Name compatibility is an old game: turn two names into numbers, then reduce them a digit at a time. This tool refuses to hide the arithmetic. It shows the value of every letter, every rung of the ladder, and reads the final two digits as a percentage. Latin names use Pythagorean numerology, Arabic script uses abjad, Hangul uses jamo strokes and Japanese uses kana strokes — so a name is always counted the way its own writing system counts.
There is no randomness. The same two names with the same method always give the same number — refreshing will not buy you a better score.
Order matters. The letters of the two names alternate, so swapping them changes every neighbouring pair. That is how the traditional method works; the swap button lets you see both.
99% is the ceiling. The last two digits are read as tens and units, so the combination that would make 100 cannot occur.
Turning names into numbers was not invented once and exported. It grew separately in different places: Korean tradition counts the strokes of the jamo, Western numerology assigns 1–9 across the alphabet, the Arabic-writing world uses abjad values, and Japanese practice reads stroke counts.
That is why this tool does not impose one universal algorithm. Feeding every name through, say, Unicode code points would produce a rule belonging to no culture at all — and then there is no answer to «why this number». Instead the script you typed selects the tradition that actually belongs to it.
The consequence is that mixing scripts leaves one side uncountable. A comparison needs a single ruler, and blending two traditions in one row destroys it. When that happens, transliterate one name into the other script and measure both the same way.
A Hangul syllable decomposes into an initial consonant, a vowel and an optional final consonant. Unicode defines the syllable block as exactly that composition, so the three jamo can be recovered from the syllable mechanically. Each recovered jamo is then given its stroke count and the syllable's value is their sum.
Single consonants take the strokes a pen actually makes: ㄱ is 1, ㄷ is 2, ㄹ is 3, ㅂ is 4. Doubled consonants (ㄲ, ㄸ, ㅃ, ㅆ, ㅉ) are twice their single form, and cluster finals (ㄳ, ㄺ, ㅄ …) are the sum of their parts. Among vowels ㅡ and ㅣ are one stroke, ㅏ and ㅗ are two, and compound vowels such as ㅘ and ㅞ add their components.
So 김 is ㄱ(1) + ㅣ(1) + ㅁ(3) = 5, and 수 is ㅅ(2) + ㅜ(2) = 4. Those numbers appear in the per-letter row, so the sum can be checked by hand.
Latin letters run A=1, B=2 … I=9, then start again at J=1. S is 1 again and Z is 8 — twenty-six letters folded into nine slots, so every letter carries a digit from 1 to 9.
Accents are stripped before counting: é becomes e, ñ becomes n, ü becomes u. Without that step Spanish and French names would fall out of the calculation entirely and the tool would be useless in those languages.
Case is ignored. Capitalisation is a writing habit, not a different letter.
The Arabic script carries an old numeral system called abjad (ḥisāb al-jummal). Before Arabic digits took over it was how years and quantities were written, and it still survives in chronograms and traditional reckoning.
Units run ا=1, ب=2 … ط=9; tens run ي=10, ك=20 … ص=90; then ق=100, ر=200 … غ=1000. This tool uses the Mashriqi (eastern) ordering — the Maghrebi ordering assigns ص, ض, س and ش differently, so the same name can score differently under it.
Letters written with hamza or madda (أ, إ, آ) fold to plain ا, and ة folds to ه by convention. Urdu-specific letters (ٹ, ڈ, ڑ …) fold to their Arabic counterparts.
Japanese seimei-handan reads the stroke count of a name. Kanji stroke counts, however, differ per character, and the common set alone exceeds two thousand — an unverified table would corrupt the total without showing where.
So only kana strokes are counted here. Hiragana and katakana are about fifty characters each, small enough to check by eye. A dakuten adds two strokes, a handakuten adds one, and small kana (っ, ゃ) count the same as their full-size forms.
Enter a kanji name by its kana reading. The scope is narrower, but every number printed on screen is one that was actually verified.
Once every letter has a value, the letters of the two names are laid out alternately in a single row. Then each pair of neighbours is added and only the last digit is kept, which shortens the row by one on every rung.
Repeat until two remain and you have a tens digit and a units digit — a percentage. Stopping at three digits gives something that cannot be read as a percentage; reducing all the way to one leaves only ten possible outcomes and the resolution collapses.
Systems with large values (abjad reaches 1000) are reduced modulo ten at the first step so every tradition enters the ladder as single digits. Otherwise Arabic names would always begin with large numbers and the ladder would behave quite differently for them.
This tool guarantees exactly one thing: that the values in the tables were applied by the stated procedure, correctly. It makes no claim that the tables or the procedure predict anything about people.
Stroke tables and numerology tables are folk tradition, not standards maintained by any body, so they vary between sources. The method in use is named on screen — when another site disagrees, the table is usually the reason.
Enjoy it, but do not carry the result over into a judgement about a person. That is not something this arithmetic can do.
It is not the kind of number that can be accurate or inaccurate. The method did not come from observing people or relationships — it comes from an old habit of turning letters into numbers. What this tool guarantees is that the arithmetic is done correctly and shown openly.
The first step lays the letters out alternately, one from each name. Swapping changes every adjacent pair, and therefore every rung above it. The traditional method is order-sensitive, so we did not force it to be symmetric — we added a swap button instead.
Kanji stroke counts differ per character and the common set alone runs past two thousand. Shipping an unverified table would quietly mix wrong counts into the total. Japanese names are therefore entered as their kana reading, and the kana strokes are counted.
They are marked as uncountable and left out of the arithmetic — and listed on screen, not silently dropped. You cannot interpret a result if you do not know which characters were excluded.
No. Everything is computed in your browser and no name leaves the page. The shareable image is drawn locally too.
What this tool bases its numbers on, and how far those numbers go.
① Each letter is valued by the tradition of its script (Hangul = jamo strokes, Latin = Pythagorean A=1…I=9, Arabic = abjad ا=1…غ=1000, kana = stroke counts). ② The two names' letters alternate into one row and each value is reduced modulo 10. ③ Adjacent pairs are added modulo 10, repeatedly, until two digits remain. ④ Those two digits read as tens and units are the percentage.«Michael ↔ Sarah» (Pythagorean): M4 I9 C3 H8 A1 E5 L3 interleaved with S1 A1 R9 A1 H8, reduced rung by rung until two digits remain.