Method

Two competent engines can produce different charts from the same birth data, because a dozen choices sit between the sky and a kundali and the tradition does not settle all of them. Most software hides those choices. Here they are.

What we pinned

Choice	Ours	Why it matters
Ayanamsa	`LAHIRI` (`SIDM_LAHIRI`)	Picking a different family moves every chart by up to 2° — more than half a pada.
Lunar node	`TRUE_NODE`	Mean and true node differ by up to 1.7°. Indian software usually prints true.
Houses	`WHOLE_SIGN`	The house is the sign. No cusps, no intercepted signs, no polar degeneracy.
Dasha year	365.2425 days	No classical basis. A declared choice — and it moves boundaries by weeks over decades.
Dasha arithmetic	`DAYS_AS_TIMEDELTA`	Calendar addition and day arithmetic disagree. Also undetermined by the texts.
Dasha start date	`LOCAL_BIRTH_MOMENT`	A 02:00 birth falls on the previous day in UTC. That would shift every boundary by a day, forever.
Sunrise	`UPPER_LIMB`, -0.8333°	Disc and refraction move sunrise by minutes, which flips the weekday for a dawn birth.
Vargas	`D1, D9, D10`	There is no single formula. D10 breaks if you count continuously — a real bug we fixed.

Derivation 1.2.0 · conventions 1.0.0. A change to any of these changes every chart, so the versions travel with your result.

What we decline, and why

Declined	Reason
Longevity and death timing	Absolute. No tier, no request, no exception.
Muhurta (electional timing)	Infinitely subdividable — there is always another acceptable window, so nothing can be wrong.
Prashna (horary)	No natal anchor, so nothing to check a claim against.
Gemstone prescription	No mechanism and no counterfactual. We do state when a stone is not indicated — that is checkable.
Numerology	Its input is a name you can change, and a spelling that varies by transliteration. Nothing about it is falsifiable.
Palmistry	Requires biometric image capture, for a system with no timing structure and nothing to verify against.
Predictions about minors	Chart data only. No profiling of children.

What we do not yet know

The honest part. This engine computes positions well and can prove it. What it cannot yet prove is that the classical rules built on those positions predict anything.

None of our rules has a measured predictive value. Doing that needs dated claims recorded before the fact and scored afterwards, and that archive does not exist yet. Until it does, we publish computation and classical attribution — and we mark which rules rest on a text nobody has checked, and which rest on no text at all.

Rules we compute but do not show you

How precisely we can point at a source — and why most citations are chapter-level

Every rule prints a source like BPHS 46.8, which looks like a verse. For 58 of them it is not: the same locus is given for several rules at once. BPHS 46.8 is the stated source for twenty-two separate dasha rules. A shared locus is a chapter-level pointer in a verse citation's clothing — you cannot check it, and neither can we.

	Count
Citations confirmed against the printed edition	34
Distinct loci cited	91
Loci pointing at exactly one rule	75
Rules sharing a locus with another rule	58
Rules attested as practice, with no text to check	38

The last row is not a backlog. Those rules state openly that no admissible text carries them; there is no citation to verify, and counting them as unverified would pad the number with work that cannot be done.

The 16 loci carrying more than one rule

Locus	Rules
`Phaladeepika 26.12`	10 — The Moon transiting the 1st house counted from the natal Moon; The Moon transiting the 2nd house counted from the natal Moon; The Moon transiting the 3rd house counted from the natal Moon; The Moon transiting the 4th house counted from the natal Moon; The Moon transiting the 5th house counted from the natal Moon; The Moon transiting the 7th house counted from the natal Moon; The Moon transiting the 8th house counted from the natal Moon; The Moon transiting the 10th house counted from the natal Moon; The Moon transiting the 11th house counted from the natal Moon; The Moon transiting the 12th house counted from the natal Moon
`BPHS 24`	8 — The 10th lord; The chart supports a career; The 1st lord; The chart supports personal strength and direction; The 4th lord; The chart supports a home of your own; The 7th lord; The chart supports fortune and travel
`BPHS 75.1-75.2`	5 — Bhadra yoga; Hamsa yoga; Malavya yoga; Ruchaka yoga; Sasa yoga
`Phaladeepika 20.22`	4 — The twelfth lord's antardasha; The sixth lord's antardasha; The eighth lord's antardasha; A vargottama mahadasha lord delivers at roughly double strength
`BPHS 34.13`	4 — The yogakaraka in a dusthana; The yogakaraka in a kendra; The yogakaraka in a trikona; The chart supports the yogakaraka's period
`Phaladeepika 26.22`	4 — Saturn transiting the 8th from the natal Moon; Saturn transiting the 4th from the natal Moon; Sade Sati phase 2 of 3; Sade Sati phase 3 of 3
`BPHS 48.2-48.4`	3 — The 5th lord's mahadasha; The sixth lord's mahadasha; The 1st lord's mahadasha
`Phaladeepika 6.57`	3 — The 12th lord in a dusthana; The 6th lord in a dusthana; The 8th lord in a dusthana
`BPHS 27.5`	3 — Guru in a kendra; Chandra in a kendra; Shani in a kendra
`Phaladeepika 6.14`	2 — The Moon in a kendra from the lagna, or in its own sign; Shakat
`BPHS 47.5-47.6`	2 — This period's lord is in its fall; This period's lord is exalted
`BPHS 3.49`	2 — A planet in its fall; A planet in its exaltation sign
`BPHS 3.55-3.58`	2 — A graha in an enemy's sign; A graha in a friend's sign
`BPHS 3.56-3.58`	2 — The lagna lord standing in a sign whose lord is its GREAT enemy; The lagna lord standing in a sign whose lord is its GREAT friend
`BPHS 3.51`	2 — A graha in moolatrikona; A graha in own sign
`Phaladeepika 26.18`	2 — Jupiter transiting over the Moon itself; Jupiter transiting the fourth

10 of the 181 rules in the corpus fail our own validation — a citation pointing at a chapter about something else, or a condition so common it cannot carry weight. They are computed, they are listed here, and they are never rendered into a reading.

Sources

Rules cite a pinned edition, chapter and verse. Where a rule is widely practised but no admissible text states it, it is marked practice tradition rather than given a citation it does not have.

Place data from GeoNames, CC-BY 4.0. Ephemeris: XALEN (Apache-2.0), validated against JPL DE440.