Parameter Guide

Pick the countries you want to model as Blue. More members usually means more depth, logistics, and complexity.

• Start small (1–3) to understand the mechanics.
• Add allies when you want coalition effects.

Pick the opposing coalition. Large coalitions can increase compute time and widen uncertainty bands.

• If you want a clearer outcome, avoid overly symmetric matchups at first.

The theater is the “map anchor” for the run. Geography influences operational reach, supply realism, and domain emphasis.

• Naval-heavy theaters increase sea weighting.
• Landlocked theaters shift weight to land/air.

Changing the year changes baseline capability inputs. Use it to explore “near-future” vs “current” balance.

• Keep year fixed when comparing strategies.

Intensity is the master “tempo knob.” If outcomes feel too slow or too explosive, adjust intensity first.

• Higher intensity usually reduces “draw” probability (decisions happen faster).

A longer horizon lets the simulation “breathe.” A short horizon makes it harder to separate evenly matched sides.

• Use 60 for fast tests, 120 for more realism.

Iterations control statistical stability. If your probabilities jump around, increase iterations.

• Mobile: 200–400
• Desktop: 600–1200

Use a fixed seed for fair comparisons. Use random seed for exploratory browsing.

A 60/30/10 split is a strong edge. A 40/39/21 split is uncertain and will look more mixed in the phase separation.

When the edge is strong and draw risk is low, you’ll see cleaner “oil-water” separation. Otherwise it stays partially mixed.

Plies are the length of the duel story. Think of it as “how many chess beats occur during the simulation timeline.”

Use depth 2 as your default. Use depth 3 when you want the cleanest decisions and your device can handle it.

If the duel feels jumpy, slow it down. If you want fast demo runs, speed it up.

Chess follows the evolving uncertainty. When the MC edge is small, the duel can look tactical and volatile.

Cinematic emphasizes a coherent storyline and quick feedback. Losses are easier to interpret but less granular.

Dense introduces more knobs and internal multipliers. It can be more expressive, but needs careful parameter control.

Integrated is your “serious mode.” It’s slower, but it makes the geo + operational constraints matter more.

• If results feel too deterministic, increase MC iterations instead of changing many knobs.

You’ll be able to bring your own country metrics and run the same engine against your custom assumptions.