Our Accuracy, Out in the Open
No astrology site we know of measures its own accuracy and publishes it. We do — in two parts: are our calculations correct (checked by machine on every build), and what real people report about our predictions (live from votes, nothing invented).
1. Calculation checks (automatic)
Each test below re-runs on every build of the site. Our engine computes the same births again and compares them with the reference values: AstroSage (a widely used astrology software) and Swiss Ephemeris (the standard astronomical library used by astrology software). All of them match today. If any test fails, the build stops and the new code does not reach the site.
Kundli matching (36 gunas, 8 kootas)
✅ 119/119 couples matchTotal gunas compared for every couple; differences in the 8 individual koota scores: 0.
Lagna, planet signs, nakshatras, yogas & doshas
✅ 180/180 checks matchMahadasha/antardasha dates & Jaimini Chara dasha
✅ 589/589 checks matchLargest dasha-date difference: 6 day(s) (limit 7 days). Chara-dasha years of dual-lord signs (Scorpio/Aquarius) are not counted here — we follow the classical rule there, and 4 years differ from AstroSage.
Vimshottari dasha dates & birth balance
✅ 8/8 checks matchPlanet degrees & lagna (Lahiri ayanamsa)
✅ 10/10 charts matchLargest difference 20.4 arc-seconds (limit 30″, under 1/120 of a degree).
16 divisional charts, KP lords, Yogini dasha
✅ 65/65 checks matchAshtakavarga, D10 & advanced calculations
✅ 117/117 checks matchD7, D9, D10, Chara karakas & Chara dasha
✅ 59/59 checks matchConfidence-score rules (High/Medium/Low)
✅ 12,935/12,935 checks matchLast run: 16 Oct 2026 08:35 IST. The last row is an internal rule check (not a comparison with other software): it verifies on 100 random charts that the confidence score never breaks its own rules. Matching AstroSage shows our calculations are the same as established software — it does not prove that astrology predictions come true.
2. Prediction feedback (from real people)
When Pandit Ji (our AI) gives a time-bound prediction in the app, we ask you after that time has passed: did it happen? Only logged-in users can vote, one vote per prediction; votes are counted without names.
Loading live numbers…
"⏳ Not sure yet" answers are not counted here — that prediction is simply asked again 60 days later. Topic-wise numbers (career, marriage…) are not collected yet; we will show them here once votes carry the topic.
3. How we measure — and what we do not claim
- Astrology is not scientifically proven. We do not claim that predictions will come true. What we can show honestly is how correct our calculations are and what people actually experienced.
- Calculation correctness is measured against outside references (AstroSage, Swiss Ephemeris) — numbers on this page come straight from the test output, nothing typed by hand.
- Real experience is measured only from users' own ✅/❌ votes. Below 30 votes we show only the count, never a percentage.
What the confidence score means
Our "good time" windows carry a confidence label from a 0–100 score, built from five layers: promise in the birth chart, dasha, Jupiter–Saturn transit, Jaimini Chara dasha and Ashtakavarga.
"High" means a higher chance, not a guarantee.
How the rules will improve from real results
- Once enough votes come in, we will check which kinds of windows (which layers, which confidence level) actually came true more often.
- A rule that keeps turning out wrong gets less weight, or is re-checked against the classical texts — and every such change will be written on this page.
- For kundli matching, a quarterly re-check with fresh couples against AstroSage is set up (script ready).
- Right now votes are not yet linked to a topic or rule; adding that link is the next step. Until then, the feedback above is one overall number.
Found a calculation that looks wrong? Write to support@vedicdost.com — we check it, fix it and tell you what changed. Our promise →