VICO offers an AI‑powered platform that delivers probabilistic forecasts, scenario simulations, and early‑warning risk indicators for political, geopolitical and economic events. By ingesting real‑time news from over 150,000 sources and applying Bayesian inference and ensemble modeling, it provides calibrated event probabilities and customizable watch centers accessible via a web dashboard, API, Bloomberg Terminal, and alert channels. The service helps analysts, traders, and risk professionals quantify event risk and make more informed investment and policy decisions.
Funding
Funding not disclosed
Founders
Product
Problem
Analysts, traders, and risk professionals often rely on fragmented news sources and human judgment to assess the likelihood of political, geopolitical, and economic events, leading to delayed or inaccurate risk estimates and missed market opportunities.
Solution
VICO provides an AI‑powered platform that generates probabilistic forecasts, scenario simulations, and risk indicators for macro, policy, and geopolitical events. The system ingests real‑time news from over 150,000 sources, transforms unstructured text into a structured event taxonomy, and applies multi‑agent LLM workflows, Bayesian inference, and ensemble modeling to produce calibrated probability estimates. Users can query “what‑if” scenarios, run if/then simulations, and receive early‑warning indicators for specific events. Forecasts are updated daily and delivered via a web dashboard, API, Bloomberg Terminal integration, and customizable alert channels (email or SMS). The platform’s back‑tested Brier Scores (0.12‑0.15) demonstrate accuracy comparable to or better than prediction markets and top human forecasters.
Target Audience
Primary customers are institutional investors, hedge funds, corporate risk teams, and government analysts who need quantitative event risk assessments to inform trading, investment, and policy decisions.
Features
- Real‑time extraction of news from 150,000+ providers using a proprietary LLM‑driven data pipeline
- Structured event taxonomy with atomic components (actors, actions, impact, geolocation) and over 1 million proprietary data transformations
- Probabilistic forecasts expressed as time‑bound event probabilities (e.g., ceasefire likelihood by 2026)
- “What‑if” scenario analysis and if/then simulations for policy and market impact assessment
- Indicator generation that highlights leading signals preceding a forecasted event
- Daily forecast updates with ensemble AI models, Bayesian reasoning, and validity checks to reduce hallucinations
- Access via web UI, API, Bloomberg Terminal, and automated email/SMS alerts; customizable watch centers for specific sectors or actors