How These Numbers Are Computed

No third-party astrology API

There is no external calculation service behind this site, no licensed ephemeris data files, and no imported panchang tables. Planetary positions are computed from orbital elements, sunrise from spherical astronomy, and every panchang element and muhurat window from those two results by the classical rules. That is a deliberate trade: it means the accuracy is ours to state honestly, which is what the rest of this page does.

1. Planetary positions

Geocentric tropical ecliptic longitudes are computed for the Sun, the Moon, the five visible planets, and the lunar nodes (Rahu and Ketu) using Paul Schlyter's How to compute planetary positions — osculating Keplerian orbital elements plus the short perturbation series for Jupiter, Saturn and the Moon. Julian Day conversions follow Meeus, Astronomical Algorithms, chapter 7, and every input is resolved to UTC before it reaches the astronomy.

That method is accurate to roughly one arcminute for the planets and a few arcminutes for the Moon. We chose it over a full VSOP87/ELP2000 implementation because it stays a few hundred lines of plain, checkable code with no external data dependency — and because it is far inside what a reading actually needs. The smallest division this site ever reports is a nakshatra pada, 3°20′, which is 200 arcminutes. A one-to-three arcminute uncertainty is around one percent of it.

These are true computed positions — a drik system — not the mean-position arithmetic of the older Surya Siddhanta tradition. Where the two disagree, this site follows the sky.

2. The sidereal frame — Lahiri ayanamsa

Vedic astrology reads the sidereal zodiac, so every tropical longitude above is shifted by the ayanamsa: sidereal = tropical − ayanamsa. We use Lahiri (Chitrapaksha), the standard set by the Indian government's Calendar Reform Committee and the one used by most published panchangs and Vedic software.

It is implemented as a linear model anchored to published true-Lahiri reference values — 22°27′55″ at 1 January 1900, advancing 0.0139289° per year. That anchor and rate reproduce the published intermediate values for 1950, 1956, 2000 and 2020 to within about 20 arcseconds.

3. Sunrise, sunset and solar noon

Almost every muhurat on this site is derived from sunrise and sunset, so these are computed with exact spherical astronomy rather than the day-of-year approximations common in quick sunrise formulas. The Sun's longitude comes from the same ephemeris as the birth chart; declination and right ascension follow from the exact ecliptic-to- equatorial relations; Greenwich Mean Sidereal Time gives the local hour angle.

Solar noon is found by fixed-point iteration — at true solar noon the local sidereal time equals the Sun's right ascension — converging to sub-second precision in two or three passes. Sunrise and sunset then follow from the hour angle at an apparent solar altitude of −0°50′: 34 arcminutes of standard atmospheric refraction plus the Sun's 16-arcminute mean semi-diameter, the conventional "upper limb touching the horizon" definition.

Above the polar circles there are days with no sunrise or sunset at all. On those days every daylight-derived value — Rahu Kaal, Choghadiya, Hora, Abhijit — is reported as unavailable rather than invented. Tithi and nakshatra do not depend on daylight, so they are still shown.

4. Time zones and daylight saving

Every city carries its own IANA time zone, and every date and time is resolved in that zone. Nothing defaults to IST. A city's day boundary is its own, which is why the same festival can legitimately fall on different calendar dates in Auckland and Toronto, and why each city page states its own date.

Daylight saving is handled by the zone database rather than by a fixed offset, including the southern hemisphere's reversed transitions. Our test suite checks sunrise across six DST changeovers — London, New York and Sydney, in both directions — because an hour's error there is an hour's error in every window derived from it.

5. The five limbs of the Panchang

"Panchang" means five limbs. Four of them are angular divisions of the sidereal longitudes computed above; the fifth is the weekday.

A synodic month holds 30 tithis, split into the waxing Shukla and waning Krishna fortnights, and 60 karanas — one fixed Kimstughna at the start, seven movable karanas cycling eight times through the middle, then three fixed karanas at the end. That is the standard classical scheme, not a simplification of it.

6. Rahu Kaal, Yamagandam, Gulika and Abhijit

The daylight period — actual sunrise to actual sunset for that city on that date — is divided into eight equal parts. Rahu Kalam, Yamagandam and Gulika Kalam are each one of those eighths, with the segment fixed by weekday according to the standard tables.

This is why our times differ from sites that publish a fixed clock window. An eighth of daylight is only 90 minutes when the day is exactly twelve hours long. In Rotterdam in June it is well over an hour and a half; in December it is under an hour. The window moves with the city and the season because it is defined that way.

