Kundli Matching (Guna Milan)

What this page covers that the match tool does not print

The button above opens the free match tool, which takes two birth dates, optional times and places, and returns a Guna Milan score out of 36 with all eight kootas listed. The full tables behind that score - which sign is in which Vashya group, which nakshatra has which Gana, Nadi and Yoni, who counts as whose friend - are already printed on the Compatibility feature page, together with the verdict bands and the numerology cross-checks. This page does not repeat them. It does three other things.

• It walks through the mechanics of the kootas that decide most marriage scores - Varna, Vashya, Tara, Bhakoot and Nadi - as the code actually computes them, including the arithmetic of the nine-star Tara cycle and the sign distances Bhakoot marks down.
• It shows how the scores fall across every possible pair of Moons, so a number like 24 can be placed against the whole range rather than read in isolation.
• It explains, with three real runs of the tool, what happens when one or both birth times are missing: why the tool still gives a number, which points it calls settled and which it calls at risk, and how far apart the possible answers really are.

Graha Maitri, Gana and Yoni are read as temperament, and the Business and Friendship match types together with Mulank and Bhagyank harmony are covered on the sibling page, Love compatibility. They are mentioned here only where they move a marriage score.

What the classical books in this library say about matching

Before the mechanics, the provenance. For this page the app's whole source library was searched for the matching vocabulary: Bhakoot, koota or kuta, melapaka, Nadi, Gana, Yoni, Tara and marriage. The result confirms what the header of compatibilityEngine.js already says. None of the books listed on the sources page - Brihat Parashara Hora Shastra (BPHS), Saravali, Bhrigu Sutras, Advanced Predictive Astrology, Daivajna Vallabha or the numerology books - states a rule for comparing two birth Moons by kootas. The word Nadi occurs only as part of a book title quoted in Advanced Predictive Astrology. Daivajna Vallabha has a section on marriage, but it is horary: it reads the chart for the moment a question is asked, not two birth charts against each other.

What the books do supply is raw material that the kootas reuse. Four passages matter here.

• BPHS Ch. 4 gives a social class for nine of the twelve rashis (for example Mesh is "related to kings", Kark "represents Brahmins"). The engine's Varna table follows those nine and fills Mithun, Makar and Meen from their element. The book uses the classes to describe signs; it does not compare two of them.
• BPHS Ch. 3 Sl. 55 derives natural friendship from each graha's moolatrikona sign and also names neutral and enemy states. The engine's friend and enemy lists match that derivation, and since October 2026 Graha Maitri is graded on all three states.
• BPHS Ch. 44 Sl. 15-21 names the 3rd, 5th and 7th stars from the birth star Vipat, Pratyak and Vadh, in a chapter on maraka periods. Those are the three counts the Tara koota marks down.
• BPHS Ch. 46 Sl. 207-209 describes a Tar Dasha in which "Janm, Sampat etc." replace the grahas in order - the only place in the library where the nine tara names start to appear as a sequence. The text names the first two and leaves the rest as "etc.".

So the honest position is this. The ingredients - sign classes, friendships, the counted stars - are classical and cited. The scoring rules that turn them into 36 points are this app's implementation of the matching system as it is commonly practised, and no book in the library can be quoted for the points themselves. For Bhakoot, Gana and Graha Maitri the engine's header names the published web references it follows where they agree (October 2026). Where the engine differs from how the system is usually described, this page says so koota by koota.

The eight kootas as a marriage score, one at a time

Every koota reads one of two inputs from each chart: the Moon's sign or the Moon's nakshatra. Varna, Vashya, Graha Maitri and Bhakoot read the sign; Tara, Yoni, Gana and Nadi read the nakshatra. The controller keeps that split in a table called KOOTA_DRIVER, and it matters later, when a birth time is missing. If you are unsure what your own inputs are, the Moon sign calculator and the nakshatra calculator compute them from the same engine.

The table below is new material: it counts how often each koota returns each value when the engine is run over every ordered pair of the 108 nakshatra quarters - 11,664 pairs in all, one Moon placed at the middle of each quarter. It shows what a typical pair earns on each koota, and which kootas actually separate one couple from another.

