Do Zi Wei Dou Shu Charts Repeat More Often Than Ba Zi Charts?
"Zi Wei charts collide all the time, Ba Zi charts almost never do." You've probably heard this one.
It's something you can count.
I took the same set of birth times and cast them in both systems. Ba Zi split them into about 380,000 kinds. Zi Wei, even with the Four Transformations and the malefic stars all included, split them into only about 220,000, and after sixty years Zi Wei barely grows at all.
Curious about your own chart? Cast it free in 30 seconds →
Almost every comparison between Zi Wei Dou Shu (紫微斗数, Purple Star Astrology) and Ba Zi (八字, the Four Pillars, literally "eight characters") is made by analogy. A microscope versus a line graph. Space versus time. They all sound reasonable, and none of them can be checked.
Both systems actually take the same input: one birth time, year, month, day and hour. Feed the same batch of times into each, and how many categories each one produces is something you can count. As far as I can tell, nobody has.
I happen to have the chart engine. Over the years of studying Zi Wei, I wrote each step, placing the stars, setting the Element Bureau, assigning the Four Transformations, into code, and it can produce Ba Zi pillars as well. Since it can cast both, I counted.
Here is exactly how.
The first run covered January 1, 1965 to December 31, 2024: sixty years, twelve two-hour periods a day, 262,980 birth times in total. The second run went back to January 1, 1905: one hundred and twenty years, 525,960 birth times.
For each birth time, one side cast a Zi Wei natal chart, using the San He (三合, Three Harmonies) school's Four Transformations table, without true solar time, for a male native. The other side cast the Ba Zi pillars: the year pillar changes at Lichun (立春, Start of Spring), the month pillar changes at the solar terms, the day pillar comes from the perpetual calendar, and the hour pillar uses the Five Rats rule (五鼠遁). Both runs had zero casting failures.
I counted Zi Wei at two levels. The main-star level looks at the Earthly Branch of the Life Palace (命宫), the Five Element Bureau (五行局), and which main stars sit in each of the twelve palaces. When people say a chart has a good structure, or that four people a month share your chart, this is the level they mean. The strict level adds the Body Palace (身宫), the auxiliary stars, the malefic stars, the minor stars, and the natal year Four Transformations on top of that.
The unit of counting is the birth time. Each time counts once, regardless of how many people were actually born in that two-hour period.
The script is at xiaoS/census/ziwei-vs-bazi.test.ts. The results are stored in xiaoS/data/20261004-census-ziwei-vs-bazi.json (sixty years) and xiaoS/data/20261005-census-ziwei-vs-bazi-120y.json (one hundred and twenty years).
First, the sixty years.
Ba Zi: 262,932 kinds. Out of 262,980 birth times, that's only 48 short. All 48 sit at the two ends of the window: January 1 to 4, 1965, and December 17 to 20, 2024. Both ends fall in a Jia Chen (甲辰) year and a Bing Zi (丙子) month, and the day pillars happen to come round to the same point in the cycle, so the 48 times pair up exactly. Within one complete sixty-year cycle, no two Ba Zi charts are the same.
When people say Ba Zi charts don't collide, within this window they're right.
Zi Wei at the main-star level: 720 kinds.
On average each chart pattern is shared by 365.25 birth times, and the most crowded one by 602. When I saw 720, I assumed the script had lost the palace ordering somewhere. I went back and checked palace by palace. Nothing was lost. 720 is exactly the 144 basic charts multiplied by the 5 Element Bureaus.
The strict level is much wider: 225,754 kinds. Of these, 194,498 appear only once in the sample; counted by birth time, 73.959% of birth times map to a chart pattern of their own. The remaining 31,256 each sit on top of 2 to 5 different Ba Zi charts.
225,754 against 262,932: the number of Zi Wei chart patterns is about 86% of the number of Ba Zi charts. Ba Zi is a bit finer, but Zi Wei isn't far behind.
What happens if the window gets longer?
On October 4, in a comment on Zhihu, I said that a sixty-year window is the most favorable one for Ba Zi, and that at one hundred and twenty years Ba Zi charts would start to come in pairs. I also threw in a guess that Zi Wei would close the gap.
So I ran it a second time.
Ba Zi: 388,548 kinds. The collisions have started.
Every collision is between birth times sixty years apart. Midnight hour (子时) on January 1, 1905 and midnight hour on December 17, 1964 both have the pillars Jia Chen, Bing Zi, Geng Zi, Bing Zi. The first half of my prediction came true.
The main-star level is still 720 kinds. On average each one is shared by 730.5 birth times and sits on top of 721.83 kinds of Ba Zi, with a maximum of 1,198.
The strict level: 227,166 kinds. The script prints progress every fifteen years, and at the end of 1964, the first sixty years had produced 225,738 kinds. By the end of 2024, 227,166. The second sixty years added only 1,428.
I was waiting for Zi Wei to close the gap, and it barely moved. My first suspicion was that the progress printout used the wrong variable. I checked against the results file. The numbers were right.
The strict level is almost saturated within sixty years. The Zi Wei chart count fell from 86% of the Ba Zi count to 58%, and the gap widened. Of these 227,166 kinds, only 2,840 sit on top of a single Ba Zi; 98.75% of chart patterns sit on top of two or more. Counted by birth time, 99.46% of birth times share their chart pattern with some other time. On average each pattern sits on top of 2.29 kinds of Ba Zi, with a maximum of 9.
