Why annual weather scores mislead

The mistake, exactly as we made it

We wanted a list of warm winter bases. The obvious move was to filter the dataset for a high weather rating and a low cost of living, and see what came out. Near the top of the result were Burgas and Varna — both on the Bulgarian Black Sea coast, both rated 4.2 for weather, both around $1,000–$1,100 a month. On paper, a warm cheap winter.

In reality, Varna sits at 43.2 degrees north. On the shortest day it gets 8h 59m of daylight and the sun goes down at 16:36. Its winter is a cold grey Black Sea winter. The rating was not wrong — Varna genuinely has a pleasant year overall, with a long warm summer and a mild autumn. The rating was answering a question about twelve months, and we had asked it a question about one.

Every city where the same trap is set

There are 311 cities in our dataset rated 4.0 or better for weather. 25 of them sit north of 42 degrees, which is far enough north that the December day is under nine and a half hours and the winter is a real winter. These are the cities where the annual rating and the winter reality diverge most sharply.

High annual weather ratings at high latitudes — the trap set.

None of these is a bad city. Several are excellent. But a 4.4 annual rating in Montenegro and a 4.4 annual rating in the Canaries are describing two different shapes of year, and only one of them is describing a January you would want to spend outdoors.

What a warm winter actually requires

Two things, and the annual rating expresses neither directly. The first is latitude low enough that the sun is still meaningfully high in the sky in December. The second is a climate that is dry or mild in the specific months you are there, which for many subtropical places is not the same as being pleasant in July.

Compare like for like. Las Palmas sits at 28.1°N and gets 10h 21m on the shortest day, with sunset at 18:10. Dahab is at 28.5°N with 10h 20m. Varna, rated the same for weather as neither of them but close to both, gets 8h 59m. That is roughly an hour and twenty minutes of daylight difference every single day, and it is the difference between finishing work in daylight and finishing work at night.

Why we did not just fix the rating

Because an annual rating is the right tool for the question it was built for. If you are choosing somewhere to live for a year, the shape of the whole year is what matters and collapsing it to one number is reasonable. Splitting it into twelve monthly ratings would need twelve times the sourced data, and by the standard on our temperature verification page most of that data does not exist in a form we would publish.

The honest fix is not a better number. It is telling you what the number means, which is what this page is. The same principle applies to the headline city score, which turns out to measure something quite different from what most people assume — see what our city score really measures.