Two things stand out. First, Bhakoot and Nadi carry 15 of the 36 points and each is all or nothing, so between them they produce most of the spread: a pair that loses both starts 15 points down before anything else is counted. Second, Yoni barely discriminates in this engine - 85.6% of pairs get the middle value of 2 - because only exact animal matches score 4 and only the seven named enemy pairs score 0.

Varna: one point, and the only koota that cares about order

Varna is the classical name of this koota and of the four sign classes it compares; on this page it means nothing more than a class assigned to a Moon sign. The engine gives each sign an index from its element and awards the point when the first person's index is equal to or higher than the second's. Of the 144 ordered sign pairs, 54 score 0 and 90 score 1.

Because the rule compares "first" with "second", it is the one koota whose value can flip when the two people are swapped. Every one of the 108 sign pairs with different classes flips; only the 36 same-class pairs do not. Over the 11,664 quarter pairs, swapping the order changes the total in 8,748 of them (75%), always by exactly one point. The free tool has no field for gender and does not reorder anyone, so whoever is typed first is "first". The engine header records this direction-dependence as an open audit point, left unchanged because no book in the library says which partner's class should be compared with which.

Vashya: five groups, and a floor of half a point

Vashya puts each Moon sign into one of five numbered groups. Same group scores 2; groups whose numbers differ by one score 1; everything else scores 0.5. There is no 0, so the lowest Vashya can go is half a point. Over the 144 sign pairs, 36 score 2, 60 score 1 and 48 score 0.5. The groups carry no names in the code, and the "adjacent number" rule depends on how the groups happen to be numbered rather than on a stated relationship between them - one more reason the engine header treats the Ashtakoot point scales as an implementation, not a citation.

Graha Maitri, Gana and Yoni in a marriage score

These three are read as temperament on the Love compatibility page. For a marriage score, what matters is their weight: together they hold 15 points, the same as Bhakoot and Nadi, but they rarely all go to zero. Graha Maitri goes to 0 only when the two sign lords are natural enemies of each other (the Sun with Venus or Saturn). Gana goes to 0 only for a Deva Moon with a Rakshasa Moon, which happens in 162 of the 729 star pairs. Yoni goes to 0 in 52 of the 729. Across all 11,664 quarter pairs, these three average 8.5 of their 15 points.

The nine-star Tara cycle, counted out

Tara counts from one person's birth star to the other's, both ways, inclusive of the starting star. The engine divides each count by 9 and looks at the remainder. A remainder of 3, 5 or 7 fails; any other remainder passes. Both counts passing scores 3, one passing scores 1.5. Those three failing positions are exactly the 3rd, 5th and 7th of each nine-star round - the Vipat, Pratyak and Vadh stars of BPHS Ch. 44 Sl. 15-21, where the text uses them for dasha periods. Using them to match couples is the common practice; the book itself does not.

Because 27 stars make three rounds of nine, every count from 1 to 27 lands on one of nine positions. The table shows all nine positions, the forward counts that land there, and what the return count is. For two different stars the two counts always add up to 29, so a forward count of 4 means a return count of 25.

Read down the last column and the shape is clear. A pair scores the full 3 only when the stars sit at positions 1, 2 or 9 from each other - the same star, the next star, or the star before, in any of the three rounds. Every other relationship has one failing direction and scores 1.5. The failing positions never meet: if one direction is at 3, 5 or 7, the other is at 8, 6 or 4, which pass. So in this engine Tara can return 3 or 1.5 and never 0. Over the 729 ordered star pairs, 243 (one third) score 3 and 486 score 1.5, which is why the average Tara score is exactly 2.

The same arithmetic explains a puzzle people notice in practice: two partners whose Moons are one star apart always score full Tara, while two stars apart always scores 1.5. A day is about one star along the zodiac. Stars one apart sit at positions 2 and 9 relative to each other - both pass; stars two apart sit at positions 3 and 8, and the 3 fails.

Bhakoot: the sign distances, one by one

Bhakoot counts from one Moon sign to the other, inclusive, and checks the result both ways. If either count is 2, 5, 6, 8, 9 or 12, Bhakoot scores 0; otherwise it scores the full 7. Since a count of 2 one way is 12 the other way, 5 one way is 9 the other, and 6 one way is 8 the other, there are three failing relationships: neighbouring signs (2/12), signs in trine (5/9) and signs six apart (6/8).

