How Compatibility Scores Work

A Zodiux sun-sign compatibility score is arithmetic, not an opinion. It takes exactly two inputs — the position of each sign in the zodiac, a whole number from 0 to 11 — and returns four dimension scores and one overall percentage. There is no database of couples behind it, no machine learning, and no editor deciding that Capricorn and Taurus feel like an 83. Every constant is printed below, one pair is worked end to end, and one section takes the model apart to measure which of those constants is doing the work. That last part is the reason for publishing any of it: once you can see the whole machine, it is obvious how little it knows.

Two numbers in, five numbers out

Each sign has an index: Aries 0, Taurus 1, on through Pisces 11. The entire model is a function of two of those indices. Nothing about either person reaches it, because nothing about either person is ever collected — a pairing page such as Capricorn and Taurus is a static file written at build time, with no idea who is reading it.

The algorithm, in the order it runs

1. Each sign's index modulo 4 gives its element; modulo 3 gives its modality.
2. For each of four dimensions, look up both elements in the element table and average the two values. That is the element blend.
3. Do the same with both modalities in the modality table. That is the modality blend.
4. Take the sign distance — the smaller of the two ways round the wheel — and read one modifier per dimension from the aspect table.
5. Dimension score = 0.58 × element blend + 0.42 × modality blend + aspect modifier, rounded to a whole number, then clamped into the 40–97 band.
6. Overall = the four dimension scores weighted 1.2 chemistry, 1.1 emotional sync, 0.9 communication, 0.9 staying power, and rounded.

Two consequences follow immediately. The model is symmetric — averages and the shorter distance do not care which sign you name first, so Leo and Aries scores identically to Aries and Leo. And it is deterministic: a pair of signs returns the same percentage for everyone who shares it, forever, until someone edits a constant in the build script.

Step 1: element and modality from the index

The two shortcuts in step 1 look like tricks but are not. Because the zodiac cycles through Fire, Earth, Air and Water in fixed order, and independently through Cardinal, Fixed and Mutable, taking the index modulo 4 and modulo 3 reproduces the traditional assignments exactly — every sign, no exceptions. That is a property of how the zodiac is built, not an approximation adopted for speed.

If you want the background rather than the arithmetic, the four elements and the three modalities explain what each grouping traditionally means.

Step 2: the element table

Four dimensions are scored: chemistry, emotional sync, communication and staying power. Each element contributes a fixed value to each of them. These numbers are the model's editorial layer — they encode the traditional character of each element, and they are the part of this page most open to disagreement. They are not measurements of anything.

The element blend for a pair is the plain mean of the two rows. Two signs of the same element therefore get that element's own profile back unchanged, undiluted — which is why same-element pairings run high on the dimension their element owns and flat on the one it does not.

Step 3: the modality table

Modality carries 42% of the blend to element's 58%. The stated reason for the split is that element is the older and coarser division and is allowed to dominate, while modality is what separates two Fire signs from each other. Whether that split earns its precision is measured further down this page, and the answer is not flattering.

The spread here is narrower than in the element table, and deliberately so: 64 to 92 against element's 58 to 92, on top of a smaller weight.

Step 4: sign distance becomes an aspect

Take the absolute difference between the two indices, then take the smaller of that and 12 minus it. The result is a whole number from 0 to 6, and for sun signs that number is the classical aspect between them: signs four apart start 120° apart, which is a trine. This is the one place where real geometry enters a sun-sign score, and even here it is idealised. Two suns four signs apart are almost never exactly 120° apart — each sits somewhere in a 30°-wide sign, so the true angle is anywhere within 30° of 120°.

Read the square and the opposition rows carefully, because they are where the model refuses the obvious cliche. A square carries the second-largest positive chemistry modifier in the table and simultaneously the largest negative on staying power: friction, scored as friction, not as failure. An opposition gets the largest chemistry bonus of all and close to nothing anywhere else. What aspects are covers the underlying idea in a real chart, where aspects have orbs and these do not.

Twelve of the 78 pairings are a sign with itself, and they are not a special case in the code: the distance is 0, the conjunction row applies, and both blends return the sign's own profile undiluted. Scorpio with Scorpio is the highest of the twelve at 83; Gemini and Libra are the lowest, tied at 78.

