How Auspica works out a BaZi chart

How is a BaZi chart calculated?

In five steps, the same in the app and in the free chart on this site. The engine that does them runs on your phone or in your browser; nothing is looked up from a table of approximate dates. If the terms are new, What is BaZi? explains them in plain words.

1. The birth date, time and place become an exact moment, using the time zone that applied at the birthplace on that date.
2. The sun's position at that moment gives the year and month pillars.
3. The calendar date gives the day pillar, and the clock hour gives the hour pillar.
4. Each branch's hidden stems, the Ten Gods and the luck pillars follow from the four pillars by fixed rules.
5. Judgements that tradition argues about, such as how strong the Day Master is and which elements help it, use a named method with its reasons attached.

Does Auspica use clock time or true solar time?

Clock time, by default. The birth time is read as the clock showed it at the birthplace, including daylight saving, and the time zone comes from the international time zone database, which records historical changes. Singapore, for example, kept UTC+7:30 until the end of 1981, and China used daylight saving from 1986 to 1991; a chart for those years uses the offset that was in force.

True solar time is an option in the app's settings, off by default. It shifts the clock by the birthplace's longitude (four minutes for each degree) and by the equation of time, since the real sun runs up to about sixteen minutes ahead of or behind the clock over the year. It changes only the day and hour pillars; the year and month pillars always use the exact moment, whatever the clock said.

Where clocks were set forward and a recorded time never happened, the time is moved forward by the gap. Where clocks were set back and a time happened twice, the earlier one is used. In both cases the chart says so.

When do the year and month pillars change?

At solar terms, not on the 1st of the month or at Chinese New Year. The year pillar turns at the exact moment of 立春 Li Chun, when the sun reaches 315 degrees of ecliptic longitude: in 2027 that is 4 February 2027 at 09:46 UTC+8. A birth before that moment belongs to the old year, and to its zodiac animal. Why the year starts at Li Chun has the full explanation and the date for every year from 1930 to 2040.

The month pillar turns at each of the twelve 节 terms, thirty degrees apart, and its stem follows from the year's stem by the Five Tigers rule (五虎遁). The terms are computed from the sun's apparent position, to the minute:

How are the day and hour pillars found?

The day pillar follows the cycle of sixty stem and branch pairs, which has run unbroken for centuries: the engine counts days from a fixed point, so no table is needed. The hour pillar follows the twelve two-hour periods, starting with 子 at 23:00, and its stem comes from the day's stem by the Five Rats rule (五鼠遁). The stem of the day pillar is the Day Master; the Day Master lookup finds it from a birth date.

The hour from 23:00 to midnight is where schools differ. Auspica's default keeps that hour in the day that is ending, with the hour's stem taken from the next day (夜子时). The alternative starts the new day pillar at 23:00. Both are in the app's settings, and a chart for a birth in that hour names the day pillar the other convention would give.

The defaults together are the "Classic (MY/SG)" preset: clock time, the day changing at midnight, and luck start ages counted in calendar days (see luck pillars). They are the conventions of the calculators most people in Singapore and Malaysia compare against.

What if I don't know my birth time?

You get three pillars: year, month and day, with the hour marked unknown. The engine uses local noon to place the year and month, and takes the day pillar from the calendar date. If a month began on the birth day itself, the chart says the month pillar is uncertain and asks for the time. Luck pillar start ages are then approximate. When the time is known, a birth within an hour of a month change, or within ten minutes of an hour change, is flagged too, because a slightly different recorded time would change a pillar.

A worked example

Someone born on 10 February 1991 at noon in Singapore, a woman. Every value below is computed by the engine when this page is built.

Day Master

辛 Yin Metal, born in a 寅 wood month

Strength

Weak: support is 32% of the chart's weight

Deciding rule

Support (扶): a weak Day Master is helped by resource and companions

Favourable

金 Metal, 土 Earth

Unfavourable

木 Wood, 火 Fire, 水 Water

Luck pillars

Run forward, because the year stem 辛 is yin and the chart is a woman's. The first, 辛卯, starts at age 8 in 1999; then 壬辰 from 2009 and 癸巳 from 2019.

What are hidden stems and Ten Gods, and how are they found?

Each earthly branch holds one to three heavenly stems, its hidden stems (藏干): a main one and, for most branches, one or two lesser ones. Auspica uses the standard Ziping table, so 寅 holds 甲, 丙 and 戊, and 午 holds 丁 and 己.

A Ten God (十神) names how any stem relates to the Day Master: by element (the same, produced by it, controlled by it, controlling it, or producing it) and by whether the two share yin or yang polarity. Five relations times two polarities give the ten. Branches are read through their hidden stems.

The element balance counts every stem, plus each branch's hidden stems shared out as 1, or 0.7 and 0.3, or 0.6, 0.3 and 0.1. Those shares are a common convention, not a rule from the classics, and the strength method below weighs the chart its own way.

How are luck pillars worked out?

Luck pillars (大运) are ten-year periods that step on from the month pillar: forward through the sixty pairs for a man born in a yang year or a woman born in a yin year, backward otherwise. The age the first one starts is the distance from birth to the next 节 term (forward) or back to the last one (backward), with three days counting as one year.

By default the distance is counted in whole calendar days between the birth date and the date of the term, as printed almanacs count it. The app also offers the exact time to the term. The two usually agree to within a year; for someone born on the day of a term but before its moment, they can differ by about ten years, and the chart says so.

How is Day Master strength decided?

By weighing every stem and branch for or against the Day Master, following the classical texts. The method follows 子平真诠 with 徐乐吾's commentary (评注), 滴天髓 with 任铁樵's 阐微, and the seasonal tables of 穷通宝鉴; one rule about when a chart does not follow (有食神制煞则不从) is taken from 三命通会. No classic gives numbers, only rankings, so the weights were set to read the classics' own example charts as their authors did, rather than taken from a modern points table.

