Learn

How we test astrology

AstroTradeLab takes popular financial-astrology claims and asks a single question of each: does the real sky do better than a shifted one? Here is how the tests are kept fair.

1. Claims fixed before results

Twelve claims, the exact windows, stock groups and pass rules were written down and timestamped on the Bitcoin blockchain before any result was computed. Changes made later are listed with their date and whether any result had been seen.

2. Every NSE stock, real astronomy

About 2,600 NSE main-board stocks from 2016; portfolios use those trading at least ₹1 crore a day. Planet positions come from NASA JPL's DE421 ephemeris with the Lahiri ayanamsa, checked against published Panchangs and eclipse dates. Companies get a listing-day chart.

3. Placebo calendars

The heart of the method. For each claim we take the real astrological calendar and shift it by a random number of days (at least 60), 2,000 times. Each shifted calendar keeps everything about the real one, how long retrogrades last, how often eclipses come, how the windows cluster, but puts them on the wrong dates. If the claim is true, the real calendar should beat almost all of them. The p-value is the share of shifted calendars that do as well as the real one: p = 0.03 means only 3% of shifted calendars matched it.

4. Correcting for many tests

Test thirteen ideas at the usual 5% bar and the chance that at least one passes by luck alone is about 49%. We therefore apply the Holm correction across all tests: the smallest p-value is multiplied by the number of tests, the next by one fewer, and so on. A claim counts only if it still clears 0.05 after that. Gochar stock picking (H11) shows why this matters: p = 0.025 on its own, 0.32 after correction.

5. A sealed holdout

Results so far use 2016–2023. Prices from 2024 onwards stay locked in the code and are opened once, after the methods are frozen, as a final out-of-sample check.

6. Costs and realism

Portfolio tests trade at the close with 0.10% cost per side plus 0.10% securities transaction tax on sales, and use only data available at the time. Stocks that delisted are missing from NSE's current list, a survivorship bias we state rather than hide.

7. Live paper portfolios

Since 27 October 2026 the astrology portfolios run live on paper, each next to a twin that follows the same rule with the planets scrambled. Every day's results are hashed into a public ledger and timestamped, so nobody can edit the past. See the portfolios.

What a "no" means

A claim that fails here has not been proven false for all time; the test shows the effect, if any, is too small to tell apart from chance in eight years of NSE data. All results are on the home page, with one page per test below.

For learning and curiosity, not investment advice.