What paircast actually computes

1 · From birth date to four pillars

Your birth date + time + city goes through three steps:

1. True solar time correction — if you supplied a birth city, we adjust your clock time by (longitude − standard meridian) × 4 minutes. Singapore at 103.85°E vs the GMT+8 standard meridian (120°E) means ~−65 minutes. A child born at clock-time 12:30 in Singapore was at true solar time ~11:25, which can land in a different 时辰 (2-hour pillar) entirely.
2. Solar → Lunar conversion — via a peer-reviewed astronomical calendar engine that encodes the official 节气 (solar term) boundaries used by 协纪辨方书.
3. Pillar resolution — year, month, day, and hour pillars are read from the lunar calendar. Each pillar is a stem + branch pair (天干 + 地支). The 60-jiazi cycle is deterministic; given the same corrected datetime, every BaZi tool produces the same four pillars.

2 · The ten stems and twelve branches

Every chart is built from these 22 symbols. The element + polarity for each is what drives all downstream computation.

Heavenly stems (天干)

Earthly branches (地支)

3 · Day Master, strength, and the Use God

The "Day Master" (日主) — the stem of your day pillar — represents you. Three things determine how the rest of the chart relates to it:

• Strength — how strong the day-master's element is in the chart, weighted by pillar position. Month branch carries the most weight (×1.6 for branch element, ×1.0 for stem). Categories: 旺, 偏旺, 中和, 偏弱, 弱.
• Use God (用神) — the element that helps your day master most. Weak DM → use the element that generates you. Strong DM → use the element that controls you, or the one your DM controls. This single value drives most compatibility scoring.
• Favourable God (喜神) — a secondary helper element. Carries half the weight of the use god in compatibility scoring.

4 · Five-Element relations

The relations between the five elements are deterministic. They drive shi-shen (Ten Gods), compatibility scoring, and the year-by-year outlook.

5 · The Ten Gods (十神)

A stem's "Ten God" label describes its relationship to your day master, derived from element relation + polarity. Five relations × 2 polarities = 10 labels:

6 · Chart pattern (命格)

We classify every chart into one of eleven patterns by looking at the month branch's main qi (本气) and its Ten God relation to your day master:

For earth-branch (辰/戌/丑/未) months — known as 杂气 (mixed-qi) months — we also check whether a secondary hidden stem is "transparent" (appears as a visible stem elsewhere in the chart); if so, the pattern upgrades to that stem's Ten God.

7 · The 21 Shen Sha we check

Shen Sha are auspicious / inauspicious stars whose positions are looked up from classical tables. paircast checks 21 of them — many apps stop at 5-6. Each has a one-line interpretation that the LLM uses when narrating the reading. Rules sourced from 三命通会 / 神峰通考 / 协纪辨方书; where sources disagree we picked the most-cited modern form.

Auspicious nobles

Relationship / honour stars

Cautionary stars

8 · True solar time correction

Time zones are political; the sun is not. Cities far from their timezone's standard meridian see a meaningful gap between clock time and solar time. paircast applies the flat-longitude correction used by 子平 practitioners:

offsetMinutes = (longitude − standardMeridian) × 4

24 cities are pre-loaded; the table below covers the ones most relevant to the Chinese diaspora:

DST handling is deliberately out of scope for v1 — the SG/MY/HK/CN/TW target market doesn't observe DST. US/UK/AU users who observed DST at the time of birth would need to adjust their clock-time entry manually.

9 · How the compatibility score is computed

The 0–100 compatibility score is the sum of six independent components, each scored against its own maximum. The LLM never decides the score — it only narrates the engine's output.

10 · The 5-year couple outlook

Each year in the 5-year heatmap gets a 0-100 score, computed independently for each partner against the year's 流年 (year ganzhi), then averaged. Per-partner deltas added to a base of 50:

Final per-year score is clamped to [5, 95] so extremes feel earned and pure 0/100 never appears. The four per-quarter sub-scores apply small modifiers based on the seasonal branches (寅 spring, 巳 summer, 申 autumn, 亥 winter).

11 · How the date picker scores a candidate day

The picker scores each candidate day against both partners' charts. Hard taboos (day-pillar clash, Yáng-gōng forbidden day) downgrade the band even if the numeric score is high. For wedding purpose specifically, weights are:

Other purposes (engagement, ROM, moving-in, business launch, conception) share the same scaffolding with adjusted weights — for example, business launch drops marriage-specific terms (不将日, 三娘煞) and emphasises the use-god match factor.

12 · The boundary between engine and LLM

We use a frontier reasoning model to write the 6,500-word reading. Here's a clear line for what it does and doesn't decide:

If the LLM started inventing rules — claiming, say, that one stem-pair combination implies a specific marriage outcome — that's a violation of the boundary. We test the prompt to keep it grounded; the engine numbers do the talking.

Verify it yourself

Drop in two birth dates and see the structured output this page describes — chart pillars, scores, the lot — before any AI prose. The free reading exposes every number we just listed.