How Does the 7-Day Forecast Decide Its Colors?
Each day in Bokki's weekly calendar is one comparison: today's element against your chart's favorable and unfavorable elements — and the order you ask in is the whole rule.
Bokki Editorial
The short answer
Bokki's /week screen paints each of the next seven days ☀️, ⛅ or 🌧️ — good, steady or care. Nothing about you besides your birth date (and birth time, if you have it) goes into that call: not your name, not your gender, not where you live. Each day's color comes from exactly one comparison: does that day's element belong to your chart's favorable elements (the ones your chart is read as needing more of), or does it match your chart's single unfavorable element? The function that decides this, moodOf(), asks those two questions in a fixed order and stops at the first "yes" — favorable first, unfavorable second, and only if neither applies does a day fall back to steady. That order matters more than it sounds like it should, because a chart can occasionally answer yes to both.
The rule — an order, a fixed score, and a ten-day cycle underneath
The engine reads today's day pillar (the same stem-and-branch pairing used everywhere else in a Saju chart) and pulls out just the stem's element. It calls moodOf(element, favorable, unfavorable): if the element is one of your chart's three favorable elements, the day is good; failing that, if the element equals your one unfavorable element, the day is care; otherwise it's steady. Each mood carries a fixed score — good is always 82, steady always 70, care always 60 — and that score never wiggles. It used to: an earlier version added a small random jitter to keep the numbers from feeling too mechanical, but that jitter could push a care day's score across the boundary that decides which word the screen prints, so two days with the identical mood could display two different tiers depending on nothing but a hash. The fix removed the jitter entirely — a fixed mood now always prints the same word. Underneath all of this sits a cycle that has nothing to do with your chart: the ten heavenly stems (甲乙丙丁戊己庚辛壬癸) pair off with the Five Elements two stems at a time — 甲乙 is Wood, 丙丁 is Fire, 戊己 is Earth, 庚辛 is Metal, 壬癸 is Water — so any single element's day comes back exactly twice in every ten-day stretch. A seven-day window is shorter than that ten-day cycle, so depending on exactly where the week starts, it can catch both of an element's two days back to back, just one of them, or neither. Nothing about your chart controls that part; the calendar does.
Worked example — the same three-element streak, twice
Take the chart daeun, sipsin and guardian have already used: born 15 May 1990 at 14:30 in Seoul, a chart read as weak, with favorable elements Water, Earth and Metal and one unfavorable element, Fire. Run the week starting 8 September 2026 (today, as this article publishes) through weeklyForecast(), and the day stems for that week happen to be 乙丙丁戊己庚辛 — stems two through eight of the ten-stem cycle. That window catches both stems of Fire, both of Earth and both of Metal, only one of Wood's two stems, and neither of Water's. The result: Tuesday (Wood) comes back steady; Wednesday and Thursday (Fire, Fire) both come back care — two care days in a row; Friday and Saturday (Earth, Earth) both come back good; Sunday and Monday (Metal, Metal) both come back good again. Not one Water day appears all week. Every one of those repeats and gaps is the ten-day cycle showing through a seven-day window — the chart's own favorable/unfavorable list never changes mid-week, only which element the calendar hands it each day.
Now the tie case the rule was built for: a birth on 15 January 1950 at noon reads as a balanced chart with favorable elements Fire, Wood and Metal — and its one unfavorable element is also Metal. Scanning the 15th of every month from 1950 through 2025 (912 charts, one birth time, noon, held fixed) turns up 91 charts like this — about 10.0% — where the favorable list and the unfavorable element share an element. Running that January 1950 chart's own week forward to a stretch that includes a Metal day shows exactly what the order predicts: the Metal day comes back good, not care, at the full score of 82 — because moodOf() checks favorable first and stops there. A chart that can want and dislike the same element at once still needs one answer on the calendar, and the order the code asks in is that answer.
What this calculation doesn't decide
The daily line of text under each color ("Good flow. Move first." or "A careful day — tidy up rather than push.") is chosen by a hash of your birth date and the day's date, from a pool of three per mood — that choice adds variety to the wording, never to the score or the color. The calculation reads only your birth date and, if you supply it, your birth time; it never reads your gender, your name, or your location, and it never checks your day master's own element the way some other Bokki calculations do. It also isn't a probability or a prediction about what will happen that day — it's a single structural comparison, repeated seven times, between a fixed three-element list and a fixed one-element list. And it doesn't smooth out the clustering the ten-day cycle produces: a week that happens to land on both of an unfavorable element's stems will show two care days in a row, and a week that misses a favorable element entirely will show none of it, by construction — not by chance in either direction.
Closing note — a fixed order, a fixed score, a cycle that isn't yours
In short: Bokki's 7-day forecast never rolls a die. Each day, it takes that day's element from the ten-stem cycle and asks two fixed questions in a fixed order against your chart's favorable and unfavorable elements — favorable wins ties, and the score attached to each answer never moves. The only randomness left is which of three near-identical sentences gets shown under an already-decided color. What looks like week-to-week variety — two good days in a row here, a run of care there, a color that never shows up at all — is the ten-day stem cycle running underneath a seven-day window, doing the same arithmetic for everyone, on a schedule nobody's chart controls.