The 5/9 row changed in October 2026. Until then this engine gave Moons five and nine signs apart the full 7, where most descriptions of Guna Milan treat them as a Bhakoot failure alongside 2/12 and 6/8; an internal audit (point #29) flagged it. No book in the library contains a Bhakoot table, so the engine now follows two published references that agree on all three failing pairs, named in its header. A score computed here before that change for a 5/9 pair was 7 points higher than it is now.

In practice: 6 of the 12 sign distances fail, and since every distance occurs equally often across the 144 sign pairs, exactly half of all pairs lose Bhakoot - 5,832 of the 11,664 quarter pairs. Bhakoot reads only the Moon sign, so two people whose Moons are in Leo and Virgo, or Leo and Sagittarius, lose it whatever their nakshatras are, and two whose Moons are in Leo and Libra keep it.

Nadi: eight points on a three-way split

Nadi sorts the 27 nakshatras into three groups of nine - Aadi, Madhya and Antya - in a repeating zig-zag that the feature page prints in full. Same Nadi scores 0, different Nadi scores 8. Because each group holds exactly nine stars, any given star shares a Nadi with nine of the 27, itself included. So exactly one third of pairs lose Nadi: 3,888 of the 11,664.

Since Nadi reads the nakshatra and Bhakoot the sign, the two fail independently of each other in the counts, and the joint figures follow almost exactly from the thirds. Of the 11,664 pairs, 1,700 (14.6%) lose both and score at most 21; 8,020 (68.8%) lose at least one; only 3,644 (31.2%) keep both. In other words, more than two thirds of possible Moon pairs give up at least 7 points on these two kootas alone.

The tool's note for a shared Nadi reads "traditionally flagged; worth discussing openly". It does not attach any health claim, it does not mention exceptions or cancellations, and it suggests no remedy. Exceptions that some practitioners apply - for instance when the two stars differ though the Nadi is the same - are not computed. If the engine does not compute something, it does not change the score.

Where any one score sits among all of them

The feature page gives the floor, ceiling and median. The table below adds how the 11,664 quarter pairs split across the four verdict words the tool prints for a marriage reading. The words come from the score as a share of 36: 75% and above, 55%, 35%, and below.

The mean total is 21.07 (median 21.5, range 4 to 34.5). The single most common total is 23, reached by 740 pairs; 561 pairs score 28, among them every identical pair of Moons, which wins every koota except Nadi. The distribution is lumpy rather than smooth, because two kootas move in steps of 7 and 8: totals ending in .5 are common around 20 to 24 and rare above 30.

Two cautions follow. These counts treat every quarter as equally likely, which is a property of the tables, not of any real population, so they say how the scoring behaves, not how couples behave. And the verdict words have no special status: the code has no pass mark, nothing is blocked or approved, and a score sits in a band only because of the thresholds written into computeCompatibility().

When a birth time is missing: what the tool actually does

The match tool asks for dates and treats times as optional, and the controller's comment gives the reason: the Moon is computed geocentrically, so the birth place changes nothing in Guna Milan - only the birth time does, by moving the Moon along the zodiac. A place is still used to find the time zone; with no zone at all, the tool assumes Indian Standard Time and says so in the response. When a time is missing, four steps follow.

1. 1.readBirthInput() casts the chart for 12:00 noon on the birth date, in the birth zone. The comment calls noon "the least-wrong single choice" - the midpoint of the unknown day. The headline score uses this noon Moon.
2. 2.birthDayInstants() finds the first and last minute of the birth date, local 00:00 and 23:59. If the zone changes clocks that day, each end is taken on its own offset.
3. 3.moonCertainty() computes the Moon at both ends and records the nakshatra and sign at each. If both match, that person's inputs are settled whatever the time was. If not, it records which changed.
4. 4.pointsAtRisk() then looks up each scored koota in KOOTA_DRIVER. A sign-driven koota is at risk only if a Moon changed sign during its day; a nakshatra-driven koota is at risk if a Moon changed nakshatra. One uncertain person is enough, because every koota compares the two. The response marks the score provisional and gives the points at risk and the points settled.

