Methodology & Transparency
Effective: June 20, 2026
1. There Is No Single Right Answer in Saju
Saju (Four Pillars of Destiny) has branched into multiple schools of thought over centuries. The same birth chart can yield different interpretations depending on the school. Differences in body strength assessment, useful god selection, and chart pattern classification are well-acknowledged within the field. This Service does not hide these differences — it discloses them transparently. Every criterion described below is not "the one correct answer in Saju" but a single consistent standard adopted by this Service for coherence. This page distinguishes two categories. First, standard/consensus items shared by virtually all schools (stem-branch conversion, element assignments, generating/controlling cycles, five-state labels, hidden stems structure, etc.) are adopted without naming a school. Second, every item where the standard disappears the moment values are quantified or a method is chosen (score coefficients, positional weights, root multipliers, strength thresholds, time-correction methods, etc.) is explicitly disclosed below with the adopted school, theory, or source name.
2. Adopted School: Eokbu (Suppression-Support) + Johu (Climatic)
This Service adopts Eokbu (抑扶, Suppression-Support) as the primary method for selecting the Useful God, and Johu (調候, Climatic Balancing) as a secondary supplementary method. The Gyeokguk (格局, Chart Pattern) method carries significant inter-school variance and a high risk of automated misclassification, so it is not used for Useful God selection; chart pattern classification and its success/failure narration are provided as supplementary information, and borderline cases are withheld — no narration is emitted when the verdict is undetermined (the no-auto-confirmation principle, see Section 10). Eokbu: Assesses the Day Master's strength (strong/weak), then selects the element that supports the weak or suppresses the strong as the Useful God. Johu: Suggests a climatic Useful God based on the birth month's seasonal temperature (winter births→Fire supplement, summer births→Water supplement). When the two methods conflict, Eokbu takes priority. The Johu result never automatically overrides Eokbu. This priority itself is a school choice — other schools may reach different conclusions.
3. Ten Gods Counting Criteria
The assessment of excess and absence in Ten Gods groups follows these criteria. Counting scope: Only the 7 positional Ten Gods assignments are counted (5 when birth time is unknown), excluding the Day Master. The Day Master represents "oneself" and is not double-counted as a Companion star. Excess threshold: A Ten Gods group with 3 or more is classified as excess. Absence threshold: A group with 0 is classified as absent. Tiebreak rule: When multiple groups tie for the highest count, the order Companion → Resource → Output → Wealth → Authority determines priority. All criteria above are consistent design values. Other schools may include the Day Master in the Companion count or use different thresholds. Zi-Wu-Si-Hai Ti-Yong (体用) Two-Axis Policy: The earthly branches Zi (子), Wu (午), Si (巳), and Hai (亥) have a unique property where their formal (Ti/体) yin-yang and functional (Yong/用, hidden-stem main-qi) yin-yang are reversed. This Service uses Ti (the conventional yin-yang table: Zi=yang Water, Wu=yang Fire) for the yin-yang balance gauge and Yong (hidden-stem main qi: Zi→Gui=yin Water, Wu→Ding=yin Fire) for Ten Gods determination, never mixing the two without a label within the same output. This two-axis separation policy has no single official standard (Foundation Definition §4a) and is an explicit design solution of this Service.
