94 K-pop acts peaked in January. 38 in September. February never wins big

We already knew, for 774 Korean musical acts, the single month each one was read most on English Wikipedia. We used it to ask how fast attention drains away after that month. We never asked the simpler question sitting in the same column: which month is it?

It is not spread evenly.

Five of the twelve months sit more than two standard deviations from expectation: January (+3.64) and April (+3.34) above it, September (−3.11), February (−2.73) and November (−2.48) below it. The seven that remain are unremarkable, and that is worth saying plainly — the calendar is uneven, not chaotic.

The correction matters more than the finding

The third column is the part worth reading first. February has 28 days. A short month collects fewer reads, so it is less likely to hold an act’s best month for reasons that have nothing to do with music or scheduling. Our window also opens in July 2015 and closes in August 2026, which means July and August appear twelve times while every other month appears eleven.

Both of those are instrument artefacts, and both point the same way as the result — which is exactly when a number should be distrusted. So the expectation above is not a flat one-twelfth. It is proportional to the days each calendar month actually contributes.

The correction changes the answer:

February is still low. July is still high but much less so — a third of its apparent excess was the calendar. September barely moves, which is why it survives as the strongest low.

Among the 100 biggest peaks, February is empty

Restricting to the 100 acts whose best month drew at least 50,000 reads removes the risk that a small article with a noisy month is driving the pattern.

Not one of the hundred biggest peaks landed in February. Everything else in this table is inside two standard deviations of expectation — April, the highest, reaches only +1.73. The January and September extremes that were so clear in the full set are gone.

That last point is the honest one. The pattern in the 774 is real; in the 100 it mostly dissolves. Whatever produces it works on acts read in the thousands, not the hundreds of thousands. The more quotable version of this article — “the biggest K-pop moments happen in January” — is not what the numbers say.

What we did not measure

We do not hold release dates. This is the month an act was looked up most, and reading it as guidance about when to schedule a comeback is a claim about data we never collected. There are at least three ordinary explanations we cannot separate with what we have: awards season and year-end lists sitting in January, the academic and fiscal calendar, and the plain fact that a Wikipedia article is usually written after an act becomes known.

What the table does establish is narrower and still useful: the calendar is not flat, the flatness you would assume is wrong by a third in both directions, and two of the three months that stand out do so only once you stop dividing by twelve.

The full table, with the day counts and both subsets, is on the peak-month page. The same 774 acts lose half of that peak within two months of reaching it, and Korean titles halve on the same timescale measured on entirely different data.