Statistical Simulations (Multiple Runs)

Execute many runs and analyze the aggregated distribution

Statistical Simulations (Multiple Runs)

A Statistical run executes your simulation many times and returns aggregated statistics instead of per-step logs. Use it once your graph is stable and you want to understand the distribution of outcomes — averages, variance, extremes.

Setting up the run

Open the Statistical tab on the simulation page. The Runner shows:

Number of Runs — how many times to execute the simulation (e.g. 100). Bigger N = smoother curves, longer wait.
Game Version — same as Evaluation.
Steps Limit — same as Evaluation.
Profiles — pick a profile (or stay on runtime values). To compare several profiles, see Comparing Profiles.

There is no seed input — each run uses its own random seed so the distribution is meaningful.

What comes back

For each step position (step 0, step 1, …), the backend computes aggregated statistics across the N runs and returns them in a step statistics array. The graph then plots whichever aggregation you pick.

Available aggregations:

Mean — average value
Median — 50th percentile
Standard deviation — spread
Min / Max — extremes
(and more, depending on the datapoint type)

Pick the aggregation from the dropdown above the chart.

Step range

A step range slider above the chart lets you zoom to a sub-window. This is useful for long simulations where the early steps compress everything else.

Important: no step-by-step log

Statistical runs do not produce an executable log — you can't step through individual decisions, and the Step Insights panel is limited to aggregated values at each step. If you need to inspect a specific outcome, switch to Evaluation and pin a seed.

Tip: Start with a small N (10-20) to confirm your graph is well-shaped, then crank it up to 100-1000 for the real analysis.

Continue to Graph Analysis & Views to learn how to read and save chart layouts.

Statistical Simulations (Multiple Runs)