Love
Relationship timing
59 public-figure relationship rows across 46 scored events: 31 union rows and 28 break rows. Venus/Jupiter union leads and Saturn/Mars break leads remain unverified.
Screening only59 rows
Astral EdgeEntertainment · Reflection · Not prediction
Know your sky
before it moves.
Astral Edge reads classical astrological timing against your birth chart and shows you, hour by hour, when the traditional markers stack supportively or challengingly across love, career, money, and high-amplitude life windows. A reflective tool grounded in tradition with a public research audit that now includes lottery windfall backtests, asteroid/Lilith timing signals, and many claims that did not hold up.
Rendering the sky…
Uranus in TaurusMay 2018 – Apr 2026
Get the app →
Free to start. $2.99/month for premium timing.
How it works
We tested 53 astrological hypotheses.
One contrarian signal replicated.
Most claims did not.
Most astrology apps weight techniques by tradition or gut feel. We built a cohort of 596 documented peak events and 450 setback controls, pre-registered 53 astrological hypotheses, ran 10,000-permutation label-shuffle tests with a subject-level discovery / held-out split, and corrected for multiple comparisons.
Only H34 replicated out of sample, and it replicated in a cautionary direction:Thursday / Friday "benefic weekday" lore looked less favorable for peak outcomes and more common in setback rows. The production scorer stays a traditional-astrology-informed timing heuristic — a structured read of classical techniques — not an empirically validated predictor. Use it as a reflective tool, not as a forecast. Full null-result writeup is linked on the methodology page.
596
Peak events
450
Setback controls
53
Hypotheses tested
1
Held-out replications
Cross-validated classification AUC on the archived fitted model was 0.505 — chance. We ship the scorer as a classical-astrology heuristic, with the honest label on the tin.
Recent research lanes
We're not only testing generic "luck."
We split it into love, loss, windfall, and forensic stress-tests.
The first rebuild narrowed the claims dramatically, so the newer research program gets more specific: relationship events, mortality and setback case studies, lottery/draw timing, and retrospective forensic control tests each get their own controls. None of this is framed as advice or fate. It's a public audit trail for what the app should and should not claim.
Death / loss
Setback and mortality checks
73 deceased-public-figure rows, 10 mortality hypotheses, and 4,394 generated mining features. No BY-corrected result verified; never for living-person prediction.
Safety-gated0 BY passes
Lottery
Windfall and draw structure
Lottery work separates personal windfall timing from draw-shape research. The active model is research-anchored-v4-N593; H108 Jupiter-Apophis and H109 Jupiter-Lilith now feed the evidence layer.
Fixed-budget only0.700 mean rank
Forensic
Serial-killer controls
Retrospective public/deceased-only tests: N=60 natal charts, 59 dated attacks, and 83 expanded transit events. Scorpio, mutable, Mars-Pluto, full-moon, and spree-peak memes did not survive.
Research only0 screening use
Product translation: the mobile app can show love, lottery/money, career, and setback/loss journaling lanes, but the language stays reflective. The serial-killer work exists only as a retrospective stress-test for astrology claims: it must never be used to score, screen, or predict behavior for living people. It should help users log outcomes and compare timing against reality, not tell them who to love, when someone dies, or that a ticket will win.
New lottery research post
Apophis and Lilith are now in the windfall evidence model.
The update publishes the H108 Jupiter-to-Apophis result (14.4% observed vs 4.5% null, p=0.0035), the H109 Jupiter-to-Lilith result (13.3% observed vs 5.3% null, p=0.0185), and the calibrated active-model backtest: 0.700 mean same-day percentile across lottery-winner rows.
Boundary kept intact: no BY survivors at q=0.10, so these remain evidence-weighted timing hypotheses, not proof or gambling advice.
Read the model update →New forensic research post
Serial-killer charts did not support the easy true-crime astrology memes.
The public writeup posts the numbers: no Scorpio or mutable-sign overrepresentation, no Mars-Pluto natal lift, no full-moon or Mercury-retrograde spree trigger, and no within-spree day-level FDR passes. The only surviving signals are narrow retrospective transit/asteroid findings that need independent replication.
Read the findings →Who we studied
Verified birth times, at the moments lives turned — in both directions.
Peak events: Nobel laureates, Oscar winners, Olympic golds, Grand Slam finals, Billboard #1 singles, championship athletes, documented lottery jackpots. Setback controls: convictions, resignations, bankruptcies, career-ending scandals. Rodden ratings range from AA (birth-certificate) down to noon-proxy where no birth time is publicly available. Peaks skew more timed than setbacks — a real asymmetry we don't hide.
We pre-registered 53 hypotheses and ran the rigorous test the data deserves. The result is a null: no hypothesis replicated across a clean discovery / held-out split at FDR-corrected significance. The scorer ships as a classical-astrology heuristic with that honest label — not as a validated predictor.
What we falsified
Popular astrology memes that didn't hold up in our cohort.
These claims people repeat confidently showed no verified support in the current pipeline. If the app ever pushes them as proven, call us out.
- Mercury retrogradeSignature stressor periodPopular meme; empirically null in the shipped cohort
- Via CombustaInauspicious arc of the zodiacNo verified held-out signal
- CazimiStrongest amplifier in traditionBorderline-negative, not product-grade
Fixed-star luck conjunctions, waxing-moon favorability, Chaldean decan rulerships, Jupiter/Uranus lottery midpoint lore, and several natal speculation-house claims also failed to reach product-grade evidence. The full list lives in the methodology notes.
Honest answers
Questions you're probably asking
Does this actually work?
Not as a validated predictor, and we don't claim that it is. We tested 53 astrological hypotheses with permutation tests and a subject-level discovery/held-out split. Only one signal replicated, and it was contrarian: popular benefic weekdays behaved more like caution flags than lucky proof. Classification AUC on the archived fitted model was 0.505 — chance. The scorer is a classical-astrology-informed timing heuristic, not a forecast.
So why ship it at all?
Because classical astrology has been scored by hand for two thousand years, long before anyone had the data to validate it. Nearly every astrology app you've used is scoring a theory-anchored tradition. We make that visible, show the caution flags, and let you journal whether the timing feels useful in real life.
How do I know you're not just hiding the bad results?
The opposite — we publish the uncomfortable parts. Earlier versions of this page overstated survivors. After an audit, we rebuilt the analysis with a subject-level split: five discovery leads failed confirmation and only one contrarian caution signal replicated. Transparency cuts both ways.
Is this financial advice?
No. Astral Edge is for reflection and entertainment. Nothing in the app should be interpreted as financial, medical, or relationship advice. Decisions about money, health, or other people are yours.
What does it cost?
Free to try. $2.99/month unlocks premium timing windows, push alerts, and the full calendar. Cancel anytime in your App Store or Google Play subscription settings.
Your sky is already moving.
Get the app. Find out when the day turns in your favor.
Coming soonApp StoreComing soonGoogle Play
Coming to the App Store and Google Play. Enter your email to know first: support@astraledge.app