Astroyana research · methodology

How we run the research

Every figure in our research studies comes from a large, synthetic and reproducible sample of birth charts computed with the production Astroyana engine. This page explains what the numbers mean, how we sample, the limits of the method and how to cite the work.

Independent review, 11 October 2026: birth clustering is a scenario under assumed weights, not an estimate of India’s population. The monthly weights lack a traceable source table; the hourly weights are adapted from a US report and the 40% scheduled share is assumed. A rounded 0.00% is not proof of impossibility. The precision table applies to sample proportions, not means, ratios, rule uncertainty or uncertainty in the weights.

What these studies measure

Each study measures the internal structure of the astrological system: how often a pattern occurs across possible birth moments (its base rate), and how the classical rules interact — for example, how often a dosha is cancelled.

These studies make no claims about real people or outcomes. They use no surveys or data about anyone’s career, marriage or health.
A figure such as “5.5% have a full Kaal Sarp” means 5.5% of sampled birth moments produce that chart pattern — a base rate per birth moment, not a census of any population.
They use no personal data. No real user data is read or stored. The scripts generate random birth moments and write only aggregate percentages — there is no personal data involved.

How we sample

For each chart we draw a birth moment at random:

Date: Uniform, 1950–2009
Time: Uniform across the 24-hour clock
Place: 25 major Indian cities, weighted by population
Timezone: IST (UTC+5:30)
Sample size: 50,000 charts; doshas 100,000; birth priors 100,000 per regime; Saturn 5,000
Reproducibility: A fixed seed reproduces every statistic exactly when the study is rerun on the same engine commit.
Precision: 95% half-width ±0.44 pp at N = 50,000 (±0.31 pp at 100,000; ±1.39 pp at 5,000)
Dataset revision: 2 (10 October 2026); revision 1 was June 2026

The selected 25 cities have population-based weights; that does not establish population representativeness. Most studies draw dates and times uniformly. Birth clustering compares that baseline with assumed monthly and hourly weights.

The engine

Charts are computed by the same functions that power our live calculators and reports: the Swiss Ephemeris with bundled offline data, the Lahiri ayanamsa (the Indian government standard — except the ayanamsa study, which deliberately compares four systems), mean nodes and whole-sign houses. Each dataset records the exact engine version, the engine git commit, the Swiss Ephemeris and pyswisseph versions, the generation date and its dataset revision, so any figure can be pinned to the code that produced it. Because every figure is “how often this pattern occurs under the rule the engine applies”, the datasets are regenerated whenever a source audit corrects a rule, and the change is recorded in the revision history below.

The studies at a glance

Compare your chart with the base rates
Ashtakoota baseline computed over 50,000 random couples. Enter your own birth details to see where your match lands on the distribution.
Strengths and weaknesses across the whole chart
Per chart, strengths (exalted or own-sign planets, auspicious yogas) against weaknesses (uncancelled natal doshas, debilitated and combust planets) across 50,000 charts; negative yogas and Kemadruma never count as strengths.
Dosha prevalence
Mangal (detect_mangal_dosha, the matchmaking rule: Mars in houses 1/4/7/8/12 from the Lagna or Moon), Kaal Sarp (detect_kala_sarpa_dosha) and Saturn’s house from the natal Moon on the transit date, tallied over 100,000 charts. Sade Sati is split by house (12th/1st/2nd); the 4th- and 8th-house transits are reported by house, not as Dhaiya or Ashtama Shani.
Guna Milan score distribution
Ashtakoota (ashtakoot_score) over 50,000 random pairs — the 36-point score, the 18-point pass rate, Nadi/Bhakoot present vs. cancelled, and the share of perfect scores that are same-nakshatra pairs. Bhakoot dosha on the 2/12, 5/9 and 6/8 axes; Tara malefic only on remainders 3/5/7.
How ayanamsa changes a chart
Each chart recomputed under Lahiri, KP, Raman and True-Chitra; how often the Moon sign, nakshatra, ascendant and Manglik verdict differ.
Birth-time sensitivity
Each chart recomputed at +15/+30/+60 minutes; how often the ascendant, nakshatra and Manglik verdict change.
Sign distributions under assumed birth patterns
Uniform versus assumed monthly/hourly priors, with 100,000 synthetic charts per regime; Sun, ascendant, Moon sign and nakshatra distributions. Not a population estimate.
How common are named yogas?
detect_all_yogas across 50,000 charts; the share of charts carrying each named yoga, with Gajakesari reported under both the qualified BPHS Ch.36 rule (the engine default) and the popular angle-only rule. Negative yogas (Asubha, Daridra, Kemadruma) are counted separately and never as auspicious.
Time in Saturn-from-Moon transits
compute_sade_sati uses a 7-day scan with cusp refinement across each chart’s 90-year lifetime for Sade Sati (Saturn in the 12th/1st/2nd from the Moon); a 15-day scan of the same ephemeris measures Saturn in the 4th and 8th from the Moon by house, with a published cross-check between the two methods.
Planetary dignity at birth
Exalted / own-sign / debilitated (classical tables), combust (per-planet Sun orbs) and retrograde (apparent speed) per planet.