4. Element Strength Gauge: Per-Element Independent 0~100 Model
The core analysis model of this Service is a per-element independent 0~100 gauge. Unlike the mainstream 610-division method (which distributes element scores within a fixed total), each element has an independent absolute strength unconstrained by a fixed sum. It is structurally possible for all five elements to be classified as "strong." Basis: Adapted from the Andogyun-style 25-point band approach (per-element independent absolute bands, non-sum-preserving), with the Diti Tianshui (滴天髓, Jeokcheonsu) Soe-Wang chapter referenced as classical background for per-element coexistence. This stands in contrast to the mainstream 610-division and Wangdu (旺度, sum-preserving) models. The model choice itself is school-dependent, and whether to normalize (sum-preserving vs. non-preserving) is the top-level fork in the formula. The mainstream fixed-sum approach (610-division) is referenced only as a supplementary distribution indicator, not as the headline identity. Gauge Formula: Each element's strength is computed as a weighted sum of three factors. Strength = (Seasonal Score x 0.5) + (Root Score x 0.3) + (Presence Score x 0.2) The Deukryeong (得令, seasonal power) + Deukji (得地, root power) + Deukse (得勢, momentum) three-factor framework is orthodox Saju methodology (Paljabalchaehae, Japyeongjinjeon). However, the specific weight distribution (0.5 : 0.3 : 0.2) is a calibration provisional value, not a scholarly consensus. Placing the highest weight on the seasonal factor follows the consensus that "the month branch is the single most important variable for strength," but the exact ratio (in the 40~50% range) varies by source. These weights may be adjusted after further grid-sweep calibration and are currently unsettled. Important: The month branch (seasonal power) is the largest single variable in strength assessment, but the proposition "if an element is in season, it is automatically strong" contradicts orthodox Saju doctrine. An element in season can still be weak if it has no roots (rootless) or its momentum is isolated — actual strength is determined by the comprehensive assessment of all three factors: Seasonal Power, Root Power, and Momentum.
5. Wang-Xiang-Xiu-Qiu-Si Score Coefficients
The five seasonal states (旺 Wang, 相 Xiang, 休 Xiu, 囚 Qiu, 死 Si) assigned to each element based on the birth month's seasonal element are a consensus shared by virtually all schools (Diti Tianshui Soe-Wang chapter; Foundation Definition §3). However, there is no scholarly consensus on the numerical coefficients that convert these five labels into scores. The coefficients adopted by this Service are: Wang (旺, Prosperous) = 100 points (coefficient 1.0) Xiang (相, Prime) = 80 points (coefficient 0.8) Xiu (休, Resting) = 60 points (coefficient 0.6) Qiu (囚, Imprisoned) = 40 points (coefficient 0.4) Si (死, Dead) = 20 points (coefficient 0.2) These values are explicit design values of this Service, not a scholarly consensus. Similar quantification attempts exist (Bazi Wangshuai Jingsuan, Wuxing Wangdu Lilun), but some Chinese sources (e.g., guoyi360) have noted these are "one instructor's quantification, not grounded in classical texts." Earth (土) Seasonal Assignment: Earth's prosperous season is assigned to the transitional months (the four Earth branches: Jin-Sul-Chuk-Mi). This differs from the Traditional Chinese Medicine assignment to "Long Summer" and is a school-dependent design choice.
6. Root Tier Hierarchy and Quantitative Multipliers
When a Heavenly Stem's element finds the same element in an Earthly Branch's hidden stems, it is called "rooted" (通根, Tonggeun). The strength varies by root type. The hierarchy (Jangseong-Nok-Wang > main qi > middle qi > residual qi) itself is a consensus grounded in Japyeongjinjeon (子平眞詮) and the Diti Tianshui (滴天髓) commentary by Im Cheol-cho: "长生禄旺，根之重者也；墓库余气，根之轻者也." However, there is no unified standard for the quantitative multipliers that convert this hierarchy into scores. The multipliers adopted by this Service are: Strong Root (main qi of month/day branch, or main qi in 2+ branches) = 1.5x Medium Root (main qi of year/hour branch, or middle/residual qi of month/day branch, etc.) = 1.3x Weak Root (middle/residual qi of year/hour branch, single occurrence only) = 1.1x No Root (element absent from all hidden stems) = 1.0x Transcendence bonus (same element appears in a Heavenly Stem) = +0.1x (not applied when rootless; cap 1.6x) These quantitative multipliers are calibration provisional values and may be adjusted after further grid-sweep measurement. The hierarchical ordering is standard, but the specific score conversion is a design choice.