Step 5: the blend, the clamp, the weighted overall

Every dimension is then one line of arithmetic: multiply the element blend by 0.58, the modality blend by 0.42, add them, add the aspect modifier for that dimension, round to a whole number, and hold the result inside 40 to 97. Rounding happens before clamping, which is why the clamp can only ever bite on a whole number.

The overall percentage is a weighted mean of those four. Chemistry counts 1.2, emotional sync 1.1, communication 0.9 and staying power 0.9 — a total weight of 4.1. Chemistry leads because the pages carrying these scores are read as romantic-compatibility pages. Those four weights are the same four the live compatibility calculator applies in its love mode, so the two never disagree about which dimension counts most — but they are otherwise separate models producing separate numbers, and nothing forces them to agree about a pair.

Worked example: Capricorn and Taurus, 83%

Capricorn is index 9, Taurus is index 1. 9 mod 4 and 1 mod 4 are both 1, so both are Earth. 9 mod 3 is 0, Cardinal; 1 mod 3 is 1, Fixed. The distance is the smaller of 8 and 4, so 4 — a trine, and the trine row of the aspect table applies.

Both signs being Earth, every element blend is just Earth's own value. The modality blends average Cardinal with Fixed: chemistry (80+74)/2 = 77, emotional sync (70+78)/2 = 74, communication (74+64)/2 = 69, staying power (74+92)/2 = 83.

Staying power is the interesting row. Two Earth signs in a trine produce a raw 101.22, which rounds to 101, and the clamp cuts it to 97. That single clamp costs the pair a point overall: (72 × 1.2 + 87 × 1.1 + 78 × 0.9 + 97 × 0.9) / 4.1 = 339.6 / 4.1 = 82.83, which rounds to 83. Left unclamped it would have been 343.2 / 4.1 = 83.71, rounding to 84.

Those are not illustrative numbers. Open Capricorn and Taurus compatibility and the breakdown table on that page prints the same element blends, modality blends, modifiers and dimension scores, because both were produced by the same function.

Which constants actually move the number

Printing a model is not the same as justifying it. So here is the model taken apart: the same scoring function run over all 78 pairs, once as shipped and then four more times with one part of it changed, to see how much each part is really contributing to the answer.

Two rows in that table deserve to be said out loud. First: the 0.58/0.42 split, the parameter this page has just spent a section explaining, barely does anything. Flatten it to 50/50 and only 20 of the 78 overall scores move at all, none of them by more than a single point. It is a rounding detail dressed as a tuning knob, and a reader would be right to treat the two decimal places in the worked example as false precision.

Second: the aspect modifier, the smallest-looking column in the model, is carrying it. Set those seven rows to zero and every pairing in the zodiac collapses into a six-point band and a single headline — 78 pairs, one verdict. Removing it moves the average pair by 5.9 points and the worst-affected by 10. So the honest one-line summary of this model is that the element and modality tables decide roughly where a pair sits, and the distance between the two signs decides almost everything about how far it moves from the middle.

Why the band is 40 to 97

Nothing in the astronomy sets those limits. They are an editorial position stated as a constant: no sun-sign pairing is hopeless, and none is a guarantee. A model this crude has no business printing a 12% or a 100%, because either number would carry a certainty it cannot support.

Running the model across all 78 unordered sign pairs — 66 different-sign pairs plus the 12 same-sign ones — gives 312 dimension scores. The 40 floor binds on none of them: the lowest raw value anywhere is 57.6, Gemini and Pisces on staying power. The 97 ceiling binds exactly twice, both times on staying power and both times on an Earth trine: Taurus with Capricorn at 101.2, Taurus with Virgo at 99.1. So the floor is a promise about the shape of the model rather than an active constraint, and the ceiling trims two pairs by four points and two points respectively — the rounding happens before the clamp, so 99.1 becomes 99 and then 97.

Across the whole set, overall scores run from 66 — Gemini and Capricorn, the single lowest — to 87, shared by the three Water trines: Cancer with Scorpio, Cancer with Pisces, and Scorpio with Pisces. The bottom two headlines never appear on a sun-sign pairing page at all. They exist because the same five thresholds are used by the compatibility calculator, which reads two real charts and can genuinely land there: start every dimension at its neutral 50, give it two charts with no aspect inside orb and sun signs in the friction combination, and it returns 50 — squarely in the fourth tier that no sign pair ever reaches.

