RAAZ

Research preview / source notes

What the pressure gauges mean

Raaz estimates annual country-level risk from historical observations. It does not forecast a specific location, date, or incident. Data availability and reporting vary across countries and years.

Revolution pressure marker score

The three-country list, revolution map and global revolution dial use a separate 2025 indicator index, scored from 0 to 100. It is an editorial weighting of observable conditions, not a probability that a revolution will occur in 2026 or 2027. A revolution in the outcome dataset requires mass action aimed at replacing a government through noninstitutional means; a coup alone does not qualify. The index covers only countries with every input observed in 2025. Its fixed bands are Calm 0–24, Watch 25–49, Elevated 50–74 and Severe 75–100.

MarkerWeightInput and scaling
Pro-democracy public mobilization30%V-Dem v2cademmob_ord, 0–4 divided by 4.
Groups opposing the political regime25%V-Dem v2regoppgroupssize_ord, 0–4 divided by 4; excludes ordinary electoral opposition.
Economic distress25%60% real GDP per person contraction, capped at an 8% fall; 40% consumer inflation, capped at 40%.
Political polarization10%V-Dem v2cacamps_ord, 0–4 divided by 4.
Restrictions on peaceful assembly10%Inverse of V-Dem v2caassemb_ord, 0–4.

We also record V-Dem v2csreprss_ord, civil society repression, inverted from its 0–4 ordinal scale to a 0–1 regime-response proxy. The five-marker raw score is multiplied by (1 − 0.20 × that proxy), so observed repression can lower the score by at most 20%. This is an editorial scenario adjustment, not a trained coefficient. Past repression does not tell us whether security forces would obey orders during a future uprising. It can also be a response to existing unrest. The adjusted score is never presented as a probability.

The weights express a transparent theory of grievance, economic strain, mobilization and organized opposition. They have not been fitted to revolution outcomes or validated on later years. The annual revolution probability stays hidden because its chronological backtest did not beat the historical-frequency benchmark. V-Dem v16 and its indicator definitions provide the political markers; World Bank real GDP per person growth and consumer inflation provide the economic markers. Data attribution and reuse terms accompany the derived values.

Outcomes

GaugeEvent counted in training
Armed conflictAt least 1,000 battle-related deaths in a country during a calendar year (UCDP war intensity).
RevolutionA qualifying revolution episode beginning during a calendar year, per the HSE Revolution Dataset v1.1. Coups alone are excluded.
Economic downturnAt least a 3% year-over-year decrease in real GDP per person, calculated from World Bank WDI NY.GDP.PCAP.KD.

Model and evaluation

A regularized logistic regression learns from prior country-year observations. Inputs known at the end of each year include current and previous war status, five-year war history, real GDP per person, recent real growth, inflation, and missing-value flags. No 2025 revolution label is treated as a confirmed negative. The evaluation uses features from two years before each outcome: it trains on outcomes through 2017 and tests on later outcomes through 2025 (2024 for revolutions), then refits on all labeled years for the displayed 2027 calendar-year baseline. It uses no Raaz story text as a supervised outcome.

The benchmark always predicts the training period event frequency. A category is shown only if its later-period Brier score is at least 1% lower and the test set contains at least 25 events. The backtest is one chronological holdout, not evidence of live forecasting accuracy. Retraining and independent external evaluation are still needed.

CategoryModel BrierBaseline BrierTest eventsPublic score
War0.02580.042966Shown
Revolution0.02560.025734Hidden
Downturn0.10870.1086159Hidden

Displayed country probabilities are raw model outputs. The band compares each probability with the historical rate among covered country-years: Calm below 0.65 times that rate; Watch from 0.65 to under 1.35 times; Elevated from 1.35 to under 2.5 times; Severe at 2.5 times and above. The global war and downturn gauges are means of covered country estimates, not the probability that at least one event will happen anywhere. The global revolution gauge is instead the mean of covered countries’ adjusted 0–100 marker scores. Its fixed numeric scale and bands differ from the annual probability gauges.

Time and coverage limits

The current input snapshot ends in 2025, but its official release became available in September 2026. The 2027 number is a next-calendar-year baseline using information available by September 2026; it is not a live or rolling twelve-month forecast. One-, three-, and six-month scores remain unavailable because the annual labels cannot be divided into monthly probabilities. A scheduled source check runs daily, but V-Dem political observations remain annual; a new check does not create new daily indicators. New source years and model claims require fresh validation.

Raaz holds a large geolocated historical site corpus. A location and an article about an event do not by themselves create reliable country-month event labels. The corpus is useful for historical context and explanation, but the model does not currently use it as a predictive feature. Options Oracle news feeds are current snapshots rather than an archived, consistently geocoded time series; they are not yet used for training or scoring. Adding them requires a persistent archive, rights review, leakage checks, and a new chronological backtest.

Sources and reuse

Code and reproduction notes are in the Raaz repository. This is a research preview and not safety, travel, financial or policy advice.