One Astrology Finding Just Survived the Strictest Multiple-Comparison Correction

When you test 26 astrological hypotheses on the same dataset, you have a multiple-comparison problem. Any one hypothesis might cross p<0.05 purely by chance — after 26 tries, the probability that at least one of them does is about 73%. "We tested a bunch of hypotheses and some were significant" is not a finding. It's a statistical error waiting to be pointed out in peer review.

The standard fix is false-discovery-rate correction. There are two versions:

• Benjamini–Hochberg (BH) — controls false-discovery rate at q=0.05 under the assumption that your tests are independent (or positively dependent in the right way).
• Benjamini–Yekutieli (BY) — controls FDR at q=0.05 without any independence assumption, in the worst case. Much stricter.

Of our 26 hypotheses at 10,000 permutations, four pass BH. One — and only one — passes BY.

The one: Vedic benefic mahadasha

Hypothesis (H6, Parashara/Vimshottari): At the time of a major life event, the active Vimshottari mahadasha ruler is one of the four Vedic benefic planets — Jupiter, Venus, Mercury, or the Moon.

Result at 10,000 permutations:

• Observed rate at peaks: 47.5% (134/282)
• Null rate under shuffled pairings: 37.3%
• Cohen's h: +0.207
• Two-sided p: 0.0001
• BH-FDR at q=0.05: pass
• BY-FDR at q=0.05: pass

That p=0.0001 is the lowest in the study. It's not close to the correction thresholds. It just passes them cleanly.

What "dasha ruler" means here

Vimshottari is the most widely used Vedic time-lord system. Your life, from birth to death, is divided into 9 planetary periods (mahadashas) ranging from 6 years (Sun) to 20 years (Venus). Which mahadasha you're in at any moment is determined by the Moon's position at your birth.

The four "Vedic benefics" — Jupiter, Venus, Mercury, Moon — total about 56 years of any 120-year cycle. So you'd expect any random moment to fall in a Vedic benefic's mahadasha roughly 47% of the time. The null rate in our permutation test came out to 37.3%, slightly below this, because our cohort ages skew toward earlier parts of the cycle. The observed rate for peaks is 47.5%.

Peaks are disproportionately likely to happen during Vedic benefic mahadashas.

The catch

Setbacks fire this at 48.0% — essentially the same as peaks. The gap between peaks and setbacks is −0.5 percentage points.

So H6 is not a peak indicator. It's a major-event indicator. Something important happens to you — positive or negative — and you're probably in one of Jupiter, Venus, Mercury, or Moon's mahadasha when it does.

That changes how we treat it in the production scoring model. If we were building a general "eventful days" prediction, we'd give H6 a big weight. We're building a luck score — peaks good, setbacks bad — so H6 gets a smaller weight than its raw significance suggests. It's still predictive for the model, because the background rate in the general population is lower than the eventful-moment rate, so firing H6 does push a day's score toward "something important might happen."

But the useful product question is: of those eventful moments, which direction? And H6 can't tell you that.

Why this matters for astrology research

Vimshottari is not a fringe technique. It's the backbone of modern Vedic astrology, a 9-planet time-lord system that predates the Hellenistic tradition. If you ask a competent Vedic astrologer what period of life you're in, they'll answer in Vimshottari terms.

What we've shown is that the direction of Vimshottari's classical benefic claim — that benefic dashas coincide with life events — is statistically robust. Not just robust to single-hypothesis testing but robust to the strictest multiple-comparison correction you can apply across 26 hypotheses.

What we've also shown is that the valence claim — that benefic dashas specifically produce good events — is not supported. Good events and bad events both fire above baseline in benefic mahadashas.

That's a correction to the classical doctrine, not a refutation of it. Vimshottari is measuring something real. The real thing it measures is timing of amplitude, not timing of luck.

Context for practitioners

If you're using Vimshottari in client work, here's what you can and can't say based on this study:

• You can say: "During a benefic mahadasha, the chance of a high-amplitude life event is meaningfully elevated." That's supported at p=0.0001 with the strictest correction.
• You cannot say (based on this data alone): "During a benefic mahadasha, things are likely to go well for you." Peaks and setbacks fire at equal rates.
• You should be careful saying (based on this data): "Benefic mahadashas are safe periods." The eventfulness floor is raised, not lowered.

One cohort isn't the whole story. Our peaks include people who eventually achieved documented success; their "benefic mahadasha setbacks" may have been productive setbacks en route to those peaks, rather than terminal ones. But even allowing for that, the symmetric peak/setback response pattern is striking enough to write down.

The other three that survived BH (but not BY)

• H21 Climacteric year — Valens's age-based timing (p=0.001)
• H7 Jupiter mahadasha specifically (p=0.002)
• H18 Saturn-return window (p=0.006)

Each is discussed in the full findings document at findings-2026-04-17-10k-permutation.md. Methodology code and raw outputs are reproducible from the scripts in services/astro-api/scripts/.