Monte Carlo valuation

A point estimate hides how much you do not know. This skill takes a finished DCF, replaces two or three of its drivers with probability distributions, and runs the same model thousands of times. What comes back is a distribution of value.

That distribution answers questions a single number cannot. How wide is the plausible range? Where does today's price sit inside it? How often does this company end up worth nothing?

The model is not reimplemented here. Every trial calls the engine in dcf-valuation-engine, so the simulation and the base case are the same code with different inputs.

The script

resources/simulate.py — pure standard library, no installation needed. JSON on stdin (or --in FILE), JSON out.