How Accurate Is Saju? What the Evidence Shows

Accuracy can mean a correct chart, a recognizable description or a fulfilled prediction, and only one of those can actually be checked.

"Is Saju accurate?" contains at least four separate questions: was the chart calculated correctly, did the interpretation feel personally specific, did a prediction occur, and did the reading actually improve a decision? Saju calculation can be checked against a stated method. Its personality and predictive claims have not been scientifically validated. Separating these four questions honestly is more useful than answering with a single percentage.

Accuracy 1: Is the chart calculated correctly?

This is the most testable layer, and the one worth checking first. A Saju service should correctly process the birth date, time, place, time-zone history and solar-term boundaries, using its own stated day and hour conventions, to produce the appropriate Year, Month, Day and Hour stem-branch pairs. Two calculators can disagree with each other for a defensible reason — different conventions, not a simple arithmetic error — but the service should disclose which conventions it uses and flag boundary cases rather than presenting one silent answer. Worth asking directly: which time zone and daylight-saving rule was used, whether the year changes at Ipchun, whether months are based on solar terms, whether longitude or local solar time gets adjusted for, and what happens when the birth time is unknown. A chart that cannot be shown cannot be audited.

Accuracy 2: Does the interpretation feel recognisable?

Recognition matters to the reader, but it isn't proof by itself. A statement can feel accurate for several unrelated reasons: it might be genuinely connected to a recurring pattern, or simply broad enough to fit most people, flattering enough to feel true, interpreted through whatever just happened recently, quietly supported by details the reader already supplied, or remembered as a hit while the misses fade from memory. The Barnum effect describes the tendency to accept general personality descriptions as uniquely personal, and confirmation bias adds to it by making supporting examples easier to notice than contradicting ones. These effects do not prove that every Saju interpretation is useless. They explain why "it felt exactly like me" is weaker evidence than it first appears.

A stronger test converts the claim into behavior. "You value freedom but also need security" is a weak statement — it fits almost anyone. "When a decision becomes irreversible, you may keep multiple options open beyond the point at which optionality is useful" is a stronger pattern hypothesis, because it can actually be checked: against situations where it occurred, one situation where it didn't, the trigger right before the response, the real cost of the pattern, and one alternative action worth testing. Specificity is valuable precisely because it creates the possibility of being wrong.

Accuracy 3: Did a prediction come true?

Predictions require stricter evaluation than reflective descriptions. Before an outcome happens, a fair test records the exact predicted event, the time window, what would count as failure, the base rate of that kind of event, and whether several alternative outcomes were also mentioned at the same time. "Career energy changes this year" is close to impossible to falsify. "You will receive a job offer between March and May without applying" is more testable, but even that needs repeated evidence across many readings, not one striking anecdote, before it means anything. Retrospective matching, fitting broad language to events after they've already happened, is unreliable for exactly this reason. There is no strong scientific evidence establishing Saju as a reliable predictor of specific future events.

Accuracy 4: Did the reading improve a decision?

A reading can be useful without being predictively true. Suppose it prompts someone to notice that they repeatedly underestimate maintenance costs after a launch; they add a capacity threshold to a business plan and avoid overextending. The practical benefit comes from the question and the resulting action, not from any proof that the birth chart caused or predicted the pattern. Decision usefulness can be checked directly: did the reading clarify the actual options, identify a risk that could be verified, preserve alternative explanations instead of collapsing to one story, lead to a reversible test, and reduce rather than increase fear and dependence. This is the standard Kairo prioritizes.

What the 2026 Korean survey does and doesn't show

Gallup Korea reported that among 607 respondents who had paid for fortune-telling or Saju, 59% said the content matched reality. That figure describes a self-reported perception among previous customers, and Gallup explicitly noted that the survey cannot establish a causal relationship between belief and perceived real-world accuracy. It demonstrates cultural use and subjective experience. It does not validate Saju predictions scientifically.

Red flags in accuracy claims

Be cautious of a service that states a precise accuracy percentage without a published study behind it, that implies a longer report must be more accurate than a short one, that treats disagreement as proof your birth time must be wrong, that claims scepticism blocks the reading from working, that offers medical, legal or investment certainty, or that predicts disaster unless you buy a remedy or an additional session. Each of these claims makes a service harder to evaluate, not easier.

A seven-day accuracy audit

After receiving a reading:

Highlight five specific claims.
Rewrite each as observable behaviour.
Record supporting and contradicting examples.
Mark which claims could fit most people.
Separate natal interpretation from future prediction.
Test one low-risk action suggested by the pattern.
Review whether the reading increased clarity or only certainty.

Clarity allows uncertainty to remain visible. False certainty hides it.