Abhijit Muhurat is centred on true solar noon, extending one-thirtieth of the daylight length either side of it.

7. Choghadiya and Hora

Both are planetary-hour systems built on the same shape: a sequence of ruling planets chosen by weekday, sliced evenly across the day and then across the night. Choghadiya gives eight periods each; Hora gives twelve.

Crucially, the daytime span (sunrise to sunset) and the night span (sunset to the following sunrise) are divided separately. Outside the equinoxes those two spans are not equal, so a night Choghadiya is not the same length as a day Choghadiya — which is exactly why dividing 24 hours into 16 equal slices, as some calculators do, gives the wrong answer everywhere except twice a year.

The names and qualities follow the standard table: Amrit, Shubh and Labh auspicious; Chal neutral; Udveg, Rog and Kaal inauspicious.

8. Festival dates — the vyapini rule

A festival is not simply "the day the tithi starts". Each one is defined by which tithi it falls on and which time of day classically governs that choice — sunrise for most vratas, madhyahna (midday) for Ganesh Chaturthi, aparahna (afternoon) for Vijayadashami, pradosh (evening) for Diwali, nishita (midnight) for Janmashtami and Shivaratri. The tithi that prevails at that moment decides the date. This is the classical vyapini rule.

Applying one uniform "tithi at sunrise" test — perfectly adequate for a daily panchang — gets several of these dates wrong. Bhai Dooj 2026 is the clean example: Dwitiya begins at 13:58 on 10 November, so a test taken mid-afternoon claims the 10th, while the rule actually requires Dwitiya to prevail from the start of the Aparahna window, putting it on the 11th — where published panchangs place it.

Makar Sankranti is the exception: it is purely solar, computed as the moment the Sun crosses a sidereal longitude, with no tithi involved.

9. Muhurat windows

Once a festival's date is settled for a city, its window is derived from that city's own sunrise and sunset — never from India's.

• Pradosh Kaal (Diwali Lakshmi Puja, Dhanteras) — from sunset, running 144 minutes.
• Aparahna (Bhai Dooj) — the fourth of five equal divisions of the daylight period.
• Pratahkal — the first of those same five divisions.
• Sandhya Arghya (Chhath) — the sunset instant itself, with Usha Arghya at the following sunrise.

Where a computed window overlaps that day's Rahu Kalam, the page says so rather than quietly publishing both.

10. What we verify

The engine is pinned by an automated test suite that compares it against externally published values and against physical invariants that must hold regardless of tradition. It runs on every change. Among the checks:

• Lahiri ayanamsa against published tables for 2000 and 2026.
• All four 2026 Sankranti dates — 14 Jan, 14 Apr, 17 Aug, 17 Oct — which test solar longitude and ayanamsa together, and so are the sharpest single check on the sidereal frame.
• Thirteen 2026 festival dates against published panchangs, including Holika Dahan and Holi as the separate days they are.
• Rahu and Ketu exactly 180° apart, to within 0.01°.
• Saturn's longitude genuinely moving backwards during its 2026 retrograde.
• The ascendant equalling the Sun's longitude at sunrise.
• Vimshottari dasha periods totalling exactly 120 years in the classical order.
• Sunrise across six daylight-saving transitions in three hemisphere-spanning cities.
• Polar latitudes degrading to "unavailable" instead of throwing or inventing a window.

11. Known simplifications

No calculation is exact, and a methodology page that claims otherwise is not worth reading. Here is where ours is approximate:

• Positions carry about an arcminute of uncertainty (a few for the Moon). The effect you would notice is on boundary times: the exact minute a tithi or nakshatra changes over can differ by a few minutes from a full-precision ephemeris. The tithi or nakshatra itself is not in doubt except within those few minutes of a changeover.
• No ΔT correction. Positions are computed from Universal Time rather than Terrestrial Time. In 2026 the difference is around 70 seconds, which moves the Moon roughly 40 arcseconds.
• Nutation is not modelled separately. It is absorbed into the published reference values the Lahiri model is calibrated against.
• Positions are geocentric, which is the standard convention for panchang work — no topocentric correction is applied for the observer's position on the Earth's surface.
• The modern era is what this is tuned for. Both the orbital elements and the linear ayanamsa model are calibrated for roughly 1900–2100. Accuracy degrades outside that range.

Found something wrong?

If one of our figures disagrees with a panchang you trust, we would genuinely like to know — tell us the date, the city and the value you expected. Disagreements usually come down to a documented difference in tradition, but not always, and the ones that do not are worth fixing.

See it in use