Crowdsourced Data

Every country and city is scored 0–100 by an equal-weight composite of three crowdsourced pillars. Higher is better. The weights are fixed and identical for every row.

Composite = 100 × mean( norm(savings) , norm(lifestyle) , norm(people_supported) )

norm(x) = (x − min) / (max − min)     # min & max taken across all ranked
                                      # countries (or cities), per pillar

The three pillars (⅓ weight each)

• Annual savings — euros kept per year after tax and cost of living (absolute €).
• Lifestyle level — self-reported quality-of-life, scale 1 (frugal) to 3 (luxurious).
• Avg. people supported — how many people the income supports (household-size signal).

• Each pillar is min-max normalized across the ranked set, so a score is relative to the other places shown — not an absolute index.
• The score always uses absolute € savings. The “Savings as % of income” toggle is a display view only and never changes a ranking.
• The Happiness & Infrastructure ranking uses the same method with two equal-weight pillars (happiness, tax & infrastructure satisfaction).
• Rows backed by fewer than 25 submissions are flagged low n and can be hidden with the “high-confidence only” toggle.

The Overall (Livability) rank — the default view

Blends the two rankings above into one 0–100 score, weighted 60% financial outcomes / 40% livability (happiness + tax & infrastructure). Financial leads because it rests on far more submissions. Three safeguards keep the blend honest:

• Common scale — both pillars are min-max normalized over the same gated set of places, so 0–1 means the same thing on each axis (we don't average the per-tab composites, which are each scaled over a different population).
• Bayesian shrinkage — a place's happiness/tax score is pulled toward the population average in proportion to how little data backs it (adj = (n·mean + k·prior) / (n+k), k=10), so a country with two ratings can't swing the board.
• Sample gate — a place must clear at least 5 submissions and 5 happiness ratings to be ranked; those that don't appear in the separate “not yet ranked” list.