My prediction was wrong.
Now count the other way round: how many Zi Wei charts sit under a single Ba Zi?
Of the 388,548 kinds of Ba Zi, 257,386 have only one main-star Zi Wei chart under them. 131,154 have two. And 8 have three.
Those with two or more make up 33.757% of Ba Zi kinds, a third. Counted by birth time it's 49.877%: for half of all birth times, the same Ba Zi can be produced by another time in the sample, and that other time's main-star Zi Wei chart is different. Counting at the strict level instead, it's 131,360 kinds, 33.808%, covering 49.952% of birth times.
Why would the same eight characters produce two Zi Wei charts? Ba Zi follows the solar terms and the stem-branch cycle, and returns to the same set of characters every sixty years. Zi Wei places the Life Palace using the lunar month and the hour, and places the Zi Wei star (紫微, the Emperor star) using the lunar day and the Element Bureau. The same stem-branch day sixty years apart doesn't fall on the same lunar date, so the palace where Zi Wei starts changes, and all the main stars shift along with it.
People often ask whether there are people with the same Ba Zi but different Zi Wei charts.
There are. Two people born sixty years apart. In this sample, half of all birth times are in that situation.
What these numbers can show is how the two systems classify.
Ba Zi packs all the information into eight characters and produces the most categories, but every layer of meaning has to be drawn out by you. Zi Wei lays the same information out in advance across twelve palaces, which makes it more direct to read; the cost is that if you only look at the main stars, the number of categories drops from over two hundred thousand to 720.
The classical Zi Wei texts say this themselves. The Complete Book of Zi Wei Dou Shu (《紫微斗数全书》) has the line: "Stars may share the same positions, yet fates have their distinctions; one must understand the essentials of generation and control, and examine whether the stars are well placed or out of place." Under the same star map, fates still differ, depending on whether each star sits in a palace where it is strong or weak.
On the Ba Zi side, the Yuan Hai Zi Ping (《渊海子平》) describes the four pillars this way: "Take the day as the master, the year as the root, the month as the governing outline, the hour as the assistant." Each of the eight characters has its own job, and every one of them does classifying work.
Each passage speaks for its own school. I'm not using them to rank one above the other.
What these numbers can't show needs a few more words.
More categories doesn't mean more accurate readings. All I counted is how many categories each system produces. Which one is right is a question this data can't answer, and this post doesn't try to.
I only ran the San He Four Transformations table. With a Flying Star (飞星) table, the strict-level numbers would change. By how much, I haven't run.
True solar time wasn't used. Birth times right on an hour boundary would shift. By how much, I haven't checked.
I only ran male charts. Gender only affects the direction of the major periods, not the natal star positions, but I genuinely haven't run female charts.
Neither side included the ten-year major periods (大限) or annual charts (流年). I only counted the natal chart.
The 525,960 is an exhaustive enumeration of times, with every two-hour period weighted equally. Real births aren't spread evenly across the day, so this distribution doesn't match the real population.
And one hundred and twenty years isn't the ceiling for Ba Zi either. Stretch the window further and Ba Zi will keep splitting, while Zi Wei's strict level is already nearly saturated. The 58% figure belongs to these one hundred and twenty years only.
Do Ba Zi charts really never collide? Within one sixty-year cycle, they don't. Stretch to one hundred and twenty years and they do, always between times sixty years apart.
Why do Zi Wei Dou Shu charts collide more than Ba Zi? It depends on which level you count. Main stars alone give only 720 chart patterns. With the Body Palace, auxiliary stars, malefic stars, minor stars and Four Transformations all included, sixty years give 225,754, still fewer than Ba Zi.
Can the same Ba Zi produce two different Zi Wei charts? Yes. In this sample, a third of Ba Zi charts sit on top of two or more main-star Zi Wei charts, coming from birth times sixty years apart.
Can I reproduce these numbers myself? Yes. San He Four Transformations table, no true solar time, male native, Ba Zi year pillar changing at Lichun and month pillar at the solar terms. Use whatever charting program you have to cast the same batch of times both ways, then deduplicate and count. The paths to the script and results files are above.
Back to the line we started with: "Zi Wei charts collide a lot, Ba Zi charts almost never do."
Over sixty years, true. Over one hundred and twenty years, half true.
Next time someone uses chart collisions to argue which system is finer, you might first ask them: is your Zi Wei the main stars only, or with the Four Transformations and malefic stars included? And then ask: how many years is your window?
⚠️ FateStar generates and interprets your chart based on the traditional Chinese discipline of Zi Wei Dou Shu (紫微斗数). All content is for informational and reflective purposes only.
About the Author

Founder of FateStar. He began structured study of San He Zi Wei Dou Shu in 2020 and is responsible for product method, source organization, and content review. Personal experience is not presented as scientific proof.
More about Louis →Related Articles

I Cast 262,980 Charts to Find Out How Rare Zi Wei Dou Shu Configurations Really Are
At What Age Do You Reach Each Palace's Decade in Zi Wei Dou Shu?
How Many Kinds of Marriage Can Zi Wei Dou Shu Actually Tell Apart? I Counted All 262,980 Birth Moments
Born in a Leap Month: How Do You Set Up a Zi Wei Dou Shu Chart?
