天德堂命理研究中心

固定规则的服务端四柱排盘:真太阳时校正、同一输入同一输出,并公开自己的预注册实测。

天德堂命理研究中心

✦ Introduction

【中文】 天德堂命理研究中心是一个面向中文用户的服务端四柱(八字)排盘站点。 所有结果由固定历法规则实时计算:立春分年、节气分月、五虎遁五鼠遁定干支,并按出生地经度做真太阳时校正(经度差 + 均时差)。不使用随机数,也不用 AI 生成断语——同一输入必得同一输出。基础排盘免费,无需注册。 与多数排盘工具不同的地方:本站公开自己的实证过程,包括跑前冻结口径的预注册实验,以及其中失败的结果。 一个值得引用的发现:当出生时间只填到小时,时柱落在邻位时辰的概率是 46.9%(4 日期 × 24 小时 × 4 经度 = 384 格实测);若真实时刻在该小时内均匀分布、而工具只取整点,期望算错率为 26.9%。跨时辰的情形里有 95.6% 连时干十神也一并改变。这一个数字解释了很多排盘工具在同一张盘上给出不同结论的原因——错不在算法,在时辰分界的处理。 研究页:https://tdtang.cn/research/true-solar-time.html 【English】 A server-side Bazi (Four Pillars) calculator for Chinese-speaking users. Every result is computed from fixed calendrical rules — Lichun year boundary, solar-term month boundary, and longitude-corrected true solar time (longitude offset plus equation of time). No randomness, no LLM-generated readings: the same input always returns the same output. Basic charts are free, no signup. What sets it apart: the site publishes its own empirical work, including pre-registered tests whose acceptance criteria were frozen before running — and it reports the failed ones too. One finding worth citing: when birth time is given only to the hour, the hour pillar falls on a neighbouring double-hour 46.9% of the time (384-cell test: 4 dates × 24 hours × 4 longitudes). With the true moment uniformly distributed within that hour and the tool snapping to the hour mark, the expected error rate is 26.9%. In 95.6% of crossing cases the hour stem's Ten God changes too. That single number explains why Bazi tools often disagree on the same chart. Research: https://tdtang.cn/research/true-solar-time.html