7. Per-Element Strong/Weak Band Thresholds and Special Patterns
The gauge (0~100) classifies each element into strong, moderate, or weak bands. The cut thresholds are set independently for each element, inspired by the Andogyun-style independent-band approach (each element's band is determined within its own distribution). Earth (土) has notably higher thresholds than other elements because its gauge distribution sits structurally higher (Earth appears in all four transitional month branches' hidden stems); the per-element cuts absorb this structural bias. The cut thresholds were calibrated via a full grid sweep of 17,280 charts (4,320 for unknown-time charts), targeting approximately 22~23% of each element's distribution in the strong band. These cut values are calibration provisional values and may be adjusted after further measurement. Jeonwang-gyeok (專旺格, Special Dominance Pattern): When one element is overwhelmingly dominant and the element that controls it is completely rootless, the chart is classified as a special pattern (Gokjik-gyeok for Wood, Yeomsang-gyeok for Fire, Gasaek-gyeok for Earth, Jonghyeok-gyeok for Metal, Yunha-gyeok for Water). This is based on the Ilhaengdeukgi-gyeok concept from Japyeongjinjeon, but the quantitative thresholds and detection conditions (dominant gauge ≥ 92, margin over the second-highest gauge ≥ 20, the controlling element completely rootless, and the Day Master belonging to the dominant element) are design values of this Service — always evaluated, under these conservative thresholds. The combination of strong/moderate/weak across all 5 gauges yields 3^5 = 243 character types. Seven special patterns (5 Jeonwang + 1 all-weak + 1 all-moderate) are evaluated as a priority layer. This type identification system is a design of this Service.
8. Positional Scoring Weights and Summation Method
The score weight for each position in the Eight Characters is as follows. Month Branch: 30 pts, Day Branch: 15 pts, Hour Branch: 15 pts, Year Branch: 10 pts, Each Heavenly Stem: 10 pts This positional weighting system must be understood in three distinct layers. (1) The methodology of assigning different weights by position = Patent KR101017211B1 (documents a position-based weighting method; the specific score values and whether the month branch receives the highest weight are not covered by the patent — month-branch primacy is this Service's design choice) (2) The specific score values of 30/15/10 = Published school values from Manselyeok (Ten Thousand Year Calendar) community and Sinbinara (2012) (variations exist in the range of 20~36 for the month branch; these are not the sole standard values) (3) The raw sum total of 1,100 (with time) / 850 (without time) = This Service's design value Unlike the mainstream approach of normalizing the total score into fixed proportions (sum-preserving), this Service does not normalize (non-sum-preserving). This applies the design principle of the per-element independent gauge model (Section 4) to positional scoring as well.
9. Time Correction Policies
Birth time correction determines the Eight Characters (especially the Day Pillar and Hour Pillar) and is the primary reason different Manselyeok apps return different Day Pillars for the same birth date and time. For this section in particular, we disclose exactly what is applied, on what basis, and which limits remain. Zi Hour (子時) Boundary: Among the three methods — Night Zi (夜子時), Early Zi (早子時), and Standard Zi (正子時) — this Service adopts the Unified Standard Zi (UNIFIED) method. 23:00 onward applies the next day's Day Pillar. As noted in Korean sources, "there is no definitive conclusion to this day," and the Night Zi split (23:00~00:00 uses next day's Hour Pillar only, keeping the current day's Day Pillar) is not implemented. This choice means the same birth time may yield a different Day Pillar compared to other apps. Birth-City True Solar Time Correction: When a birth city is selected (major cities across Korea, the US, Japan, and China — using official coordinates), the birth time is corrected in proportion to the difference between the city's longitude and the standard meridian actually in force at the birth instant (1 degree of longitude = 4 minutes, an astronomical constant). When the correction crosses midnight, the Day Pillar rolls over with it. The default city is Seoul, and when the birth time is marked "unknown," time correction is skipped entirely. Some research and practitioners (e.g., KCI Choi Won-ho (2023), Korean Professional Saju Association) also recommend applying longitude correction. Standard Time and Daylight Saving Time History: A static transition table — built from the pinned IANA tzdb 2026a release and cross-checked against official gazettes and statutes — applies the standard time and DST actually in force at the birth instant. It covers Korea's 4 standard-meridian changes (1908: 127.5°E → 1912: 135°E → 1954: 127.5°E → 1961: 135°E; Presidential Decree No. 876 and Act No. 676), Korea's 3 DST periods (1948-51, 1955-60, 1987-88 — with 1955-60 running on the 127.5°E base, i.e. UTC+9:30), Japan's 1948-51 DST, China's 1986-91 DST, and the US per-zone regime (from the Standard Time Act of 1918 through the current 2007 rules). Lunar Input Conversion: Lunar birth dates are converted to solar dates by computing the actual new moons with the Meeus astronomical formulas, with the day boundary following the birth country's calendar authority. Korea (and US cities — as a Korean-audience service, a lunar birth record from the diaspora is assumed to follow the Korean lunar calendar) uses Korean Standard Time midnight, matching the modern KASI (Korea Astronomy and Space Science Institute) convention; Japan uses JST; China uses UTC+8 midnight per the Purple Mountain Observatory convention. For months whose new moon falls between the two midnights (roughly 15:00–16:00 UTC), the same lunar date maps to solar dates one day apart by country (about one lunar month every 2.2 years — e.g., Lunar New Year 1997 is February 8 in Korea vs. February 7 in China). When a new moon falls within minutes of midnight, a residual ephemeris uncertainty of about plus or minus 2 minutes leaves a theoretical possibility of a one-day difference from official almanacs. Equation of Time (max approximately plus or minus 16 minutes): The gap between clock time and actual solar time — caused by Earth's orbital eccentricity and axial tilt — is computed at each birth instant using the Meeus astronomical formula and applied. Determinism preserved: All timezone calculations run on fixed tables built into the engine. Regardless of the client device's timezone setting or browser version, the same input always produces the same result. Transparent limits: Chinese timezone records before 1949 are incomplete, so births in that period are shown with an uncertainty notice. In the US between 1945 and 1966, DST observance was decided city by city, so the calculation uses each zone's representative city, also marked with an uncertainty notice. Births near a time transition are handled by a fixed convention — a wall time that never existed (right after a DST start) is read on the pre-transition clock, and a wall time that occurred twice (near a DST end) is read as the earlier instant. The corrections applied to your own reading can be reviewed item by item in the "Time correction details" panel on the result screen. What remains on the roadmap: Support for cities outside these 4 countries, and a selectable Zi-hour method (Night Zi / Early Zi), are items under future consideration.
10. Johu / Chart Pattern / Annual Fortune Disclaimers
Johu (Climatic): The birth month's season is classified as winter (Hai-Zi-Chou)→Fire supplement, summer (Si-Wu-Wei)→Water supplement, or neutral (defers to Eokbu). This 3-zone classification follows the orthodox directional combination seasonal groups but does not apply the per-Day-Master precision table, which varies significantly by school. Chart Pattern (Gyeokguk): Automated chart pattern classification carries the highest inter-school variance and error risk. This Service currently applies chart pattern classification and its success/failure (Seonggyeok/Pagyeok) narration. The special-pattern cutoff (dominant element score ≥ 770 of 1,100) and standard-pattern classification (based on month-branch main-qi stem transparency) are both design values. The verdict rests only on deterministic evidence — presence of the supporting god, exposure of the opposing god, month-branch clash/combination, and combination-away of the transparent pattern stem — and borderline cases are marked "undetermined," in which case no narration is emitted (the no-auto-confirmation principle). Combination-transformation (Hap-hwa) conditions carry significant classical contradictions (Sammyeongtonghoei vs. Japyeongjinjeon), and even English BaZi calculators handle them as manual settings — so combinations are referenced only as verdict evidence, and elemental transformation is not implemented. Annual Fortune (Seun): The annual pillar uses the Gregorian calendar year; the Start-of-Spring (Ipchun) division is not applied. Branch relations (clash, combination, punishment, harm) compare only the annual branch with the day branch, prioritized as: clash > punishment > combination > harm. Combination-transformation of elements is not implemented (display only, no score change). All items above may differ significantly by school. This Service takes a conservative approach that minimizes automated determinations.
