Unknown birth hour? How to narrow it with multi-version comparison
Forgot your birth hour? Don't paste a birthday into AI. Hour-narrowing logic first—parallel candidates, compare differences, narrow with facts—then concrete steps and what written chart data AI actually needs.
TL;DR: If you don't know your birth hour, two common dead ends are guessing a default hour or asking AI to “sense” it from your birthday. The steadier approach is hour narrowing: cast a chart under each plausible hour, compare structural differences, and cross-check against life facts you already know—until you're down to one or two candidates.
Below: that logic first, then the steps. If you plan to use ChatGPT or Claude to help compare versions, the end notes what kind of written chart data actually works.
What hour narrowing is trying to do
In chart-reading systems, birth hour often fixes the Life Palace, hour pillar, or ascendant—get it wrong and everything built on top is off. Plenty of people have the date but not the clock time; that's normal.
Hour narrowing isn't about making anyone compute your hour. It's about finding which of a few reasonable candidates best matches facts you already know.
list candidates → cast each → spot structural differences → cross-check with life facts → rule out / narrow → deep-read 1–2 left
Practitioners have long asked about birth order, family, or remembered years before settling an hour—same idea. Today you can cast multiple versions in software and use AI to summarize differences and draft comparison questions.
What you can and cannot conclude
- Can: shrink “no idea” to likely / second likely / still unclear
- Can: say which facts would help next if data is thin
- Cannot: replace a birth certificate or hospital record as final proof
- Cannot: get a reliable chart from a one-line birthday—models don't run charting engines and often invent pillars or palaces (see our accuracy comparison)
Why one chart—or one screenshot—isn't enough
Casting a single hour is a bet. Hour narrowing needs parallel assumptions: same birthday, different assumed Chinese hour slots—卯 vs 辰 vs 巳, each as its own chart to compare. Don't cast every clock hour (9 AM and 10 AM share the same 巳 slot in BaZi and Zi Wei).
Usually 2–4 candidate hour slots is enough. Convert “morning” into the slots it might span—卯, 辰, 巳, one cast each—not every clock hour. Don't default to midnight when unsure.
If AI will help compare versions, full copyable text beats a screenshot—images drop fields and misread labels. Eight characters or a birthday sentence isn't enough to contrast versions. Each candidate needs complete fields (pillars, palaces, ascendant, etc.) plus a label: “this version assumes X o'clock.”
Some people call that organized export a chart packet—the cast is correct, fields are complete, ready to paste into a chat. What that looks like: AI-readable chart packet.
Five steps (any charting software)
1. List candidates; verify the date
Pull from certificates, hospital notes, or family memory. Confirm the calendar date first—a wrong lunar/solar conversion wastes the whole run. Turn vague phrases into candidate hour slots: “morning” → cast 卯, 辰, and 巳; “around nine” → at least 辰 and 巳 (near a boundary, add casts like 8:50 and 9:10—see true solar time).
2. Cast each candidate in software you trust
Same date and place, change only the hour. BaZi: hour-pillar ten gods vs temperament; Zi Wei: Life Palace and family palaces; astrology: ascendant and houses—one tradition per chat.
Near a true-solar or midnight boundary? Cast an extra version on each side. See true solar time.
3. Turn each cast into comparable text
Label the assumed hour; keep full fields, not just an image. Export or copy if the site allows; if you only have a picture, at least type the biggest divergences (Life Palace, hour pillar, ascendant)—more fields, better comparisons.
4. Add life facts; ask AI to compare
Paste all versions plus facts you're willing to share—temperament, birth order, parents/siblings, remembered years, sleep habits. Ask for key differences and targeted questions: if you pick A vs B, which hour does each support?
Ready-made format: birth-hour narrowing prompt (includes don't recast, don't merge, don't declare the hour).
5. Down to 1–2? Deep-read a single chart
New chat, one packet only (see prompt guide). Still tied between two? Say which you're reading from and keep the other as fallback—don't pretend the hour is settled.
Which clues help most
Comparison works on chart differences × facts you can confirm:
- Temperament—pace, introvert/extrovert, anxiety (versions often diverge on relevant palaces or ten gods)
- Family structure—birth order, parents, siblings
- Remembered life nodes—school, job changes, moves by rough year (check luck cycles)
- Sleep and constitution—early bird vs night owl, heat/cold sensitivity
Five to ten short lines is enough. Skip what's too private. Thin data should end in “still unclear,” not a forced pick.
What AI is good for here—and not
Once multiple versions are cast in real software, AI is useful for plain-language diffs, comparison questions, and ranking candidates from your answers. It should not cast, merge versions, or sense an hour from a birthday alone.
If your chart site only outputs images, comparison gets painful—that's an input problem, not a longer prompt. Site types: how to pick an online charting site.
Astralium is one option: cast, then export copyable packets with full fields for ChatGPT or Claude. If you already use another tool that exports complete text per candidate hour, the five steps above apply the same way.
Related posts
What is birth-hour narrowing?
Can I just ask ChatGPT if I forgot my birth hour?
How many candidate hours do I need?
Can AI finally determine my exact birth hour?
The logic is universal: multiple versions, facts, narrow. If you need copyable full text per candidate, online charting can export packets—then use the narrowing prompt.