How often does this happen? The Moon moves about 13 degrees a day and a nakshatra is 13 degrees 20 minutes wide, so a nakshatra change during the day is the norm. Running the same check over every birth date from 1980 to 2005 in Indian time, the Moon changed nakshatra during the day on 9,100 of 9,497 dates (95.8%) and changed sign on 4,167 (43.9%). So a missing time almost always puts at least the 21 nakshatra-driven points at risk, and a little under half the time all 36.

Three couples, run through the real tool

The three examples below are invented couples run through the actual match() handler; every number comes from those runs. They are chosen to show the three cases: a Moon that changes nakshatra but not sign, a Moon that passes through three nakshatras and two signs in one day, and two partners who both lack a time.

Meera and Kabir: one nakshatra boundary, 21 points flagged, 6 actually move

Kabir knows his time: 21 July 1993, 08:40, Delhi, which puts his Moon in Cancer, in Ashlesha. Meera knows only her date, 12 March 1996. Her Moon is in Scorpio all day, but it moves from Anuradha into Jyeshtha at about 02:16 that morning. At noon it is in Jyeshtha.

The tool prints 26 of 36, marks it provisional, names Tara, Yoni, Gana and Nadi as at risk, and says 15 points are settled. That is exactly right as far as it goes - but only one of the four flagged kootas actually changes. Tara stays at 3 (Anuradha and Jyeshtha are neighbours, and both sit at a passing position from Ashlesha), Yoni stays at 2, Nadi stays at 8. Gana alone moves, by 6 points. And the window for the lower score is only the first two hours and sixteen minutes of the day. If Meera's family remembers that she was born in daylight, the 26 stands; if not, the two possibilities are 20 and 26, both Workable.

Riya and Kabir: a day with three nakshatras

The Moon moves faster at some points in its orbit, and on 13 March 1996 it covered more than one full nakshatra in a day. Riya, born that date with no time recorded, has a Moon in Scorpio-Jyeshtha until 01:07, in Sagittarius-Mula from 01:08, and in Sagittarius-Purva Ashadha from 23:41. Matched with Kabir:

With no time, the tool prints the noon result, 24.5, and since her Moon changed sign as well as nakshatra, all 36 points are at risk. The precision panel reports the Moon at the two ends of the day - Jyeshtha at the start and Purva Ashadha at the end - while the headline itself shows Mula, so all three possibilities are visible on the card. But the time each one covers is not: Mula holds for 22 hours and 33 minutes of the 24, and the other two together for about an hour and a half. The tool does not weight the possibilities by time; it reports them. Reading the card with that in mind, 24.5 is far more likely to be Riya's actual score than either extreme, though here all three possibilities fall in the same Workable band.

Isha and Dev: both times unknown

Isha was born on 8 March 1996 and Dev on 5 November 1994; neither has a time. Isha's Moon is in Virgo-Hasta until 03:05, Virgo-Chitra until 15:19, then Libra-Chitra. Dev's is in Scorpio-Anuradha until 17:54, then Jyeshtha. At noon the tool takes Virgo-Chitra and Anuradha and prints 13.5 of 36, Mixed, provisional, all 36 points at risk. The six real combinations are:

From the same two dates, the possible marriage scores run from 8.5 (Challenging) to 27.5 (Strong) - the entire useful range of the scale. Gana and Nadi swing together here, by 14 points: Hasta and Anuradha are both Deva with different Nadis, and so are Chitra and Jyeshtha both Rakshasa with different Nadis, but cross them and both kootas fall to 0; Bhakoot takes a further 7 once Isha's Moon crosses into Libra, which sits next to Dev's Scorpio, and Graha Maitri gives back 2 (Venus and Mars are neutral to each other). This is the case the provisional flag exists for. A 13.5 printed with "36 points at risk" is not a score in any useful sense; it is a prompt to find at least one birth time.