1. Weigh the chart. A branch counts for more than a stem (干多不如根重), and the month branch most of all; a stem with no root in any branch counts for half. A complete three-harmony or directional group adds weight to its element, and two clashing branches side by side both lose some.
2. Look for special structures first. When the Day Master combines with the stem beside it, the pair transforms (化) if the month branch holds the new element and that element carries enough of the chart. A chart overwhelmingly on the Day Master's side follows its strength (从旺, or 专旺 when a whole group of its element stands in the chart); one where the Day Master has no main root and next to no support follows what surrounds it (从财, 从杀, 从儿).
3. Otherwise, place it in a band from extremely strong to extremely weak. A weak Day Master is supported (扶) and a strong one restrained (抑). At the edges the rule turns, as 任铁樵 reads them: a chart too strong is drained rather than controlled, and one too weak to resist takes the element that controls it.
4. Check the climate (调候). In the extreme seasons, from the 穷通宝鉴 table: fire for a metal, water, wood or earth Day Master born in winter, and water for a wood, metal or earth one born in summer, while the chart lacks it.
5. Bridge a conflict (通关). When the Day Master's element and its controller, or its wealth, hold large and even shares of the chart and the element between them is present, that element leads.

The result is a band, the favourable and unfavourable elements in order, and the rule that decided them, which the app shows. It does not model the month's commanding stem by day (司令), since the pillars alone do not say how far into the month a birth fell, nor 病药, nor the full structures (格局) of 子平真诠. On some charts the classical authors themselves disagree, and a practitioner may read those differently.

Where does the almanac come from?

From the same calendar rules, worked out for each date. The day's pillar is the sixty-day cycle; the twelve day officers (建除十二神) follow from the day branch and the month branch; the yellow and black belt gods (黄道黑道), the twenty-eight mansions (二十八宿), the clash (冲) with its animal and 煞 direction, and the directions of the joy, wealth and fortune gods follow their traditional tables.

The almanac's month changes on the calendar date of each 节 term, so the term's own day already belongs to the new month, as in printed almanacs; dates are in UTC+8, the convention for Singapore, Malaysia, Hong Kong and Taiwan. What a day favours (宜) and avoids (忌), from a list of 141 entries, comes from a published table keyed by the almanac month and the day pillar, ported from an open-source Chinese calendar library. Auspica gives no medical advice, so entries about medical care are never shown.

Not every almanac item is computed yet: the daily lists of auspicious and baleful spirits, hourly 宜 and 忌, 彭祖百忌 (apart from the 丁 day rule for haircuts), 胎神 and the nine stars are left out rather than guessed.

How is a day rated against my chart?

By adding up reasons, each of which the app shows. Every day starts at 50, the middle. Each reason that applies adds or takes away points, and the total is shown as stars: a star for every 20 points, to the nearest half star. A day that clashes your day branch can score at most 40, your year branch at most 44, and both at most 30, so good reasons cannot hide a clash. A clash is a watch-out, not a verdict.

The example chart on 4 February 2027

The 甲寅 day, Li Chun 2027, read against the chart above:

The numbers rank days for one person; they are not a forecast. The reasons are there so you can weigh them yourself.

How are good days and haircut days chosen?

On the good-day pages, a day is listed for an activity when the almanac names one of that activity's entries under 宜, and does not name it under 忌 or mark the day 诸事不宜. That reading is the same for everyone, so a listed day can still clash your own chart; the app's Date Finder adds the people involved, and for a wedding or an engagement it rules out a day that clashes any of them.

Haircut days read two traditions together. The Tibetan lunar day, in the Phugpa calendar, follows a sutra chapter on cutting hair, which favours days 3, 4, 5, 8, 9, 10, 11, 13, 14, 15, 18, 19, 22, 23, 26 and 27 of the lunar month. The Chinese almanac objects when it lists 理发 under 忌, marks the day 诸事不宜, or the day's stem is 丁 (丁不剃头). A day neither objects to is Good, and Best if the almanac also lists 理发 under 宜. When the only objection is the Tibetan day or the 丁 day rule, the day is shown as your call, with the reason, rather than hidden. In the app, a day that clashes your chart is also one to avoid.

How is it checked?

By automated tests of several kinds, which run against the engine itself:

• Solar-term moments against times published by national observatories, to the minute.
• Thousands of random charts compared field by field with an independent open-source implementation of the Chinese calendar: pillars, hidden stems, Ten Gods, luck pillars and more.
• Reference charts from a widely used online calculator, which fix the default conventions: clock time, the midnight day change, exact solar-term times and luck ages by calendar days.
• Historical clock changes, such as Singapore's move from UTC+7:30 and China's daylight saving years, and the lunar calendar against an observatory's published tables.
• For Day Master strength, 89 example charts from 滴天髓阐微 and 子平真诠评注, each with its author's own verdict. When the method was set, it agreed with the author on strong or weak in 52 of the 59 charts where the author makes that call, found 21 of the 28 special structures, and named the author's useful element (用神) first in 55 charts and among the favourable elements in 67. The tests fail if it slips below that.

What Auspica does not do

It does not predict. Auspica treats BaZi as a tradition for reflection and planning: a chart describes tendencies and timing, and nothing in it is guaranteed. Clashes are watch-outs, not omens. Readings are cultural guidance, and the site says so on every page: Cultural guidance for reflection and entertainment. Not medical, legal or financial advice.

Where tradition disagrees, the app picks a default, says which, and lets you change it: the day boundary, clock or true solar time, and how the luck start age is counted. Where the engine computes nothing, the app shows nothing rather than a guess.

Questions about the method