Risk Model
WW3 Tracker
Transparent Bayesian analysis of major-power conflict escalation paths
This is an illustrative Bayesian model for educational purposes only. It is not a forecast and should not be taken as a prediction of real-world events. The model is transparent and uses publicly available data, but the choice of priors, evidence factors, and likelihood ratios involves subjective judgment. Real geopolitical risk depends on countless unmeasured factors, including decision-making by world leaders, accidents, miscommunication, and unknown unknowns.
99.6%
Illustrative Annual Risk
Live Dashboard
Key metrics tracking major-power military capacity and active conflict zones (October 2026).
Military Spending
USD Billions (SIPRI 2024)
Nuclear Warheads
Total Deployed & Reserve (FAS 2025)
Active Major Conflicts
UCDP 2024
Bayesian Model
This model combines a baseline prior (historical frequency of major-power conflict) with six evidence factors. Each factor has a likelihood ratio (how much it increases or decreases conflict probability) and a confidence weight.
P(Conflict | Evidence) ∝ P(Evidence | Conflict) × P(Conflict) / P(Evidence)
Prior Beliefs
- Baseline conflict probability
- 2.0% annual risk (historical average 1945–2024)
- Escalation likelihood
- If a major incident occurs, 15.0% chance of escalation to full war
Evidence Factors
NATO eastward expansion (1999–2024)
Likelihood Ratio (Conflict vs Peace): 1.80 : 0.90
Confidence: 85%
Source: NATO expansion timeline, academic IR literature
Data Date: 1/1/2024
Ukraine conflict (2022–2026)
Likelihood Ratio (Conflict vs Peace): 3.20 : 0.50
Confidence: 95%
Source: UN situation reports, casualty counts, SIPRI
Data Date: 10/1/2026
Nuclear-armed adversaries (USA+NATO vs Russia)
Likelihood Ratio (Conflict vs Peace): 2.50 : 0.40
Confidence: 90%
Source: Federation of American Scientists, SIPRI nuclear database
Data Date: 1/1/2026
Economic interdependence (USA–China trade)
Likelihood Ratio (Conflict vs Peace): 0.60 : 2.10
Confidence: 80%
Source: US Census Bureau trade data, IMF Direction of Trade
Data Date: 9/1/2026
Taiwan strait tensions (2020–2026)
Likelihood Ratio (Conflict vs Peace): 2.00 : 0.70
Confidence: 80%
Source: Taiwan Ministry of Defence reports, PLA statements
Data Date: 10/1/2026
Strategic stability & arms control (JCPOA collapse, New START uncertain)
Likelihood Ratio (Conflict vs Peace): 2.20 : 0.60
Confidence: 75%
Source: UN Security Council records, think tank assessments
Data Date: 6/1/2024
Data Sources
All dashboard metrics are sourced from peer-reviewed databases and open international reporting:
- Military Spending: USD 2.7 trillion
- Stockholm International Peace Research Institute (SIPRI) Military Expenditure Database, 2024. Global total across all nations. SIPRI.org
- Nuclear Warheads: ~12,000
- Federation of American Scientists (FAS) Nuclear Notebook, 2025. Deployed and reserve warheads.Nuclear Notebook
- Active Major Conflicts: 4
- Uppsala Conflict Data Program (UCDP) Armed Conflict Dataset, 2024. Defined as conflicts with 25+ deaths per year.UCDP
Key Limitations
- Subjective priors: The baseline conflict probability (2%) is debatable; different historians may give 1–5%.
- Incomplete evidence: Many risk factors are not captured here (economic depression, climate stress, internal instability in major powers, etc.).
- Unmeasured feedback: Real escalation paths are nonlinear and depend on human psychology, organizational behaviour, and fog of war.
- Model recency: This snapshot reflects October 2026 data. Conditions change rapidly, and older data may not predict future events.
- No crystal ball: Even a perfect Bayesian model cannot predict black-swan events (accidents, rogue actors, charismatic leaders) that reshape history.
Definitions
- Bayesian inference
- A method of updating probabilities in light of new evidence. Formula: P(A|B) = P(B|A) × P(A) / P(B). Here, A = conflict, B = evidence factors.
- Prior
- The initial probability before seeing evidence (e.g., historical baseline of 2% conflict per year).
- Likelihood ratio
- How much an evidence factor makes conflict more or less probable. A ratio > 1 increases conflict risk; < 1 decreases it.
- Posterior
- The updated probability after combining all evidence and likelihood ratios. This is the model's final estimate.
- WW3 / major-power conflict
- A war between two or more members of NATO and CSTO/China (e.g., NATO-Russia direct war, or USA-China conflict over Taiwan). Note: "WW3" is a shorthand; a real war might involve proxy conflicts, limited exchanges, or regional escalation.
Why This Model?
This tracker exists to clarify thinking about risk, not to alarm. By laying out priors and evidence transparently, it allows readers to:
- Disagree with specific factors and adjust their own estimates.
- Understand which pieces of evidence matter most (e.g., nuclear arsenals vs. trade ties).
- Recognize deep uncertainty: a 5% risk sounds small until you realize 5 in 100 years is odds worth taking seriously.
- Track changes over time as new evidence arrives (e.g., a ceasefire, arms agreement).
For related analysis, see the Countries & Alliances page. Data sources cited above; last updated October 2026.