11. Compatibility Scoring Criteria
The compatibility score is a weighted sum of 5 categories, computed after calculating each person's chart independently, producing a single 0~100 score. Saju has no standard point allocation such as "30 points for branch relations," so the weights below are design values without a standard, fixed and disclosed. The same two inputs always produce the same score, and every score is an integer (no decimals). Category weights (total 100): Branch Relations: 30 pts — six-harmony (+6), three-harmony (+5 per completed union), clash (-4), punishment (-2), and harm (-1) applied to a baseline of 15 across the two charts' branches (clamped to 0~30) Day Stem Relation: 25 pts — stem combination (25), generating (20), same element (16~18), controlling (10), stem clash (5) between the two Day Masters Element Complementarity: 20 pts — balance of the two charts' combined five-element distribution, plus a generating bonus between dominant elements Ten Gods Direction: 15 pts — a pair-score table over how each Day Master sees the other as a Ten God Useful God Support: 10 pts — how much each chart supplies the other's Useful God element (bidirectional average) Relation type bands: A total of 80~100 is Natural Match (complementary), 65~79 Steady Partners (stable), 50~64 Spark Synergy (dynamic), and 0~49 Growth Partners (challenging), plus the special Soul Mirror type. After the band classification, second-pass override corrections exist — same-element Day Masters + Ten-God condition → mirror, strong branch harmony + total condition → stable, then stem combination + total condition → complementary, applied in that order so the final stem-combination correction takes precedence. When both people share the same character type code, mirror takes highest priority. These band boundaries and correction conditions are also design values and may be recalibrated against sample distributions. Adopted relation constants: 5 stem combinations (甲己, 乙庚, 丙辛, 丁壬, 戊癸), 4 stem clashes (甲庚, 乙辛, 丙壬, 丁癸 — the Earth stems 戊 and 己 have no clash), 6 branch six-harmonies, 4 three-harmony unions, 6 branch clashes, punishments, and harms. For any branch pair, only one relation is scored, prioritized as clash > punishment > harm. Not currently reflected: Shinsal (神殺) — including the branch destruction (破, Pa) and Wonjin (怨嗔) relations — is not currently reflected in the compatibility score. Whether and how to include it is a future consideration; any inclusion will be disclosed on this page.
12. Lucky Elements: Standard and School Differences
The Lucky Elements associated with the Useful God's element (five colors, directions, seasons, Ha-Do numbers, five tastes) come from a classical standard that all East Asian Saju schools agree upon — there is no inter-school variance. These items reflect a traditional standard, not a design choice, and are therefore not subject to school disclosure. However, the career/field and material mappings are modern extensions of traditional correspondences and include some interpretive latitude. Pronunciation Element (Baleum Ohaeng) School Difference: Of the 5 pronunciation-element assignments, 3 are consensus (Wood=ㄱㅋ, Fire=ㄴㄷㄹㅌ, Metal=ㅅㅈㅊ), but the consonant assignments for Earth and Water are diametrically opposite depending on the school. This Service adopts the modern conventional form (Earth=ㅇㅎ, Water=ㅁㅂㅍ), which is the reverse of the Hunminjeongeum Haeryebon (訓民正音 解例本) original text (Earth=ㅁㅂㅍ, Water=ㅇㅎ). This school choice directly affects name/fortune-related content and is therefore disclosed.
13. Major Fortune Period Theme Basis
Each Major Fortune (Daeun) period's theme (coloring) is based on the Heavenly Stem's Ten God for that period. The Earthly Branch's main-qi Ten God is also displayed, but the primary theme determination follows the Heavenly Stem. Whether the Heavenly Stem or the Earthly Branch leads the Major Fortune theme may vary by school. This Service adopts the Heavenly-Stem-primary convention. The remainder-day handling for Major Fortune age calculation (rounding, truncation, or month conversion) varies by school. This Service adopts the truncation (floor) method.
14. Entertainment Content Notice
All analysis criteria described on this page are adopted to present traditional Saju interpretations as entertainment content for fun and self-understanding. No Saju type or element composition is treated as superior or inferior to another, and no claim of scientific validity is made. This Service believes that transparently disclosing consistent criteria is the minimum honesty owed to its users. Every design value is "a consistent standard," not "the right answer."