scenario-analyzer

Headline Scenario Analyzer

Overview

This skill analyzes medium- to long-term (18-month) investment scenarios starting from news headlines. It sequentially calls two specialized agents (scenario-analystとstrategy-reviewer), and generates a comprehensive report that integrates multifaceted analysis and critical review.

When to Use This Skill

Use this skill in the following cases:

• You want to analyze medium- to long-term investment impacts from news headlines
• You want to construct multiple scenarios for 18 months later
• You want to organize impacts on sectors and stocks by primary/secondary/tertiary
• You need comprehensive analysis including a second opinion
• You need report output in Japanese

Usage example:

/headline-scenario-analyzer "Fed raises interest rates by 50bp, signals more hikes ahead"
/headline-scenario-analyzer "China announces new tariffs on US semiconductors"
/headline-scenario-analyzer "OPEC+ agrees to cut oil production by 2 million barrels per day"

Architecture

┌─────────────────────────────────────────────────────────────────────┐
│                    Skill（オーケストレーター）                        │
│                                                                      │
│  Phase 1: 準備                                                       │
│  ├─ ヘッドライン解析                                                  │
│  ├─ イベントタイプ分類                                                │
│  └─ リファレンス読み込み                                              │
│                                                                      │
│  Phase 2: エージェント呼び出し                                        │
│  ├─ scenario-analyst（主分析）                                       │
│  └─ strategy-reviewer（セカンドオピニオン）                           │
│                                                                      │
│  Phase 3: 統合・レポート生成                                          │
│  └─ reports/scenario_analysis_<topic>_YYYYMMDD.md                   │
└─────────────────────────────────────────────────────────────────────┘

Workflow

Phase 1: Preparation

Step 1.1: Headline analysis

Analyze the headline input by the user.

1. Headline confirmation

   • Check whether a headline has been passed as an argument
   • If it has not been passed, ask the user for input

2. Keyword extraction

   • Main entities (company names, country names, institution names)
   • Numerical data (interest rates, prices, quantities)
   • Actions (raise, lower, announce, agree, etc.)

Step 1.2: Event type classification

Classify the headline into the following categories:

Category	Example
Monetary policy	FOMC, ECB, Bank of Japan, rate hike, rate cut, QE/QT
Geopolitics	War, sanctions, tariffs, trade friction
Regulation and policy	Environmental regulation, financial regulation, antitrust law
Technology	AI, EV, renewable energy, semiconductors
Commodities	Crude oil, gold, copper, agricultural products
Companies and M&A	Acquisition, bankruptcy, earnings, industry restructuring

Step 1.3: Load references

Based on the event type, load the relevant references:

Read references/headline_event_patterns.md
Read references/sector_sensitivity_matrix.md
Read references/scenario_playbooks.md

Reference contents:

• headline_event_patterns.md: Past event patterns and market reactions
• sector_sensitivity_matrix.md: Event x sector impact matrix
• scenario_playbooks.md: Templates and best practices for scenario construction

---

Phase 2: Agent invocation

Step 2.1: Invoke scenario-analyst

Use the Task tool to invoke the main analysis agent.

Task tool:
- subagent_type: "scenario-analyst"
- prompt: |
    以下のヘッドラインについて18ヶ月シナリオ分析を実行してください。

    ## 対象ヘッドライン
    [入力されたヘッドライン]

    ## イベントタイプ
    [分類結果]

    ## リファレンス情報
    [読み込んだリファレンスの要約]

    ## 分析要件
    1. WebSearchで過去2週間の関連ニュースを収集
    2. Base/Bull/Bearの3シナリオを構築（確率合計100%）
    3. 1次/2次/3次影響をセクター別に分析
    4. ポジティブ/ネガティブ影響銘柄を各3-5銘柄選定（米国市場のみ）
    5. 全て日本語で出力

Expected output:

• List of related news articles
• Details of 3 scenarios (Base/Bull/Bear)
• Sector impact analysis (primary/secondary/tertiary)
• Recommended stock list

Step 2.2: Calling strategy-reviewer

Based on the analysis results from scenario-analyst, call the review agent.

