Azerbaijan vs Turkey, Armenia: Defender holds after 310 days
In today's run Azerbaijan attack Turkey, Armenia; the model ends in defender holds after 310 days.
What happened today
Today's defence and geopolitics headlines, in the outlets' own words (linked):
Those stories name the sides in today's run.
The scenario we ran
Azerbaijan attack Turkey, Armenia in autumn, mobilization full / partial, the United States stays out, nuclear weapons disabled, 365-day limit. Everything else is the engine's realistic default.
How the model played it out
Result: Defender holds after 310 simulated days (prolonged stalemate). Aggregate personnel losses within the model: 3,785 for Azerbaijan and 13,811 for Turkey + Armenia.
The war moved through initial strike from day 1, then rapid mobilization from day 3, then attrition phase from day 8, then stalemate from day 21, then contested front from day 23, then attrition phase from day 24.
Turning points:
- Day 12: Turning point: a shift in momentum. This moment mattered because momentum shifted durably toward Turkey + Armenia: a combination of local factors.
What decided it
- Deployed combat power (favoured Side B): Average deployed combat power: 16 vs 121 model units.
- Air situation (favoured Side B): Average share of air control for Azerbaijan: 23%.
- Logistics (favoured Side B): Average front-line supply efficiency: 12% vs 57%.
- Mobilisation (favoured Side B): Reservists mobilised: 223,074 vs 483,816.
- Economic resilience (favoured Side A): Average economic stability index: 0.94 vs 0.78.
The same countries, one on one
WARSCAPE keeps a permanent page for every land border in Europe, each holding its own simulated war:
- Azerbaijan vs Armenia: the model ended that one in defender holds after 365 days.
Try it yourself
Open the exact run and scrub the timeline, ask why at any event, or branch an alternate timeline from a turning point. Or change one assumption in the builder and run it again.
Sources and limits
This is a model result on public, aggregate, country-level data under the stated assumptions. It is not a prediction, and no operational information is used.