How our panchang is calculated

The foundation of any panchang is knowing exactly where the Sun and the Moon are in the sky at a given moment. We compute these positions with astronomy-engine, an open-source astronomy library. For the Sun and planets it uses the VSOP87 planetary theory, and for the Moon a lunar theory following Montenbruck and Pfleger. The results are accurate to well under an arcminute for the Sun and to about an arcminute for the Moon, far finer than a panchang needs to fix times to the minute.

Everything is calculated afresh on our own server. We do not copy tables from other almanacs or websites, so every figure you see comes from the same method described on this page.

Astronomy libraries measure positions from the spring equinox point; these are called tropical positions. The Indian panchang measures positions against the fixed stars instead; these are sidereal, or nirayana, positions. Because the Earth's axis wobbles very slowly, like a spinning top (this is called precession), the equinox point drifts backwards against the stars, and the gap between the two systems keeps growing. That gap is the ayanamsha.

We use the Lahiri ayanamsha, also known as Chitrapaksha. It is the standard adopted by the Government of India's Calendar Reform Committee and is used by the Indian Astronomical Ephemeris. At present it is about 24 degrees 13 minutes, and it grows by roughly 50 arcseconds each year. To get the Indian position of the Sun or Moon, we subtract the ayanamsha from its tropical position.

A panchang day runs from one sunrise to the next, not from midnight. The values shown for a day are the ones in force at sunrise, together with the time each one ends. The five limbs of the panchang are worked out like this:

To find when a tithi, nakshatra, yoga or karana ends, our software steps forward in time until the angle crosses its next boundary, and narrows that moment down to the second.

Sunrise is the moment the upper edge of the Sun touches the horizon, allowing for normal atmospheric refraction. The air bends sunlight slightly, so the Sun becomes visible a little before it is geometrically above the horizon. Sunset is the same moment in reverse.

For moonrise and moonset we use the centre of the Moon's disc at the horizon without refraction, the convention most Indian panchangs follow. Times are shown to the minute: sunrise-based windows are rounded to the nearest minute, and the moment a tithi or nakshatra ends is shown with its seconds dropped.

Rahu kaal, yamaganda and gulika kaal all depend on the length of daylight, from sunrise to sunset. Daylight is split into eight equal parts, and the weekday decides which part each period takes. For rahu kaal the part is:

Choghadiya splits daylight and night into eight parts each. The names are Amrit, Shubh and Labh, which are considered good; Char, which is neutral; and Rog, Kaal and Udveg, which are best avoided. The weekday sets the name each sequence starts with.

Abhijit muhurat is the 8th of 15 equal parts of daylight, falling around local noon. By tradition it is not used on Wednesday. Because all of these periods depend on your local sunrise and sunset, they are never identical in two different cities, even on the same day.

Month names follow the purnimanta system used in North India, in which each month ends on the full moon. When a lunar month passes without the Sun entering a new sign (a sankranti), that month is counted as an adhik, or extra, month. The Vikram Samvat year changes on Chaitra Shukla Pratipada.

Each festival has a rule in our data file, and every date on the site says which rule produced it. By default the tithi present at sunrise (udaya tithi) decides the day. If a tithi touches no sunrise (a kshaya tithi), it is kept on the day it begins. Festivals with their own rule use the relevant part of the day: madhyahna (Ram Navami, Ganesh Chaturthi), aparahna (Dussehra, Bhai Dooj), pradosh kaal (Dhanteras, Diwali, pradosh vrat), nishita (Maha Shivratri), moonrise (Karva Chauth, Sankashti), and the three-muhurta rule for Akshaya Tritiya.

Ekadashi follows the Smarta rules: a kshaya ekadashi is kept on the day it begins, an ekadashi spanning two sunrises on the second day, and the fast moves a day earlier if dwadashi would not reach the next sunrise (so that parana can fall within dwadashi). These reproduce all 49 Smarta dates drikpanchang.com publishes for New Delhi in 2026 and 2027; they do not model every Mahadwadashi case of the texts. Our Vaishnava date is an estimate from the arunodaya rule only.

Holika Dahan follows the seven rules drikpanchang.com publishes (Pradosh with Purnima and no Bhadra first; then after Bhadra if it ends before midnight; then Bhadra Punchha; never Bhadra Mukha; otherwise Pradosh). Where conventions genuinely differ we show both dates: Janmashtami (Ashtami at sunrise, our main date, and Ashtami at nishita) and Rangwali Holi (Chaitra Krishna Pratipada at sunrise, our main date, and the day after Holika Dahan).

Puja timings are never a bare day part. Each window is the overlap of the required tithi with the permitted day part, with exclusions such as Bhadra, Chitra nakshatra or Vaidhriti yoga removed, and a lagna requirement where one applies (Vrishabha for Lakshmi puja). A window can be empty or come in two pieces, and the page says so.

Our national reference city is Delhi (28.6139 N, 77.2090 E). Sunrise, however, moves by about 4 minutes for every degree of longitude. Kolkata sees sunrise roughly 50 minutes before Ahmedabad, even though both cities keep Indian Standard Time. Latitude matters too, because it changes the length of the day through the year.

So rahu kaal, choghadiya and other muhurat times should always be read for your own city. Tithi and nakshatra work differently: they end at the same instant everywhere in the world, but the clock time of that instant depends on your time zone. That is why a festival can fall on a different date in the USA or UK than in India. For cities abroad, our times include daylight saving automatically.

Two different things matter here. Numerical precision: our searches find each transition (a tithi ending, a lagna changing) to within one second of what the model predicts. Accuracy: how close the model is to the real sky and to other almanacs. One-second precision does not mean one-second accuracy.

Measured accuracy: in our regression test against drikpanchang.com for five dates (Delhi, New York, Sydney on the solstice, New York on its daylight-saving change, Delhi in late 2027), every sunrise, sunset, tithi, nakshatra, rahu kaal and abhijit time was within 2 minutes; the largest difference was 1.6 minutes, on a tithi end. Festival windows we checked (Akshaya Tritiya, Ganesh Chaturthi, Lakshmi puja, Raksha Bandhan, Maha Shivratri, Janmashtami, Holika Dahan, Govardhan, Bhai Dooj) agree within about a minute. The moment of Makar Sankranti comes out 4 to 6 minutes earlier than drikpanchang's, probably because of a small difference in how the ayanamsha is defined; this is unresolved.

Not yet checked against an external reference: ekadashi parana windows, Dhanteras and Ghatasthapana windows, the Vaishnava ekadashi estimate, and the four muhurat lists (which are checked only by our own independent minute-by-minute script).

Rounding: windows built from sunrise and sunset are rounded to the nearest minute; the time a tithi, nakshatra, yoga or karana ends is shown with its seconds dropped. All calculations keep full precision internally. Pages refresh themselves every 30 minutes.