Magpie Tests 🐦‍⬛

The personality tests you usually see hand you 12–16 questions, then combine your answers into a type. Change an answer and the result changes; if you know which outcome you want, you can steer toward it. Magpie Tests don't work that way. The only thing asked is your birth date, which is rewritten in the notation of an old calendar system, and the eight characters that come out are dropped into the world you chose.

That means the result can't be gamed. If there's a role you want, you won't get it unless you were born on a different day. What you get instead is something easier to line up next to your friends' results: different birthdays mean the answers came from genuinely different inputs, and two people born on the same day really do land on the same result.

The eight characters themselves aren't a prophecy — they're calendar data. Traditional East Asian calendars label each day with a stem-branch pair the way a Western calendar labels weekdays; your chart is simply the year, month, day and hour of your birth transcribed into four such pairs. What those characters mean is a separate question, and Magpie Tests answer it with world-building rather than fortune-telling.

The LOVES Name Test doesn't ask questions either, but it takes two names instead of a birth date, so changing a name does change the result.

Eight of the eleven are built on the ten possible day masters — the heavenly stem of the day you were born — plus two special conditions that expand the outcomes to twelve. What differs is which energies each world rewards. There's no reason the trait that keeps you alive in a zombie outbreak should be the same one that gets you promoted in a Joseon government office. The Special Move Test runs on a different axis: instead of the day master it uses the day branch — the earthly branch of the day you were born — whose twelve values map straight onto twelve moves, while the day master's element only sets the move's element. The Combo Attack Test pairs two of those moves and lets the relationship between the two day branches — harmony, triad, season or clash — decide the result. The Career Type Test turns the day master around: it leaves the day master out of the result and counts how every other character relates to it — the Ten Gods — so people with the same day master can still land on different types.

Take several and the difference shows. A high zombie survival rate can sit next to a low Joseon success index, and the reverse is just as common. Neither is the "good" result — it means the worlds ask for different temperaments.

How each test maps the ten day masters (or, for the special move, the twelve day branches; for the combo attack, the relationship between two of them; for the career type, the ten Ten Gods) onto its roles, and which constants it uses to compute its number, is written out in full on each test's own page.

All eleven tests run the same calculation engine as the Korean-language manse-ryeok calculator (in Korean). The year pillar turns over not on January 1 and not on Lunar New Year, but at the exact moment of ipchun, the solar term that begins spring; the month pillar turns over at each of the twelve solar terms. These are timestamps, not dates — ipchun in 2026 falls at 5:02 a.m. on February 4 — so two people born on the same February 4 can end up with different year pillars. The day pillar comes from the Julian Day Number modulo 60.

The reason one engine doesn't produce identical answers is that each test carries its own base values and weights. The first seven are tuned so that feeding one birthday into all of them produces coincidentally similar numbers less than 5% of the time — it's simply more fun when the results argue with each other. The Special Move Test came later and isn't part of that tuning, but because it reads the day branch rather than the day master, people who share a day master still split across six different moves. The Combo Attack Test simply runs the Special Move calculation once for each of you and pairs the results. The Ghost Power Test shows folklore rather than a score, so it isn’t part of the tuning either; like the Joseon test it decides by the day master, so outside its two special types, sharing a day master means sharing both results. The Career Type Test looks at everything except the day master, so your Joseon result tells you almost nothing about your career type.

On the first seven, the "top N%" badge isn't an estimate. Every one of the 22,280 days from January 1, 1950 to December 31, 2010 was computed in advance, and the resulting distribution table ships inside the page. Your value is looked up against that table.

The Special Move Test's power level and top-% badge come from a larger table: all 19,359 days from January 1, 1960 to December 31, 2012, each combined with the 13 birth-time options (unknown plus the twelve two-hour periods) — 251,667 cases in all. The Combo Attack Test's top-% badge ranks your combined power against every possible pair drawn from those same 251,667 cases. The “N% of everyone” shown by the Ghost Power and Career Type tests — the share of people who get that type — comes from the same 251,667 cases.

The birth date you type is never sent to a server. The entire computation finishes inside your browser, and nothing is stored. There is no sign-up, and nothing is withheld behind an email box.

The same holds when you copy a result link to share it. The link carries only the finished result values — never your birth date. Two people with different birthdays who land on the same result get byte-identical links. Whoever opens your link sees the result with every input field left empty. For the two-player Combo Attack Test, result and invite links likewise carry only special-move result values — never names. The LOVES Name Test treats names the same way: they're counted in your browser, and its links carry only the result.

The shareable result card image is drawn inside your browser and downloaded from there. It is never uploaded anywhere.

Do I need a Korean birth date or lunar calendar date?

No. Enter your Gregorian (solar) date — the one on your passport. If you only know a lunar-calendar birthday, convert it first with the lunar-to-solar converter (in Korean) and enter the converted date.

What if I don't know my birth time?

Leave it blank. The test then computes with six characters instead of eight, dropping the hour pillar. Year, month and day pillars are fixed without a time, so the backbone of the result is unchanged — the numbers shift slightly, and if you were born around 11 p.m. or on the day a solar term begins, the role itself can change with the time.

Another saju app gives me a different day pillar

Usually it's a difference in standard-time correction and the late-Rat-hour convention. Korea's clock runs on the 135°E meridian, roughly 30 minutes ahead of true local solar time, and schools of practice disagree about correcting for it. The manse-ryeok calculator (in Korean) explains both conventions and lets you toggle them.

Is any of this a real fortune reading?

The calculation is real; the interpretation is a game. Nothing here predicts your future, and no result carries any claim about your life. If you want the eight characters themselves taken seriously, that's what the tools linked above are for.

I don't like my result

None of the tests has a losing outcome. A low survival rate comes attached to a different talent, and the roles that look like defeats — Patient Zero, the Demon Lord — are deliberately the rarest ones to draw. If a number came out low, the interesting move is to see what the other worlds say about you.

Will there be more tests?

Yes, they're still being added. New ones appear at the top of this page.