The distribution is worth sitting with. Forty-five of the 78 pairings — well over half — land in one band, and the worst pairing in the zodiac scores 66%. If you came looking for permission or a warning, the model is not built to give you either.

Where a sun-sign score stops

This is the part that matters more than the tables above it, and it is short because it is not complicated.

It is a property of two signs, not two people

The score describes a relationship between two abstractions. Everyone born under the same pair of sun signs receives the identical number, which is the clearest possible statement of its limit: a figure that cannot vary between couples cannot be describing a couple.

It also reads one placement out of ten, twice. A Zodiux birth chart computes ten bodies — Sun, Moon, Mercury, Venus, Mars, Jupiter, Saturn, Uranus, Neptune, Pluto — plus a rising sign when a birth time is given. A sun-sign score has never seen either Moon, either Venus or Mars, either rising sign, or any angle between the two charts. Synastry, and what a sun-sign match can and cannot tell you lays that gap out question by question, including which placement answers each one instead.

And it is for reflection, not for decisions

Age, history, distance, money, timing, values, effort, whether both people want to be there at all — none of it is an input, and no astrological model has access to any of it. Published in full, a score is a decent neutral vocabulary for naming a pattern you already recognise. It is not evidence, and it is not a reason to start or end anything. Zodiux publishes the model because a percentage with no visible derivation is worse than no percentage at all; what Zodiux does not claim states the wider boundary, in rules.

What the compatibility calculator does instead

The compatibility calculator is different software with a different model, and it is the one to use if you want a read on two specific people rather than two signs. It is worth documenting here because its seams are different from the ones above; what synastry is covers the same calculator from the reader's side, with its own per-contact weight table.

It computes both birth charts in the page, using the same Astronomy Engine build the birth chart calculator uses — served from this origin, not a CDN — then compares seven bodies in one chart against the same seven in the other: Sun, Moon, Mercury, Venus, Mars, Jupiter and Saturn. That is 49 ordered comparisons, and they are all genuinely distinct, because your Sun to their Moon and your Moon to their Sun are two different measurements.

What the calculator does, and where it simplifies

• Five aspects, with real orbs. Conjunction, trine and opposition within 7°, square within 6°, sextile within 5°. Those windows do not overlap, so no contact can match two aspects and the order they are tested in never decides anything.
• Conjunctions always count as harmonious. That is a real simplification: a Mars-to-Saturn conjunction is read the same way as a Venus-to-Venus one. In a chart read by a person it would not be.
• Weights are hand-set, and most contacts do not have one. Sun to Moon is the heaviest at 14, then Venus to Mars at 13, Moon to Moon at 12, Mercury to Mercury at 10. Only 23 of the 49 comparisons carry a hand-set weight; the other 26 — every Jupiter contact, every Mercury contact except Mercury to Mercury, Sun to Mars, and Saturn to Saturn — fall back to a default of 4.
• Dimensions start neutral and get pushed. All four begin at 50, rise on harmonious contacts and fall on hard ones, then take an adjustment from the two sun signs' elements, and are finally clamped to 6–97 — a far wider band than the sign-pair model, because two real charts carry enough information to justify a low number.
• A missing birth time is allowed, and costs you the rising sign. With no time the calculator assumes local noon and computes no ascendant. The ascendant never touches the score in any case; it only feeds the element tally that picks which of four advice lists you are shown. Birth time and time zones covers what else a missing time costs.
• Friendship mode reweights and relabels the same four dimensions. Chemistry becomes fun and spark and drops in weight, communication and reliability rise, and staying power reads as reliability rather than as a forecast about romance.

One asymmetry worth naming: because a contact's weight is looked up under a sorted key, your Sun to their Moon and your Moon to their Sun get the same weight and the same sentence, even though a traditional reading would treat the two directions differently.

Neither tool sends your birth details anywhere. Both charts are computed in the page, and no birth detail is transmitted or stored on our side. Two things about your visit are: the page sends a first-party counter ping on load and another when you press the button, and we keep a per-day total of each. Data, privacy and saved chart links has the full inventory; the paid report is the only path that sends a chart.