BaZi Accuracy · Test Cases
BaZi Calculator Accuracy: The Boundary Tests That Matter
A calculator is not accurate because one familiar chart looks right. Accuracy is established by testing the places where a small input or rule change can produce a different output.
- Published
- 2026-08-30
- Reading
- 9 min
- Primary question
- bazi calculator accuracy test
Evidence path
Calculation before interpretation.
Birth record
Time + place
Engine output
Bounded reading
At a glance
Three facts to keep.
- 01Boundary cases reveal rule errors that ordinary dates can hide.
- 02Golden tests should assert calculated outputs, not interpretive prose.
- 03Unknown-time mode needs omission tests as well as normal calculations.
01
What an accuracy test should prove
A useful test states the input, convention, expected calculated field, and tolerance. It should be repeatable without asking a model to judge whether an interpretation sounds plausible.
AstroZen separates deterministic chart tests from report-writing tests. Four Pillars, adjusted time, planetary longitude, houses, and aspects belong to the deterministic layer. Narrative quality is evaluated later and cannot repair a wrong chart.
02
The minimum BaZi boundary suite
The highest-value test set covers geographic extremes, civil-time transitions, solar-term boundaries, day rollover, branch rollover, and missing data. A calculator that only tests midday dates in one country leaves its riskiest behavior unexamined.
| Boundary | Failure it can expose |
|---|---|
| Zone meridian distance | Hard-coded or wrong solar correction |
| DST start/end | Wrong UTC instant or ambiguous local time |
| 23:00 / 00:00 | Undisclosed day-boundary convention |
| Two-hour branch edge | Wrong Hour Pillar assignment |
| Li Chun / monthly Jie | Lunar-date substitution for solar-term rules |
| Unknown time | Invented Hour Pillar or false precision |
03
Published AstroZen cases
AstroZen publishes a small machine-readable case set rather than claiming universal agreement. It includes a large Madrid solar-time correction, a small Shanghai correction, exact-time Western chart assertions, unknown-time omissions, and daily transit-window checks.
- Madrid: 1992-06-15 08:30, longitude −3.70256°, +02:00 meridian → 06:15 true solar time.
- Shanghai: 1990-01-15 09:00, longitude 121.47°, UTC+8 meridian → a small correction near 08:57.
- Unknown time: no Hour Pillar, Ascendant, Midheaven, houses, or time-sensitive Moon claims.
- Western exact-time fixture: ten planetary positions, twelve cusps, angles, and a nonempty aspect list.
Method note: The JSON file is a public inspection aid, not a claim that symbolic astrology has scientific predictive validity.
04
How to run your own comparison
Choose five ordinary charts and at least five boundary charts. Record each calculator's input normalization and conventions before comparing pillars. If a product hides coordinates, offset, adjusted time, and method notes, mark the result as unreviewable rather than automatically wrong.
- Use the same place granularity and birth-time precision.
- Capture software version and test date.
- Separate calculation differences from translation differences.
- Keep interpretation out of the pass/fail assertion.
Evidence register
Sources and method.
- 01AstroZen Accuracy Lab
Published calculation rules, boundary cases, and machine-readable test data.
- 02lunar-javascript
The open-source calendar library used for Four Pillars and solar-term calendar operations.
- 03IANA Time Zone Database
Historical civil-time and daylight-saving transitions used by modern time-zone software.
Questions people ask
FAQ
Can two accurate BaZi calculators disagree?
Yes, when they intentionally use different disclosed conventions. Accuracy testing should distinguish an implementation error from a school-level rule choice.
Is matching an almanac enough to validate a calculator?
No. An almanac comparison can be one fixture, but it may use its own time and boundary conventions. A robust suite tests named rules and edge cases directly.
Where can I download AstroZen's test cases?
The Accuracy Lab links to a public JSON file containing sanitized calculation fixtures with no customer birth records.
Continue the cluster
Related evidence.
Inspect your own calculation
Use the free chart before deciding whether a report is useful.
No email or payment is required for the preview. Calculated facts and omitted factors remain visible.
Calculate my free chartEducational content for personal reflection and entertainment. Astrology and BaZi are traditional symbolic systems, not scientifically validated diagnostic or predictive methods.