The Business type on the same pair shows the risk being counted only over the kootas that feed the headline. It scores 9.5 of 21 and lists 21 points at risk - Tara, Graha Maitri, Gana and Bhakoot - rather than 36, because the controller passes only the scored kootas to pointsAtRisk(). A card can never say more points are unsettled than the score is out of.

Engine and common practice, side by side

Collected in one place, these are the points where this engine differs from the way Guna Milan is commonly described, with what the library's books say and what the app does. They explain most of the disagreements people see between this tool and others. Three of them - Bhakoot 5/9, Ardra's Gana and the Graha Maitri scale - were changed in October 2026 to follow published references that agree; the table shows the current behaviour.

Disagreement can also come from the Moons rather than the tables. A different ayanamsa, time zone or birth time can put a Moon in a different nakshatra, and the app's own astronomy is not ephemeris-grade - the methodology page gives the stated accuracy. Compare the two Moons first, then the koota table; a total-to-total comparison hides where the gap comes from.

Reading the match card well

• Look at the koota rows before the total. The card marks the weakest of the scored kootas and, when a time is missing, marks each koota whose input moved. A 24 built from a lost Nadi and a 24 built from a lost Bhakoot are the same number for different reasons.
• If the card is provisional, look for the person whose Moon moved. The precision panel names the nakshatra at the start and end of that person's day. Then ask the family what part of the day the birth was in; often that settles it.
• Treat the points at risk as a ceiling. As Meera's case shows, the score may not move at all across a flagged boundary.
• Kundli matching here reads two Moons only. Mars placement is a separate check on the Mangal Dosha calculator, and the 7th house is read on the full chart; the 7th house explainer covers what BPHS says about it.
• If both birth times are lost, what still works without a birth time lists which parts of astrology survive and which do not.

The tool offers no remedies, and this page suggests none. A low koota is described, never treated. Nothing in the score predicts the course of a marriage; it reports how two Moons compare on eight fixed tables.

Why does the match tool not ask for a birth place to be exact?

Because Guna Milan reads only the two Moons, and the Moon's sign and nakshatra are the same wherever on Earth you stand at a given instant. The place is used only to find the time zone, which turns a local clock time into that instant. With no place and no zone the tool assumes Indian Standard Time and flags the assumption in its response.

My score says "provisional" and 21 points at risk. Does that mean it could drop by 21?

No. It means the four nakshatra-driven kootas - Tara, Yoni, Gana and Nadi, worth 21 together - read an input that changed during the birth day. Some or all of them may score the same either way. In the Meera example 21 points were flagged and only Gana, worth 6, actually moved.

Why does the tool use noon when there is no birth time?

The code calls noon the least-wrong single choice: it is the midpoint of the day, so the computed Moon is never more than twelve hours from the true one. The tool then checks the start and end of the day separately and tells you if anything changed, rather than silently relying on noon.

Can Tara ever score zero here?

No. A failing count sits at the 3rd, 5th or 7th position of a nine-star round, and the return count then sits at the 8th, 6th or 4th, which pass. So one direction always passes and the lowest Tara score is 1.5. Two thirds of star pairs get 1.5 and one third the full 3.

Our Moons are five signs apart. Another site took 7 points off for Bhakoot. Which is right?

This app's library contains no Bhakoot rule, but the published references the engine now follows agree with that site: since October 2026 this tool also scores 5/9 as 0. If you checked here before then, the old figure was 7 points higher.

How often do couples lose Nadi?

Across every pair of nakshatra quarters, exactly one in three, because each Nadi holds nine of the 27 nakshatras. That is a property of the table, not a survey of couples. Bhakoot is lost in half of all pairs, and both are lost together in 14.6%.

Does it change anything if my birth day had two nakshatra changes?

The card still shows all three possibilities - the noon nakshatra in the headline, the start-of-day and end-of-day nakshatras in the precision panel - but not how long each lasted. On a day like 13 March 1996 the noon nakshatra covered over 22 of the 24 hours, so it is much the likeliest.

Is the Varna koota a judgement about anyone's background?

No. Varna here is the classical name of a one-point koota built from a class BPHS Ch. 4 assigns to Moon signs. The engine reads it from the element of the Moon sign; it asks nothing about a person, family or community, and it is the smallest of the eight kootas.