Every in-game economy, currency earned, prices, upgrade costs, has failure modes that are genuinely hard to spot by hand: a currency sink that's too weak and inflates to meaningless numbers by hour ten, an upgrade path with a hidden dead end nobody notices in a quick playtest. Simulation is exactly the kind of problem AI-assisted tooling is good at, because it's checkable.
What simulation actually catches
Run thousands of simulated play sessions against your economy's rules, varying playstyle assumptions (aggressive spender, hoarder, average), and you'll surface the failure modes that only show up at scale: a currency that trends toward zero or infinity, a price curve that makes one purchase strictly dominant over all others, a grind wall that's mathematically much longer than you intended. These are exactly the bugs that slip past a five-hour human playtest because a human doesn't naturally play a thousand hours before your economy ships.
The setup work you can't skip
The simulation is only as good as the rules you feed it. That means actually modeling your economy's real formulas, not a rough approximation, before you trust any output. Garbage rules in, garbage confidence out, and treating a simulation's output as gospel when the underlying model was sloppy is worse than not simulating at all, because now you're wrong and confident.
What it flatly can't tell you
Whether grinding for gold feels satisfying or feels like a chore. Whether an upgrade purchase feels like a meaningful moment or a checkbox. Simulation gives you the shape of the numbers. It has no access to the feeling of playing your game, and that feeling is most of what makes an economy good or bad in a player's actual experience, not just its math.
Where this fits in a real balancing pass
Use simulation early, to catch structural breaks in the numbers before you spend art and UI time on a system that's mathematically broken. Then hand the balanced-on-paper economy to human playtesters to catch the feel problems simulation is blind to. Numbers first, feelings second, in that order, because fixing feel on top of broken math wastes the feel-fixing effort.