Limitations

Synthetic, not a census. Base rates per birth moment, not percentages of a real population.
Uniform time and date. Real birth-hour and seasonal-birth curves would nudge the house-based figures.
Rule definitions. Results reflect the engine’s classical rule set (which houses count for Manglik, which cancellation rules apply); other schools differ, and we state the engine’s choices.
No outcome data. Nothing here connects a chart pattern to a life event.
Rules are revised. When a source audit corrects an engine rule, the datasets are regenerated and the change is recorded below; cite a figure with its dataset revision.

11 October 2026 review: consistently including both node directions changes charts without weakness markers from 4.55% to 4.49% and average weakness markers from 2.86 to 2.89. Across 5,000 lifetimes, the 7-day Sade Sati scan averages 23.7 years. Against the 15-day scan, the absolute difference in total duration per lifetime averages 20 days, reaches 287 days, and is within 30 days for 94.5% of lifetimes. These are differences between scans, not guarantees of exact timing.

Revision history

Revision 2 (10 October 2026). All nine studies were rerun with the fixed seed. Each dataset records its generating commit, including the later Gajakesari correction. Matching rounded Moon distributions support reproducibility but do not prove individual birth-record identity. Figures that moved: Gajakesari 33.4% → 14.5% (the engine now applies the qualified BPHS Ch.36 rule as printed, an angle from the ascendant or the Moon with a benefic and no affliction; the old angle-only rule still gives exactly 33.4% and both are published); net Manglik 41.9% → 28.0% (the rule narrowed to houses 1/4/7/8/12 from the Lagna or Moon, so there is no “high” severity tier any more); couples clearing 18/36 79.1% → 78.2% and Bhakoot dosha 33.4% → 49.8% (the 5/9 axis is now a dosha); perfect 36/36 0.57% → 2.83% (janma-tāra is no longer malefic); charts with both strengths and weaknesses 96.1% → 95.5%; Manglik verdict flips under a one-hour error 3.5% → 7.4% and the system-dependent Manglik verdict 5.2% → 4.4% (both from the Manglik narrowing). Unchanged: planet dignity (all values), birth clustering, the ayanamsa core disagreement (18.5%), rising-sign changes (50%), Kaal Sarp (5.5% full) and Sade Sati now (24.8%). The Sade Sati split and the 4th/8th-house Saturn figures are numerically unchanged but are now reported by house, because the engine’s August 2026 source audit found no printed source for the phase names or the Dhaiya and Ashtama Shani labels and suppressed them. The Saturn-burden total (41.0 of 90 years, 45.5%) is unchanged; the study now also publishes the stricter occupancy basis (37.5 years, 41.6%), which counts only the days Saturn is actually inside those houses, and a published cross-check between the engine’s exact windows and the study’s own scan (their aggregate mean span differs by 0.01 years, not a per-chart error bound). Three defects in the revision-1 study code were also fixed: the yoga names “Adhi” and “Parivartana” did not match the engine’s names and so published as 0.0% (now 1.0% and 37.3%); the auspicious-yoga filter would have counted Kemadruma as a strength once the engine began emitting it; and eight of nine datasets carried no engine version.

Revision 1 (June 2026): first publication. The cause of every figure that changed between the two revisions, and the checks run before publishing, are recorded in the engine’s methodology document, scripts/data_study/README.md.

Privacy & citation

Privacy: synthetic data only — no real user data is read, stored, or derivable from the published aggregates.

Citation: free to cite with attribution to Astroyana. The figure, the study URL, and “synthetic sample, Lahiri sidereal engine” are enough. Every study page also carries machine-readable Dataset metadata for reference managers and AI assistants.

Questions about the method or a specific number? Each study restates its own parameters in its Methodology section, and the figures are fully reproducible from the committed scripts.