Game Budget
Compass

A budget reality check, before you spend a cent.

Plug in your project and we'll show you the full distribution of what comparable games cost.

v0.1.0-alpha UPDATED 2026·05·12

STATUS ALPHA

This tool is in alpha. Predictions are indicative & accuracy will improve as data and features grow.

VERSION v0.1.0-alpha
LAST UPDATE 2026·05·12
PREDICTED BUDGET MODEL v0.1.0-alpha
q50 · MEDIAN
$80.6K
$80,615
q30 · LOW
$44.5K
30% cost less
q70 · HIGH
$158K
70% cost less
DISTRIBUTION · USD

What comparable games actually cost

ATTRIBUTION

What's driving the number

Each feature's multiplicative effect on the median estimate, sorted by magnitude. Bars are clipped at ±2× for legibility.

  • Core team size 4 ×2.92
  • Game length 10h ×1.95
  • Self-published yes ÷1.30
  • Release year 2027 ÷1.27
  • Multiplayer yes ×1.19
  • Subtitle languages 8 ×1.04
  • Managers / leads 0 ×1.00
  • 2D/3D 3D ×1.00
  • Engine Unity ×1.00
  • Audio languages 0 ×1.00
  • Co-development no ×1.00
REALITY CHECK

Your envisioned budget

Type your number. Supports $50M, 240k, 1.5B.

$
AWAITING INPUT Enter a number to see how it compares.

SUBSCRIBERS ONLY

Sign in to customize and run your own scenario.

Sign in →
NEAREST NEIGHBOURS

Comparable titles

ALPHA · LOCKED

Coming after alpha

Nearest-neighbour comparable titles will unlock when the dataset moves out of alpha.

Q&A

Frequently asked

How does this tool work?
A quantile regression model trained on a curated panel of games of all scopes returns a full distribution (q10 through q90) given your inputs. Coefficients are fit independently at each quantile, then aggregated to a predicted distribution.
What data is the model trained on?
200+ published games released between 2014 and 2025. For each, we collected scope and team features (tags, 2D/3D, solo/multi, publishing model, team size, length, engine, localization, etc) and their production budget. Budget figures come from press reports, post-mortems, court filings, and private disclosures we can't name.
When are predictions less reliable?
Here are few known limits of our predictions: (i) Niche engines (Godot, GameMaker) have small samples. (ii) Large AAA productions are statistical outliers by nature, and outliers are hard to predict. (iii) IP licensing costs are invisible to the model. (iv) Country specific labor costs are not taken into account here.
Why a distribution instead of a single number?
Two studios making similar games rarely spend the same amount. Salary bands, location, process maturity, crunch tolerance, outsourcing strategy; all of those are invisible to us (for now, as we are working on the location and salary topics), and any of them can swing the budget significantly. A single-number estimate would feel reassuring but lie about its precision. The distribution surfaces that real-world variance more honestly.