement flips probability weights:

[REQUIRED: ≥100 words identifying the one factor (political, economic, procedural, or external) that most determines which scenario materializes. Examples: "EPP internal cohesion on climate votes", "Commission timing of Next Generation EU disbursement", "Council agreement on Treaty article invocation". Explain how changes in this variable redistribute scenario probabilities.]

6️⃣ Monitoring Plan

Check Date What to Monitor Decision Point
[REQUIRED: YYYY-MM-DD] [REQUIRED: specific indicator or event] [REQUIRED: what decision this informs]
[REQUIRED: YYYY-MM-DD] [REQUIRED] [REQUIRED]
[REQUIRED: YYYY-MM-DD] [REQUIRED] [REQUIRED]

Next scenario update recommended: [REQUIRED: date or trigger event]

7️⃣ Data Sources & Confidence

EP MCP tools used: track_legislation, monitor_legislative_pipeline, analyze_coalition_dynamics, get_plenary_sessions, get_procedure_events

Confidence level: [REQUIRED: 🟢 HIGH / 🟡 MEDIUM / 🔴 LOW]
Confidence rationale: [REQUIRED: where probabilities are data-backed vs. expert judgment]

🛠️ Worked example — three-scenario forecast on Critical Raw Materials Act

Scenario A — Adoption Q4 2026 in original scope (40%)

Driver: Council and EP align on "strategic raw material" definitions. Indicators that confirm: trilogue rounds <4; Council position unchanged from December 2025; rapporteur's report adopted by IMCO with ≥80% margin.

Scenario B — Adoption Q1 2027 with softened scope (35%)

Driver: industry mobilisation forces threshold raises during trilogue. Indicators that confirm: ≥3 industry letters during trilogue; 2-3 amendments narrowing strategic-supplier definitions; Council adds implementation-cost rider.

Scenario C — Adoption delayed past 2027 (20%)

Driver: HUN veto in Council on financing model OR CJEU referral on legal base. Indicators that confirm: HUN bilateral signals; legal challenge filed; rapporteur withdrawn or replaced; Commission tables revised proposal.

Scenario D — Adoption fails entirely (5%, residual)

Driver: external geopolitical shock (China-EU rare-earths embargo freezes negotiations, or unrelated political crisis displaces agenda). Indicators: China rare-earths export controls escalate; major EU MS political crisis dominates Council agenda.

Probability sum: 100%. Scenarios are exhaustive and mutually exclusive.

🚫 Anti-patterns — scenario-forecast failures

Anti-pattern Why it fails Correct approach
Probabilities sum to ≠ 100% Not a coherent forecast Re-normalise; document residual
Two scenarios that overlap Not mutually exclusive Distinguish on a single decisive variable
Probabilities all equal (33/33/33) "I don't know" disguised Differentiate; if uniform, cite reasoning
Scenario without driver Cannot be tested Each scenario has ≥1 named causal driver
Scenario without indicator Cannot be monitored Each scenario has ≥3 falsifiable indicators
Probabilities outside WEP grid Non-standard Use WEP bands from forward-projection-methodology.md §3
Update without version-stamp Loses audit trail Date-stamp each forecast revision
5+ scenarios (standard mode) Too granular for human interpretation Cap at 3-4 main + 1 residual unless long-horizon-mode active
< 6 scenarios (long-horizon mode) Insufficient coverage for multi-year horizon Must have ≥6 including EP-election branch + regime-change + 2 wildcards
Missing EP-election branch in long-horizon Omits dominant structural variable Add centre-right / centre-left / fragmented branches per §0
"Black swan" scenario in main set Misclassification Black swans → wildcards-blackswans.md; in long-horizon, 2 wildcard branches are mandatory
Forecast without monitoring plan Cannot be revised §6 monitoring plan with dates and decision points
Long-horizon WEP band out of range Violates decay table T+EP-election bands must use "Highly Unlikely" to "Almost No Chance" per forward-projection-methodology.md §3

🎯 EP MCP tool inputs

Tool Used for
track_legislation Procedure status feeds Scenario A/B distinction
monitor_legislative_pipeline Bottleneck signals feed Scenario C
analyze_coalition_dynamics Coalition-arithmetic sensitivity
get_voting_records Discipline trend across time
get_plenary_sessions Calendar constraint
get_procedure_events Trilogue round count

🔗 Controlling methodology cross-references

✅ Stage-C completeness signals

Standard mode (long-horizon-mode: false):

  • Line floor: 280 lines
  • ≥ 3 main scenarios with WEP-band probabilities summing to 100%
  • Each scenario has ≥1 named driver and ≥3 monitorable indicators
  • §5 cross-scenario sensitivity (single pivot variable)
  • §6 monitoring plan with at least 3 dated decision points
  • Confidence label per scenario (🟢/🟡/🔴) with rationale

Long-horizon mode (long-horizon-mode: true, with Stage-C scenario-count enforcement for term-outlook or election-cycle article types):

  • Line floor: 360 lines (term-outlook) / 400 lines (election-cycle)
  • ≥ 6 scenarios — enforced by Stage-C validator when article type is term-outlook or election-cycle
  • Mandatory EP-election outcome branch (centre-right / centre-left / fragmented coalition) per §0
  • At least 1 dedicated regime-change branch
  • At least 2 wildcard/black-swan branches
  • Per-scenario WEP confidence band drawn from forward-projection-methodology.md §3 decay table
  • Each EP-election branch carries reference class from historical-baseline.md §Electoral baselines

Document Control: /analysis/daily/{date}/{type}-run{N}/intelligence/scenario-forecast.md · Template v1.3 · Standard depth floor: 280 lines · Long-horizon depth floor: 360–400 lines.