I ran a simulation for the possibility of conflict based on available models and public open source intelligence:
--- This is V3 of the simulation. ---
This version adds a probabilistic simulation of armed conflict likelihood between India and Pakistan over the next 12 months (from 8 October 2025). It synthesizes open-source reporting on the May 2025 crisis, nuclear posture dynamics, and current political signals to produce scenario probabilities and sensitivity analysis.
Simulation: Likelihood of Armed Conflict (next 12 months)
Methodology:
A qualitative scoring model was used to map key drivers to scenario probabilities. Drivers considered: recent crisis behaviour (May 2025), cross-border incidents and ceasefire violations, militant attacks, domestic politics in both countries, military capabilities and postures (including nuclear weapons), and international diplomatic mediation capacity.
Weights were assigned to each driver and combined into a risk score which was mapped to scenario probabilities. This is a planning simulation, not a predictive forecast.
View attachment 152519
View attachment 152520
Next I asked the simulator to run to add a sensitivity sweep to the above scenario
View attachment 152521
The limitation for the above is that all this relies on an open-source, qualitative-probabilistic simulation.
Model Used: Each driver (historical crisis frequency, militant threat, force posture, political signalling, international mediation) was scored 0-10 and weighted.
Sources Used:
Sources: Stimson Center 'Four Days in May' (May 2025); Federation of American Scientists/Bulletin Nuclear Notebook (Sep 2025); IFPRI/IFRI analyses on Pakistan's nuclear posture (2025); recent news reporting (Oct 8, 2025).
I then used the above model and asked the simulator to run 10,000 iterations of the above to predict the probability of armed conflict between India and Pakistan:
View attachment 152522
The baseline run shows high likelihood of low-level incidents (~56%) and moderate probability of limited conventional exchange (~23%). Sustained conventional war (10%) and nuclear use (1%) remain lower, but non-zero. Only about 10% of iterations resulted in no significant escalation.
Using the above Simulation - a sensitivity sweep was performed to include the regional marginals vs the probability of using Scenario A.
View attachment 152523
The Conditional region distributions (illustrative assumptions used to allocate events to regions):
- A (Low-level incidents): Kashmir 70%, Punjab 25%, Maritime 5%
- B (Limited conventional exchange): Kashmir 50%, Punjab 30%, Maritime 20%
- C (Sustained conventional war): Kashmir 40%, Punjab 40%, Maritime 20%
- D (Nuclear use): Kashmir 60%, Punjab 30%, Maritime 10%
- E (No significant escalation): None (no region)
View attachment 152524
Out of 50,000 runs the results provide a unique insight:
- The marginal probability that Kashmir is the locus of an incident (any scenario other than E) is the highest of the three regions under these assumptions (the exact % is in the table).
- Punjab is the second-most-likely region for conventional escalations, and Maritime remains the least-likely but non-negligible in limited or sustained conventional exchanges.
====
TLDR: Based on Conflict modelling - the likelihood of a low scale conflict between India and Pakistan is estimated to be around 55% with Kashmir most probable to be the flashpoint.
The likelihood of a protracted full scale war remained at 1%
====
Sources:
Stimson Center 'Four Days in May' (May 2025); Federation of American Scientists/Bulletin Nuclear Notebook (Sep 2025); IFPRI/IFRI analyses on Pakistan's nuclear posture (2025); recent news reporting (Oct 8, 2025).
