The Korean analyst list nobody keeps

The number that looks like a rank is not one

The source hands back a field called rank. We checked what it orders before we used it, and it orders nothing: the page requests its own list sorted by analyst name, so that number is a row position in an alphabetical list. On the 2026-09-18 snapshot the names run in perfect Korean alphabetical order while the scores beside them read 8.16, 7.79, 8.10, 9.40 — not descending.

We store it as row_in_name_order and never print it as a standing. A table below is ordered by merit, and we ordered it — from the score the source published. Saying "the number 1 Korean analyst" from this data would have been false, and it would have been an easy thing to publish.

The accuracy column is blank more often than it is filled — for a third of rows

The source prints 0 in that column when it has no figure. We do not carry that through as an accuracy of zero — a zero would say the analyst was never right, which is a different and much worse claim than "not published". Those rows are null with a flag.

Scores move even when the analyst does not

Because the window rolls, a score changes on days when nothing was published. Across the 75 analysts present in three or more complete snapshots, the score moved by a median of 0.28 points between its own high and low, and by as much as 2.58.

We report the spread and not a reading of it. A score that drifts with the calendar is worth knowing about before anyone treats it as a verdict.

Who is on the list, by house

The roster below is the most recent complete snapshot (2026-09-18), 74 analysts. Report counts are for the whole collected window, 840 reports.

The two lists on the same site do not cover the same houses

The dash in that last column is not zero. The analyst list and the report list come from the same publisher and disagree about who exists.

Two of the organisations filing reports are not brokerages at all — a commercial bank and an investor-relations body — which is one reason they carry no analyst rows. We keep the category on each institution rather than counting them as brokerages that published nothing.

Ordered by score — an order we computed

Ten highest published scores on the 2026-09-18 snapshot. The source did not rank these. We sorted them by the score it published, which is a different claim from the site having called anyone first.

Analyst names are withheld here and carried in the licensed file. A public leaderboard of named individuals, built on a score that drifts with the calendar, is not something this page should create.

How we counted — including where our own collection fell short

2 of our own snapshots came back short (2026-09-12, 2026-09-13) and are excluded from every count above. Treating them as complete would have produced a headline that the roster collapsed to 20 people overnight, which is not what happened — our collection stopped early. A snapshot is flagged short when it holds fewer than half the median row count, so the rule catches the next one too without anyone remembering these dates.

• Not a ranking of analysts. The source hands us its list sorted by name, so the number it calls a rank is a row position. Any order on this page is one we computed from the score the source published, and we label it that way.
• Not a measure of who is right. Whether a target was met is a separate page.
• Not a full roster of Korean analysts. It is the list one Korean site showed on the days we collected it.
• Not back-fillable. The source recomputes a rolling one-month window every day and keeps nothing older, so these snapshots exist only because we saved them.