War Attention Index
The War Attention Index
A measure of how much New York Times coverage is occupied by war, built from 170 years of articles. It spikes in every major conflict since the Civil War.
Corrected monthly series: January 1857–March 2024 · 2,005 observations.
The chart summarizes the corrected monthly series. Use the button to download the annual chart data.
Method
What it measures.
For each 30-day window, the index is the share of New York Times articles of at least 100 words that mention both a conflict word and a military word, matched as full words including plural forms.
Why the series ends in March 2024.
The displayed series uses the corrected construction. An earlier substring-matching version incorrectly counted words such as “warehouse” and “award.” New periods will be added after validation.
Drawdown-aware allocation strategy
The research strategy converts monthly war attention into an equity weight using a 360-month rolling mean and standard deviation. The z-score is clipped to the interval [−1, 1], mapped to a 0%–100% equity weight, and held during the following month so the rule uses only information available at the time.
Performance results are under validation. Earlier Sharpe-ratio and maximum-drawdown figures were produced from the superseded index construction. They are intentionally withheld until the backtest is rerun against the corrected inputs. This is research, not investment advice.
Related research
- Hirshleifer, Mai & Pukthuanthong (2025), War discourse and disaster premium: 160 years of evidence from the stock market, Review of Financial Studies.
- Hirshleifer, Mai & Pukthuanthong (2025), War discourse and the cross section of expected stock returns, Journal of Finance.
- Buy on the Sound of Cannons: Tactical Allocation with 150 Years of War News, working draft, June 2026.
How would WAR change your portfolio?
Choose a stock–bond allocation, upload monthly returns, and compare the same portfolio with and without the WAR allocation rule.
Required columns: date,stocks,bonds. Dates use YYYYMM; returns are decimals. An optional rf column supplies the monthly risk-free return.
Historical comparison
| Measure | Your allocation | With WAR | Difference |
|---|---|---|---|
| Sharpe ratio | -- | -- | -- |
| Information ratio vs. your allocation | -- | -- | -- |
| Maximum drawdown | -- | -- | -- |
| Annualized return | -- | -- | -- |
| Annualized volatility | -- | -- | -- |
How it works. WAR is standardized against its trailing 360-month history, clipped at ±1, and applied one month later. The selected response tilts stock weight around your chosen allocation. Results use only overlapping months, assume a zero risk-free rate unless rf is supplied, and deduct 20 basis points per unit of portfolio turnover. This is a historical research comparison, not investment advice.