How Wall Street Is Using AI to Price War Risk After the Iran Conflict

TLDR

• The number of countries in external conflicts has nearly doubled since 2008, with the economic cost of violence reaching almost $22 trillion.
• Verisk’s new Predictive War Index uses machine learning to forecast the likelihood of war in a country over the next 12 months.
• Back-testing showed the model would have given a 66% probability of war breaking out in Iran just weeks before it happened.
• Citigroup and Morgan Stanley say traditional risk models built on historical data are no longer fit for purpose.
• War has overtaken civil unrest as the top political violence fear for businesses seeking insurance coverage, according to Allianz.

Wall Street and the insurance industry are turning to a new generation of AI-powered models to better predict wars and military conflicts — tools borrowed from the world of natural disaster modeling.

The shift comes as geopolitical risk has grown dramatically. Since 2008, the number of countries in active external conflicts has nearly doubled to over 100. The Institute for Economics and Peace puts the economic cost of violence at almost $22 trillion — more than 10% of global GDP.

Traditional financial models, many built on decades of historical data, are struggling to keep up. Citigroup warns against relying on “rear-view mirror” models. Morgan Stanley says it’s time to “rethink” how geopolitical risk is handled.

New Tools Built for a New Reality

Verisk, best known for modeling natural catastrophes for insurers and cat-bond investors, has released two new products aimed at war risk. Its Predictive War Index uses a machine learning algorithm trained on political, economic, and social data from 1995 to 2022. It forecasts the likelihood of war breaking out in a given country within the next 12 months.

Back-testing the model showed it would have flagged Iran with a 66% probability of war just six weeks before conflict broke out on February 28 of this year. Verisk also has a Geopolitical Relations Index that tracks tension levels between pairs of countries, looking at factors like past military clashes and geographic proximity.

A separate Verisk model, launched in October 2023, has correctly predicted six out of seven government collapses since then. These include the removal of Bashar al-Assad in Syria and Nicolas Maduro in Venezuela.

The RAND Corporation has also built an AI model that converts uncertain geopolitical scenarios into probability estimates. When run in mid-May, it put the likelihood of Iran’s regime failing to survive into 2027 at 20%.

Why Old Models Are Failing

A key problem is that events like sanctions or trade blockades don’t fit traditional financial risk frameworks. At Citigroup, a senior model risk executive explained that such events don’t behave like standard statistical moves — they change the entire distribution of possible outcomes.

The Strait of Hormuz conflict brought this into sharp focus. Shortly after war started, Lloyd’s of London was quoting marine war risk premiums as high as 1% of a vessel’s value per voyage, up sharply from a fraction of a percent before the conflict.

The US and Iran said late Sunday they reached an interim deal to reopen the Strait of Hormuz. Officials from both sides are set to meet in Switzerland on June 19 to formally sign the agreement, though key details remain unresolved.

Moody’s risk specialist Gordon Woo says conflict now mirrors terrorism modeling, where low-cost acts can cause outsized economic losses.

War has now overtaken civil unrest as the top political violence concern for businesses seeking insurance, according to Allianz’s May 2026 risk report.

The post How Wall Street Is Using AI to Price War Risk After the Iran Conflict appeared first on CoinCentral.