The Intelligence Layer
Institutional Investors Were Missing

Traditional risk models ignore public anger signals — until it's too late. We extract early-warning downside signals from global media before they reach price action.

従来のリスクモデルは世論の怒りシグナルを無視します。私たちは株価に反映される前に、グローバルメディアから早期警告シグナルを抽出します。


GDELT Project — 65,000+ Global Sources

The GDELT Project is the world's largest open-access event and news archive, monitoring print, broadcast, and web news media in over 100 languages from every corner of the globe, updated every 15 minutes.

We query the gdeltv2.gkg_partitioned table via Google BigQuery, filtering by company name across all media organization mentions, and extract the V2Tone emotional valence score for each article.

GDELTプロジェクトは世界最大のオープンアクセスニュースアーカイブです。100以上の言語で65,000以上のグローバルメディアを15分ごとに監視し、感情トーンスコアを自動算出します。

65,000+ Global Sources
Print, broadcast, web media across 100+ languages and 200+ countries
15-Minute Update Cycle
Near real-time global media monitoring — days ahead of equity market reaction
V2Tone Emotional Score
Proprietary GDELT sentiment metric: −10 (extreme negative) to +10 (extreme positive)
GDELT Theme Taxonomy
2,300+ machine-coded themes classify articles into regulatory, economic, political, and social risk categories

Claude AI — Institutional-Grade Interpretation

Raw sentiment scores mean nothing without context. We pass GDELT signals through Claude AI, prompted as a senior quantitative risk analyst with expertise in sentiment contagion and equity drawdowns.

Signal Extraction

Query GDELT for 24h rolling window vs. 7-day baseline. Calculate rage spike: current negativity ≥ 2× daily average.

spike = neg_24h ≥ 2 × avg_7d

Pattern Recognition

Claude identifies root causes from GDELT theme codes — regulatory, ESG, financial scandal, political — and maps to historical drawdown patterns.

Structured Output

JSON-schema-constrained output ensures consistent, machine-parseable reports: rage score, rating, summary, insight, confidence.

What Signals We Track

Rage Spikes
Sudden surge in highly-negative articles (tone < −5) compared to 7-day baseline
Tone Deterioration
Average sentiment dropping 2+ points below baseline, sustained below −3
Theme Concentration
Single risk theme dominating 60%+ of 24h coverage — regulatory, scandal, political
Volume Anomalies
Article volume 3× above daily average — media amplification event in progress
Cross-Market Spread
Same anger narrative appearing across unrelated geographic media markets simultaneously
Recovery Signals
Post-spike tone normalization — long re-entry timing for counter-trend positioning

A–E Downside Risk Scale

Each report delivers a single actionable rating — calibrated to institutional decision-making timelines.

A
No Meaningful Signal Rage Score: 0–20

Normal negative press cycle. No deviation from baseline. Standard monitoring posture appropriate.

通常のネガティブ報道サイクル。ベースラインからの乖離なし。標準的な監視体制が適切。

B
Mild Headwind Rage Score: 21–40

Elevated negativity above baseline. Monitor for escalation. No immediate portfolio action required.

ベースラインを上回るネガティブ感情。エスカレーションを監視。即時のポートフォリオ対応は不要。

C
Moderate Risk Rage Score: 41–60

Significant anger signal with 1–2 week price risk window. Defensive positioning and hedging analysis warranted.

1〜2週間の株価リスクウィンドウ。守備的ポジショニングとヘッジ分析を推奨。

D
High Crash Probability Rage Score: 61–80

Severe backlash. Material downside risk within 30 days. Short positioning or put protection strongly indicated.

深刻な反発。30日以内に重大な下落リスク。ショートポジションまたはプットオプションによる保護を強く推奨。

E
Extreme — Imminent Risk Rage Score: 81–100

Viral outrage / boycott-level event. Immediate material drawdown expected. Urgent risk committee escalation required.

バイラルな怒り・不買運動レベル。即時の重大な株価下落が予想。緊急のリスク委員会エスカレーションが必要。


Built for Every Role in the Investment Workflow

Equity Analyst

Screen your entire coverage universe for sentiment deterioration before earnings season. Identify which names carry elevated reputational risk.

  • Pre-earnings anger scan
  • Sector-wide risk comparison
  • Theme-based research triggers

Portfolio Manager

Add a media-anger dimension to your existing risk models. Get early warning on reputational tail risks before they impact your book.

  • Watchlist monitoring alerts
  • Portfolio-level rage score
  • Hedge timing optimization

Short Desk

Identify high-probability short setups where public anger is accelerating before price discovery. Find entries before consensus forms.

  • Rage spike short triggers
  • Entry timing with 7d trajectory
  • Cover timing on rage exhaustion

Common Questions

How is this different from social media sentiment tools? +
How far in advance does the signal appear before price moves? +
What companies can I analyze? +
ソーシャルメディア感情ツールとの違いは? +

Ready to Add the Anger Layer?

Join institutional investors using media anger signals to get ahead of downside risk.

機関投資家と同じ情報で、リスクを先読みしましょう。