Chart accuracy report

What this shows — and what it does not

Setting the Ju (working out Yin/Yang Dun and the Ju number from the time) is an independent test: the engine sees only the moment, not the book's answer. 296 match. 62 differ, and by our research all of them were set with the Zhi Run method (we use Chai Bu; near a solar-term boundary the two methods can differ by a Ju or switch Dun). Most of these calls rest on the method the author used in that period, and a few are inferred back from the book's Ju — not every book states its method outright. All 62 have a Zhi Run record or inference behind them, but we have not re-cast each one with the Zhi Run method to confirm it.

Cell by cell (Zhi Fu, nine stars, eight doors, eight deities, heaven and earth stems) is laid out on the book's own Dun and Ju, so it tests how the nine palaces are filled once the Ju is fixed. 14308 of 14308 cells match (100%).

Before a chart went in, it was checked against the engine cell by cell; charts that did not match and were judged to be errors in the book were left out — our records show 8 dropped from one batch of 147. Another 24 cases were not included: 16 because the time could not be recovered, 8 because they looked like other schools (Flying Plate, Yin Plate and so on). So "every cell matches" is the result after that screening; since then the charts serve as regression tests that fail the moment anyone changes a casting rule. This is not an independent third-party audit.

Where Tian Qin sits with Tian Rui in palace 2, some books draw one heaven stem and some draw two; we check whatever the book drew.

We do not publish the sources (titles, authors, pages) or the individual cases — this is our own research collection — and we never reproduce the books' readings or the people's stories. So this page gives totals only, with no row-by-row check: these numbers are our own self-audit, not proof you can re-check.

Breakdown