The short version
We do not describe the product as 100% accurate, the most authoritative or scientifically proven. We focus on three things: whether the calculation is stable, whether the rules can be explained, and whether the testing can be reviewed. Across the recorded tests so far, 62 of 80 questions met the preset matching criteria, or 77.5%.
1. We do not treat accuracy as a slogan
BaZi analysis is not a maths problem. Not every conclusion has a single correct answer. So Destiny Book does not describe itself as 100% accurate, the most authoritative or scientifically proven.
We focus on three things: whether the calculation is stable, whether the rules behind a judgement can be explained, and whether the testing can be reviewed.
2. Published benchmark
Destiny Book has been tested against public question sets. In the recorded results so far, 62 of 80 questions across two sets met the preset matching criteria, or 77.5%.
That 77.5% only means the result obtained under those specific questions, that scoring standard and the engine version at the time. It is not a 77.5% accuracy rate for predicting someone's life, and it does not generalise to all BaZi questions.
Scoring uses Exact Match: the final option has to match the answer key exactly to count, with no partial credit. The random baseline for the two sets is 25%.
3. Why we publish the underlying data
A single percentage is hard to verify on its own. So the goal is not just to show 77.5%, but to publish over time the question sources, the engine version, the output for each question, the scoring standard, the final judgement and the test dates.
The question sources, the scoring standard and every judged answer are published alongside the results, so anyone can check them.
4. Open tooling
Destiny Book has published the mingshu-bazi project: a CLI tool and an Agent Skill for AI agents. Anyone can use it to call the same chart plotting and judgement capabilities.
The public scope does not include the website source code or the chart engine source code. The repository documents the interface contract, field definitions, calling conventions, limitations and error handling.
The documentation also states the limits plainly: the birth-time convention has to be selected; unresolvable locations and time ambiguity are not silently downgraded; and the Five Elements and Ten Gods percentages are engine weights, not probabilities. We consider those limits part of the product's credibility too.
5. Content credibility
The core BaZi content is written by hand. Xiaohe currently produces most of it, and Xiaohe and Yuxuan review it together.
Entries state their author information, update date, reference material and notes on the analytical framework.
6. What we do not promise
Destiny Book does not promise:
- That any stated outcome will happen
- A reading offers a structured angle for reflection, not a forecast of a fixed event.
- That any framework is the only correct answer
- BaZi has different school conventions; the results here represent the framework we currently use.
- That using the product guarantees wealth, marriage, career or health outcomes
- Real outcomes depend on many factors that have nothing to do with the chart.
Destiny Book offers structured analysis and study material within a traditional framework. Major medical, legal, financial and life decisions should not rest on a reading alone.