How the War Simulator Actually Calculates Victory Probability
2026-09-19 · by Jordi Fuster
The introduction to the War Simulator covers what the tool does. This is the part that covers the actual math behind the number it shows you — which changed recently, and is worth explaining properly now that it's settled.
The old formula
For a while, every matchup on the War Simulator and the Conflict Simulator (the "Solo" and "With allies" scenarios on relation pages) came down to a blend of two things: each side's influence score, and each side's military expenditure — split 65% influence, 35% military spending. It worked, but it left out data the site had already collected elsewhere: GDP, population, and active-plus-reserve military personnel.
The new formula: five factors, weighted and compressed
The current version pulls in all five of those factors, each weighted by how much it plausibly contributes to which side would actually prevail in a prolonged confrontation:
Influence (30%) — the same aggregate score that drives the World rankings, capturing diplomatic weight and alliance coverage. Military expenditure (25%) — direct spending capacity. Military personnel (15%) — active and reserve forces combined. GDP (20%) — the economic base needed to sustain a conflict. Population (10%) — the deepest, slowest-moving factor of the five.
Here's the part that actually matters for how believable the results feel: each factor isn't split between the two sides using its raw value. It's split using the square root of each side's value instead. Without that adjustment, a wildly unequal matchup — the United States against a small Pacific island state, say — would blow out to something close to a 99.9%-to-0.1% split, which reads as more of a rounding error than a simulation. Taking the square root compresses the gap: the stronger side still wins the percentage decisively, but the result stays inside a range that feels like an actual estimate rather than a foregone conclusion.
One more rule: a factor only gets dropped from the calculation, with its weight redistributed across the rest, if neither side has any data for it at all. If just one side is missing a number, that side simply counts as zero for that factor — it doesn't get to skip the comparison.
Tested against five real matchups
Before shipping the new formula, it was checked against five real pairings already in WorldRelat's own data, comparing four candidate methods side by side. The version that shipped — weighted factors split by square root — produced: United States vs. Tuvalu at 97.3% to 2.7%, Spain vs. France at 44.3% to 55.7%, India vs. Pakistan at 68.6% to 31.4%, a three-country team of Poland, Ukraine and Romania against Russia at 46.1% to 53.9%, and Qatar vs. Ethiopia at 52.3% to 47.7%. Note that a "with allies" matchup adds a supporting country's full set of factors — military spending, personnel, GDP and population, not just influence — the same way the War Simulator handles a full team.
What it still doesn't model
No formula built from national aggregates can capture terrain, alliances that activate mid-conflict, doctrine, morale, or the sheer unpredictability of an actual war — and a nuclear-armed pairing like India and Pakistan gets a separate, more serious warning about mutually assured destruction rather than a clean percentage either side should read as reassuring. The number is a structured estimate built from real economic and military data, not a prediction, and the site treats it that way throughout — see the War Simulator itself to try your own matchup.