The GeoWatcher WW3 Risk Tracker produces a daily probability estimate of whether the current Iran-US conflict (Operation Epic Fury) will escalate to World War 3 within a 36-month horizon. It does this by running a Monte Carlo simulation: 10,000 randomized iterations of possible futures, each shaped by the same set of calibrated input variables.
The output is not a prediction. It is a probability distribution. The percentage displayed on the site represents the share of those 10,000 simulated futures in which the conflict escalated to great-power war.
The model uses nine continuous variables, each set on a 0-100 scale. These variables are calibrated daily by the operator based on verified open-source intelligence.
| Variable | What It Measures |
|---|---|
| China Aggression | How aggressively China is pushing against US interests through military posturing, diplomatic pressure, and coordination with Iran and Russia. |
| Russia Instability | How unstable Russia's internal political and military situation is. Higher values mean more reckless, desperate actions. |
| Iran Defense Capability | How much of Iran's military capability survived US and Israeli airstrikes. Lower values mean Iran is weaker. |
| US Political Will | Domestic political and public support for continuing the conflict. One of the most consequential variables in the current model. |
| NATO Cohesion | How unified the NATO alliance is behind US-led operations. Lower values mean fractures are appearing. |
| Saudi Arabia Stability | Saudi economic and political stability. Matters because Hormuz and Gulf energy routes run through its sphere of influence. |
| Economic Stress | Combined economic strain on the US and global systems: energy prices, supply chain disruption, consumer sentiment. |
| Turkey Stability | Turkey's internal stability and its relationship with NATO and the conflict parties. Turkey controls the Bosphorus. |
| North Korea Opportunism | The probability that Pyongyang exploits the conflict through weapons tests, provocations, or coordination with Iran and Russia. |
In addition to the nine continuous variables, the model includes fourteen binary event toggles. Each toggle represents a specific event or condition that is either active or inactive. When a toggle is activated, it modifies how the model weights certain pathways in the simulation.
Examples include Black Swan Events, AI Weapons Deployed, Deepfake Incident, US President Incapacitated, and Satellite Denial (ASAT). Each toggle is set based on whether the corresponding event has occurred or is assessed as imminent.
The engine uses Monte Carlo simulation. In each iteration, the model takes the calibrated input state and applies randomized perturbations drawn from probability distributions informed by historical conflict patterns. Each iteration produces a single simulated outcome. After 10,000 iterations, the model aggregates the results into a probability distribution across six possible outcomes.
The WW3 Risk percentage is the share of iterations that produced the "Loss (WWIII)" outcome. The "Most Likely Scenario" is the outcome that appeared most frequently across all 10,000 iterations.
Each daily baseline follows a seven-step process:
Step 1: Research collection from open-access sources. Step 2: Variable calibration based on the day's confirmed intelligence. Step 3: Simulation run at 10,000 iterations. Step 4: Provenance recording (every input and output documented). Step 5: Publication to the site. Step 6: Research PDF publishing. Step 7: Cross-platform distribution (X, Telegram, Threads).
The provenance record ensures that every baseline is auditable. Every input that produced the output is documented and timestamped.
The model produces six possible outcomes for each iteration:
The confidence band (the plus/minus figure displayed below the WW3 Risk percentage) represents the inter-quartile range across the 10,000-iteration run. It measures how much the simulation's output varied from run to run.
A narrow band means the model is confident in its output regardless of random variation. A wide band means the inputs are positioned near a threshold where small changes in randomization produce meaningfully different results. The confidence band is recorded at simulation run time and published alongside the baseline percentage.
The model does not predict the future. It estimates probabilities based on current inputs. If the inputs change tomorrow, the output changes tomorrow.
It does not account for unknown unknowns. The Black Swan toggle exists to add weight to tail-risk outcomes, but the model cannot anticipate events that have no historical precedent.
The output reflects the inputs, not reality. If the operator miscalibrates a variable, the output will be wrong. This is why the provenance record exists: so that every calibration decision can be audited after the fact.
All research inputs use open-access sources exclusively. No paywalled journals. No classified intelligence. No anonymous tips. The accepted source list includes Reuters, AP, AFP, BBC, Al Jazeera, CNN, the New York Times, the Financial Times, SIPRI, ACLED, GDELT, EIA, and official government statements from the parties involved.
The full source citations for each day's baseline are published in the research PDF linked from the baseline history page.