Task tool:
- subagent_type: "strategy-reviewer"
- prompt: |
    以下のシナリオ分析をレビューしてください。

    ## 対象ヘッドライン
    [入力されたヘッドライン]

    ## 分析結果
    [scenario-analystの出力全文]

    ## レビュー要件
    以下の観点でレビューを実施：
    1. 見落とされているセクター/銘柄
    2. シナリオ確率配分の妥当性
    3. 影響分析の論理的整合性
    4. 楽観/悲観バイアスの検出
    5. 代替シナリオの提案
    6. タイムラインの現実性

    建設的かつ具体的なフィードバックを日本語で出力してください。

Expected output:

• Pointing out oversights
• Opinions on scenario probabilities
• Pointing out biases
• Proposals for alternative scenarios
• Final recommendations

---

Phase 3: Integration and Report Generation

Step 3.1: Integration of Results

Integrate the output from both agents and create the final investment decision.

Integration points:

1. Compensate for oversights pointed out in the review
2. Adjust probability allocation (if necessary)
3. Final decision taking biases into account
4. Formulation of a concrete action plan

Step 3.2: Report Generation

Generate the final report in the following format and save it to a file.

Save location: reports/scenario_analysis_<topic>_YYYYMMDD.md

# ヘッドライン・シナリオ分析レポート

**分析日時**: YYYY-MM-DD HH:MM
**対象ヘッドライン**: [入力されたヘッドライン]
**イベントタイプ**: [分類カテゴリ]

---

## 1. 関連ニュース記事
[scenario-analystが収集したニュースリスト]

## 2. 想定シナリオ概要（18ヶ月後まで）

### Base Case（XX%確率）
[シナリオ詳細]

### Bull Case（XX%確率）
[シナリオ詳細]

### Bear Case（XX%確率）
[シナリオ詳細]

## 3. セクター・業種への影響

### 1次的影響（直接的）
[影響テーブル]

### 2次的影響（バリューチェーン・関連産業）
[影響テーブル]

### 3次的影響（マクロ・規制・技術）
[影響テーブル]

## 4. ポジティブ影響が見込まれる銘柄（3-5銘柄）
[銘柄テーブル]

## 5. ネガティブ影響が見込まれる銘柄（3-5銘柄）
[銘柄テーブル]

## 6. セカンドオピニオン・レビュー
[strategy-reviewerの出力]

## 7. 最終投資判断・示唆

### 推奨アクション
[レビューを踏まえた具体的アクション]

### リスク要因
[主要リスクの列挙]

### モニタリングポイント
[フォローすべき指標・イベント]

---
**生成**: scenario-analyzer skill
**エージェント**: scenario-analyst, strategy-reviewer

Step 3.3: Saving the Report

1. reports/ If the directory does not exist, create it
2. scenario_analysis_<topic>_YYYYMMDD.md Save as (e.g.: scenario_analysis_venezuela_20260104.md）
3. Notify the user that saving is complete
4. Do not save directly in the project root

---

Resources

References

• references/headline_event_patterns.md - Event patterns and market reactions
• references/sector_sensitivity_matrix.md - Sector sensitivity matrix
• references/scenario_playbooks.md - Scenario construction template

Agents

• scenario-analyst - Main scenario analysis
• strategy-reviewer - Second opinion/review

---

Important Notes

Language

• All analysis and output shall be conducted inJapaneseconducted in
• Maintain stock tickers in English notation

Target market

• Stock selection isUS market-listed stocks only
• Including ADRs

Time horizon

• Scenarios target18 monthstarget
• Describe in 3 phases: 0-6 months/6-12 months/12-18 months

Probability allocation

• Base + Bull + Bear = 100%
• Describe the probability of each scenario along with its rationale

Second opinion

• Requiredexecute with (always call strategy-reviewer)
• Reflect review results in the final decision

Output destination (important)

• Always reports/ Must be saved under the directory
• Path: reports/scenario_analysis_<topic>_YYYYMMDD.md
• Example: reports/scenario_analysis_fed_rate_hike_20260104.md
• reports/ Create the directory if it does not exist
• Must not save directly in the project root

---

Quality Checklist

Before completing the report, confirm the following:

• Whether the headline has been correctly parsed
• Whether the event type classification is appropriate
• Whether the total probability of the 3 scenarios is 100%
• Whether there is a logical connection among 1st/2nd/3rd-order impacts
• Whether there is a concrete rationale for stock selection
• Whether the strategy-reviewer's review is included
• Is the final decision based on the review stated?
• Was the report saved to the correct path?