Free birth-chart reading on your first visitTrusted by readers across IndiaPrivate and secure: your details stay yoursTalk to a real astrologer, any day of the weekDaily guidance, delivered every morningFree birth-chart reading on your first visitTrusted by readers across IndiaPrivate and secure: your details stay yoursTalk to a real astrologer, any day of the weekDaily guidance, delivered every morning
Today's panchang for your city →

How We Calculate

Every chart, every prediction, every claim: sourced and verifiable.

Updated · Applies to India

1. Calculation source

All planetary positions are computed by the Swiss Ephemeris library, the same engine used by professional Vedic and Western astrology software worldwide.

  • Ayanamsa: Lahiri 1940 (SE_SIDM_LAHIRI_1940), the standard adopted by the Indian Government for the National Panchang.
  • Lunar node: Mean Node (SE_MEAN_NODE), matching traditional Vedic practice and JHora defaults.
  • Timezone: IANA tz-database via luxon + tz-lookup, which converts birth lat/lng to the historically-correct UTC offset at the exact birth date.

swisseph v2.10.x luxon v3.x tz-lookup v3.x Node.js 20 LTS

2. Dasha system

We implement three classical Dasha systems drawn from primary Sanskrit sources:

  • Vimshottari (default): 120-year cycle, the most widely used planetary period system across India.
  • KP (Krishnamurti Paddhati): sub-lord refinement for precise event timing using cuspal stellar theory.
  • Jaimini Chara Dasha: sign-based cycle from the Jaimini tradition, cross-validated with karmic analysis.

Antardasha (sub-period) and Pratyantar-dasha (sub-sub-period) calculations follow the same classical proportional division rules cited in these texts.

3. Panchang method

Two Panchang calculation modes are supported:

  • Drik (default): based on observed (Spashta) planetary positions from Swiss Ephemeris. Used for modern Kundli software and the Rashtriya Panchang.
  • Vakya: traditional mean-motion tables used by South Indian panchangams. Available on request; accuracy varies by region.

Sunrise / sunset times are computed from Swiss Ephemeris with atmospheric refraction correction (standard 34′ model), latitude, longitude, and altitude inputs.

The five Panchang elements (Tithi, Vara, Nakshatra, Yoga, Karana) follow Drik calculation with Lahiri 1940 ayanamsa.

4. AI safety boundaries

What we will do:

  • Provide timing windows for events (not single dates).
  • Describe planetary tendencies and karmic patterns.
  • Surface remedies from classical sources: mantra, gemstone, charity.
  • Name chart strengths and growth areas with equal emphasis.

What we will never do:

  • Predict death, accidents, or serious illness. No classical Marana Karaka Sthana interpretations surface to users.
  • Use fear-based language ("bad period", "doomed", "cursed").
  • Make caste, dowry, sect, or gender-role prescriptions.
  • Give medical, legal, or financial advice under the guise of astrology.
  • Claim certainty. Every interpretation is probabilistic guidance, not fate.

5. Privacy promise

  • India data residency: all user data stored on servers located in India (Azure India Central).
  • DPDP Act compliance: aligned with the Digital Personal Data Protection Act, 2023 (India). Consent collected at onboarding; purpose-limited processing only.
  • Encryption at rest: PII (name, phone, email, birth details) encrypted with AES-256-GCM. Keys managed via Azure Key Vault.
  • Right to erasure: submit a deletion request from Settings → Account. Data pseudonymized within 14 days; permanently erased within 30 days.
  • No data selling: birth chart data is never sold or shared with third-party advertisers.

Full policy: Privacy Policy

6. System limits

  • Birth-time accuracy: Best results require birth time accurate to ±1 minute. Ascendant changes roughly every 2 hours; a 4-minute error can shift the lagna.
  • Location precision: Latitude and longitude resolved to 0.0001° (≈11 m). City-centre coordinates used when exact hospital address is unknown.
  • Ephemeris drift: Planetary positions are within <1 arc-minute vs JHora 8.0 reference output for dates between 3000 BCE and 3000 CE.
  • Divisional charts: D-1 through D-60 computed from the same Swiss Ephemeris position. D-60 is highly sensitive to birth-time accuracy; treat with appropriate caution.
  • Historical timezone changes: Pre-1900 birth data uses LMT (Local Mean Time) where tz-database records are unavailable.

7. Escalation criteria

Certain questions are beyond the scope of AI-generated interpretation and warrant a verified human astrologer. Current escalation triggers include:

  • User flags a prediction as harmful or distressing.
  • Chart contains multiple simultaneous difficult yogas (e.g., Kemadruma + Shakata + malefic 6/8/12 stellium).
  • Prashna (horary) questions about medical or legal matters.
  • User explicitly requests a human review.

Verified astrologer review: live consultation booking is coming in Track 7 (estimated Q2 launch). Until then, complex queries receive a clear disclaimer and a prompt to contact our support team.

8. Source code commitments

  • Open-source modules: Swiss Ephemeris Node.js bindings, Panchang calculation utilities, and Shadbala strength engine published on GitHub under MIT licence.
  • AI prompt versioning: Every system prompt used for chart interpretation is version-controlled in the same repository alongside the application code.
  • Weekly audit log: Internal audit log of prediction samples reviewed for tone, accuracy, and safety compliance; summary published monthly.
  • Knowledge base provenance: All 5,469 rules in the interpretation corpus are tagged with their classical source. Source metadata is accessible to our team for audit.
Public accuracy scores →

Coming soon: public Brier accuracy scores for our prediction engine, updated weekly.

Questions about any of this go to support@cosmicgeomancy.com. We reply within 48 hours.

Refunds · Cancellations · Shipping and delivery · Terms of Service · Privacy Policy