How The World Now works: methodology and data sources

The World Now operates a continuous intelligence pipeline that monitors hundreds of data sources in real time. Every event that reaches the platform has been ingested, parsed, classified, and scored before it appears on any tracker page or inside the Catalyst analysis feed. The pipeline runs continuously — not in daily batches — so that when a significant event occurs, it surfaces within minutes rather than hours or days.

Event classification assigns four key attributes to every record: type (what category of event is this), severity (how intense is it on a four-tier scale), location (country, region, and geographic coordinates), and timestamp (when the event occurred, not when it was reported). Separating occurrence time from reporting time is critical — it prevents recency bias in risk calculations and allows accurate temporal weighting.

The platform uses an all-hazards approach that spans five broad domains: geophysical events (earthquakes, volcanic eruptions, tsunamis), meteorological and hydrological events (hurricanes, floods, wildfires, droughts), conflict and political violence (wars, insurgencies, terrorism, civil unrest), infrastructure and industrial incidents (power grid failures, pipeline explosions, cyberattacks), and macroeconomic shocks (sanctions, trade policy shifts, sovereign debt crises). This comprehensive scope is deliberate: single-domain trackers like dedicated earthquake monitors or conflict databases capture only a slice of the risk landscape.

The all-hazards approach is justified by the interaction effects between domains. A drought in a politically unstable country can trigger food shortages, which trigger civil unrest, which triggers capital flight and currency collapse — a cascade that a single-domain tracker would miss entirely. The Global Risk Index and the live world map are designed to surface these multi-domain interactions as they develop, providing a more complete picture of global risk than any single specialized data feed.

Every event on The World Now is assigned one of four severity tiers. The tiers are not arbitrary — they correspond to multiplicative weights that determine how much each event contributes to risk scores and country rankings.

LOW (weight 1) covers localized incidents with limited direct casualties or economic impact. Examples: a minor magnitude-4.0 earthquake in a low-population area, a small-scale protest that disperses without violence, a contained industrial accident. These events are tracked but contribute minimally to composite risk scores.

MEDIUM (weight 2) covers significant but bounded events. Examples: a magnitude-5.5 earthquake near a populated area, a violent protest involving hundreds of participants, a sanctions package targeting a mid-sized economy, a regional infrastructure failure. These events double the risk contribution of LOW events.

HIGH (weight 3) covers major events with substantial casualties, economic disruption, or escalation potential. Examples: a magnitude-6.5+ earthquake, an active armed conflict with daily casualties, a major cyberattack on critical infrastructure, a central bank emergency intervention. HIGH events contribute three times the weight of LOW events.

CRITICAL (weight 4) is reserved for events with the potential for catastrophic, systemic, or irreversible consequences. Examples: a nuclear incident, a large-scale invasion, a financial system crisis affecting multiple countries, a category-5 hurricane making landfall on a major city. CRITICAL events have four times the base weight of LOW events.

Raw weights are adjusted by a temporal decay factor — a multiplier that decreases smoothly as events age, ensuring that recent events are weighted more heavily than old ones without creating artificial cliffs. The decay rate differs by product: the country risk rankings use a 7-day rolling window, so an event from six days ago contributes significantly less than one from six hours ago. The Global Risk Index uses a 15-minute update cycle with faster decay, giving it near-real-time responsiveness to breaking developments.

The World Now is designed to complement, not replace, existing geopolitical and risk indices. Understanding the differences helps users choose the right tool for each analytical task.

The Global Peace Index (GPI), published annually by the Institute for Economics and Peace, uses 23 indicators covering societal safety, ongoing conflict, and militarization. It provides deep structural analysis of why some countries are peaceful and others are not, incorporating factors like political instability, crime rates, and military expenditure. Its strength is longitudinal consistency — the same indicators measured the same way going back to 2008. Its limitation is obvious: annual publication means the GPI cannot capture the outbreak of a new conflict or a sudden political crisis. For long-run structural analysis, the GPI is excellent. For tracking an active crisis, it is irrelevant until next year's edition.

The INFORM Risk Index, maintained by the European Commission, models humanitarian crisis risk across three pillars: hazard and exposure (what threats exist), vulnerability (how susceptible the population is), and lack of coping capacity (how well the government and society can respond). INFORM is excellent for humanitarian planning — it identifies where a crisis is most likely to overwhelm local response capacity. But like the GPI, it is an annual product that reflects structural conditions rather than current events.

The Caldara-Iacoviello GPR Index, developed at the Federal Reserve, uses natural language processing to count newspaper articles referencing geopolitical tensions and threats. It has been constructed back to 1900 and is available at monthly frequency. Its strength is historical depth — it allows comparisons across more than a century of geopolitical history. Its limitation is that it measures media attention rather than event intensity. A conflict that receives less press coverage will score lower even if its human and economic toll is equivalent to a heavily covered one. The index can also lag when media cycles shift attention away from ongoing conflicts.

The World Now takes an event-driven, real-time approach. Rather than relying on media coverage, insurance pricing, or annual surveys, it processes actual events — their type, severity, location, and timing — as they occur. This produces a more responsive and granular risk signal, particularly for acute threats. The trade-off is coverage of structural factors: The World Now captures what is happening right now better than why a country is structurally dangerous. For a complete picture, analysts should use these approaches together. The most dangerous countries rankings and safest countries pages combine real-time event data with structural context for exactly this reason.

Catalyst is The World Now's market intelligence layer, translating global events into directional forecasts for 28 tracked assets. The pipeline runs in three phases: event processing, AI reasoning, and prediction output.

In the event processing phase, every classified event is evaluated against a mapping of 28 assets — equities, indices, commodities, currencies, and cryptocurrencies — to determine which assets have meaningful exposure to that event type and geography. An earthquake in a lithium-producing region is flagged for EV-related equities and battery metals. A central bank surprise is flagged for currencies, government bonds, and rate-sensitive equities. This asset-mapping step ensures that the AI reasoning phase focuses on genuinely relevant connections rather than generating spurious correlations.

In the AI reasoning phase, a specialized prompt architecture instructs the model to analyze each event as a senior macro strategist would. The analysis covers three layers: direct effects (what is the immediate, mechanical impact on supply, demand, or sentiment for the affected asset), second-order effects (how do the direct effects propagate through related markets and economic relationships), and positioning effects (how are current market participants likely to react, and how does that reaction affect price dynamics).

The prediction output includes a direction (bullish, bearish, or neutral), a confidence level, and a magnitude estimate across three timeframes: short-term (hours to days), medium-term (days to weeks), and long-term (weeks to months). Every prediction links directly to the source events that generated it, making the reasoning transparent and auditable. Users can follow the chain from a specific geopolitical development to a specific asset forecast and understand exactly why the system reached its conclusion.

The 28 tracked assets span gold, oil, natural gas, major equity indices, key currencies, and cryptocurrencies. Class-specific details are available on individual asset pages — for example, the tracked assets hub provides entry points to all 28. Broader market context is on the stock market prediction and crypto price prediction pages.