METHODOLOGY & VALIDATION

Every rule that ships here gets checked against something first.

This page is the engineering record behind the reading, made public: how the placement engine is checked against real data, how we cite our sources, and — just as important — exactly what's still incomplete. Not marketing copy. Written for the people who want to look under the hood.

~5 MIN READ · IN A HURRY? JUMP TO THE 60-SECOND VERSION ↓

THE 60-SECOND VERSION

If you read nothing else on this page

1

Every placement rule is checked against real charts before it ships. 529 automated checks pass right now, including matches against real, independently documented birth charts — Winston Churchill's among them.

2

We catch our own mistakes, on the record. One real example: a third-party placement library disagreed with our own validated data on Tian Kui/Tian Yue for one birth-year group. We found it, corrected it, and all 110 related checks now pass.

3

We cite by classical school, not by book title. Most rules here are corroborated across more than one independent classical source — citing the tradition is the more honest, more checkable claim than pointing at a single title.

4

Where classical schools genuinely disagree, we label our choice as a choice. Bei Pai vs. San He palace convention, for instance — both are implemented in code, and neither is silently defaulted on you.

5

What's incomplete, we say plainly. The star-magnitude table isn't built as structured data yet; 6 of 9 month-based auxiliary stars are deliberately left out; 1 of 144 wealth-interpretation profiles is flagged as a lower-confidence read.

HOW WE VALIDATE A RULE

A rule earns its place by matching a real life first

This engine didn't ship because a formula recalled from memory felt right. It shipped because it matches four real, independently documented birth charts exactly — including Winston Churchill's (30 Nov 1874, 1:30am, male) and Mao Zedong's — checked field by field: Life Palace, Body Palace, the Five-Element Bureau, all fourteen Major Star placements, and all four Transformation stars for every relevant birth-year stem. Any mismatch is treated as a real engineering bug: located, fixed, and the full suite re-run before it's called resolved.

108 of our first 110 checks passed on the very first run. The other 2 were a real bug — now fixed, and holding at 110 of 110.

Those 2 failures were the third-party placement library we depend on (iztro) computing Tian Kui and Tian Yue differently from our own already-validated data, specifically for natives of the 辛 (Xin) birth-year stem. We didn't treat it as noise — we wrote a targeted correction to bring the result in line with the validated data. It's the first real example in "What we've caught and fixed" below.

THE NUMBERS, RIGHT NOW

These are run, not estimated

529
AUTOMATED CHECKS, ALL PASSING

Across four regression suites — a mix of structural consistency checks, matches against real documented charts, and a small number of synthetic charts each built to isolate one code branch (always labeled as synthetic in the code itself).

110/110
AUXILIARY-STAR PLACEMENT CHECKS

Zuo Fu, You Bi, Tian Xing, Wen Chang, Wen Qu, Di Jie, Tian Ma, Hong Luan, Tian Xi, Huo Xing, Ling Xing, Tian Yao, Yin Sha, Tian Wu, Jie Shen, Tian Kui and Tian Yue — checked across all ten Heavenly Stems.

144/144
WEALTH-INTERPRETATION PROFILES

Every Zi Wei ring-position × Life Palace branch combination is populated. 1 of the 144 is explicitly flagged internally as a lower-confidence paraphrase, sourced from a hard-to-read scan.

2/2
INDEPENDENTLY-SOURCED SI HUA TABLES

Two independently compiled Four Transformations tables matched byte-for-byte across all ten stems — the strongest evidence this project has that the table is a shared classical standard, not one lineage's private formula.

These numbers describe what the code does when it's actually run, not what a document says about it — docs go stale; a passing test suite doesn't.Verified against the live codebase as of Sep 6, 2026.

HOW WE CITE OUR SOURCES

Cited to the tradition, not to a title

Zi Wei Dou Shu is over a thousand years old and has passed through countless practitioners' hands. When we say a rule is "sourced," we mean it's corroborated across more than one independent classical body of theory or school — the two Si Hua tables matching byte-for-byte, mentioned above, is one instance of exactly that kind of independent corroboration. So publicly, we cite only at the level of school or tradition (Wu Xing theory, the San He school's palace convention, a named classical Ge Ju formation) — never a specific modern reference title or author. That's not vagueness for its own sake: pinning everything to three modern reference texts would overstate how dependent this system is on them, and would obscure the more accurate fact — that this is a thousand-year consensus, not one author's private formula. Internally, our engineering docs cite every rule down to the specific book and page, precisely so it can be checked and corrected later — that record just isn't the public-facing one.

HOW WE LABEL CONFIDENCE

Every sentence in a reading belongs to one of four tiers

1 · Structural fact

Computed directly from your birth data

Life Palace, star placements, the Four Transformations, the Five-Element Bureau. Stated flatly, no hedge — it isn't a claim, it's arithmetic.

2 · Classical tradition attribution

Traceable to a named school or body of theory

Wu Xing production/control cycles, the San He/Bei Pai palace conventions, a named classical formation — cited at the school level, for the reasons in the section above.

3 · Engine convention

A documented judgment call where sources genuinely disagree

Bei Pai vs. San He palace convention is the clearest example — flagged explicitly as a choice this project made, never dressed up as settled classical fact.

4 · AI narrative synthesis

The prose that connects tiers 1-3 into readable sentences

Labeled as synthesis, never as an independent source of truth. How confidently a sentence is phrased always tracks how settled the fact underneath it actually is.

WHAT WE'VE CAUGHT AND FIXED

When a rule turns out wrong, we fix it like a bug — on the record

Tian Kui / Tian Yue, 辛-year natives. The third-party placement library we depend on disagreed with our own validated data here. We wrote a targeted correction; all 110 auxiliary-star checks now pass.

Wen Qu's anchor point was wrong at first. Caught against a real chart, corrected before it shipped.

Two entries in the 144-profile interpretation table had the wrong index keys at first. Not caught by manual proofreading — caught by the engine's own generated placement table disagreeing with them. The data checked itself.

Bei Pai and San He palace conventions are both implemented in code. Neither is silently chosen for you and passed off as the one correct method.

WHAT WE WON'T CLAIM

What's still incomplete, stated plainly

None of the paid products we've looked at disclose their own limitations publicly. We're doing it because it's how this kind of product should work.

The star-magnitude table (Miao/Wang/De/Ping/Xian) isn't built as structured data yet. Magnitude gets mentioned in prose for a few placements already, but the full 14-star × 12-branch, 168-cell table is 0 cells populated as structured data today.

6 of 9 month-based auxiliary stars remain unimplemented. Tian Yao, Yue Ma, Jie Shen, Tian Wu, a separate month-keyed Tian Yue, and Yin Sha — their patterns in the source material aren't consistent enough to read with confidence, and we'd rather leave a placement out than guess.

1 of the 144 wealth-interpretation profiles is a lower-confidence paraphrase. Sourced from one hard-to-read scanned passage — worth a second look if a user ever surfaces and scrutinizes that specific one.

4 more named flying-star patterns in the same family as the ones we do compute are deliberately left out. The source material we found for them is either internally contradictory or gives no way to turn it into one precise, checkable rule — we'd rather skip a pattern than invent a rule for it.

Two genuine, unexplained anomalies remain in the Tian Kui/Tian Yue placement rule, affecting a handful of birth-year stem groups. Documented in our engineering notes, still being investigated, not swept under anything.

My Life Chart exists for entertainment and self-reflection: a rigorously computed mirror, built on sourced material and a validated placement engine — not a fortune-telling machine, and not something we'll oversell to make a sale.

That's how we work. Want to see it applied to your own chart?

Free to see your Life Palace. No account required.

See what's in your chart — free