For two and a half thousand years, scholars across India, Persia, and the Arab world tracked planetary cycles and correlated them with earthly events. The question this project asks is simple: when you strip away the mysticism and apply modern statistical methods, does anything in those frameworks actually survive? This paper introduces Tempora Research — our methodology, our findings to date, and what we are building. The short answer is that most of it does not survive. But some of it does, in ways that are genuinely surprising, reproducible, and falsifiable. Individual signals reach up to 5.46 times their expected frequency on confirmed historical event dates. We have published nine research notes. We have 72 public, dated predictions through 2030. We keep score on everything.
Astrology has two problems. The first is that most of its practitioners make it impossible to falsify — predictions are vague, retroactively interpreted, and never publicly scored. The second is that most of its critics dismiss it without testing it — defaulting to prior probability rather than evidence.
Neither position is intellectually satisfying. The question — does any astronomical cycle correlate non-randomly with historical events when tested against data? — is an empirical one. It deserves an empirical answer.
This is what Tempora Research does. We take the frameworks seriously enough to test them properly: precise astronomical calculations, documented historical events, Monte Carlo baselines, statistical significance testing, and forward predictions that are dated, specific, and publicly tracked. If the model fails, you will see it in the numbers. That is the point.
We are not claiming that planets cause events, that astrology is spiritually "real," or that any individual's fate is written in the sky. We are asking a narrow empirical question: do certain measurable astronomical cycles correlate with historical event clusters at rates that exceed chance? The answer, for specific signals in specific charts, appears to be yes. We report that finding with appropriate uncertainty and invite challenge.
Vedic astronomical traditions — developed in India over roughly 2,500 years, refined by mathematicians whose orbital calculations remained accurate to within minutes of arc — produced two primary timing tools that we test.
The first is a planetary period system: a 120-year cycle divided into major and minor periods, each associated with a specific planet, determined by the position of the Moon at the moment of a nation's founding. This system assigns each entity — a country, a government, a person — a current operating "period" that cycles through all planets over time. Our research tests whether the nature of the ruling period correlates with the character of events during that period.
The second is a transit overlay system: the current positions of planets tracked against the fixed positions they held at the moment of founding. When a planet in motion crosses a sensitive point in the founding chart, the system predicts elevated activity. We test whether these transit crossings correlate with documented historical events at rates above the random baseline.
Both systems are computed with sub-arc-minute precision using Swiss Ephemeris — the same computational core used by professional observatories. There is no interpretation, no intuition, no hand-waving. The calculations are deterministic and fully reproducible.
We built a scoring engine that computes nine specific signals for any chart on any date, produces a single confluence score between 0% and 100%, and can be run forward or backward across any time period. We then ran it against history.
Six national founding charts: India (independence, 1947), United States (independence, 1776), Russia (post-Soviet founding, 1991), China (People's Republic, 1949), United Kingdom (Act of Union, 1801), Pakistan (independence, 1947).
39 confirmed historical events with documented dates: assassinations, elections, wars, financial crises, natural disasters, constitutional changes. For each event, we scored the founding chart against the astronomical conditions on that date.
300 random baseline dates per country — 1,800 random scores in total — to establish what the scoring engine produces by chance. This is the Monte Carlo baseline: the null hypothesis that the engine scores event dates no differently than random ones.
The test: do historical event dates score higher than random dates? If not — if the engine cannot distinguish a date when something happened from a date when nothing happened — the framework has no predictive power and the project stops here.
The aggregate result was disappointing. Across all countries using generic, uniform scoring, the average lift was 0.82x — meaning the engine actually scored historical events slightly worse than random dates. Generic scoring fails.
But this masked something important. When we stopped applying uniform weights across all signals and instead calibrated each country's chart against its own historical record — asking "which specific signals actually fire on this chart's event dates?" — the results changed dramatically.
The key insight: every founding chart has a unique sensitive axis. India's chart is dominated by the tension between its Cancer configuration and its Capricorn opposition. Russia's chart is acutely sensitive to a specific Mars-nodal conjunction. The US chart responds most strongly to Saturn's transit of its natal Sun position. Averaging these different sensitivities together produces noise. Isolating them produces signal.
After replacing generic weights with chart-specific calibrated weights derived from historical data, lift ratios improved significantly across all six countries:
| Country | Events Tested | Generic Lift | Calibrated Lift | Improvement |
|---|---|---|---|---|
| Russia | 4 | 1.15x | 3.12x | +171% |
| United States | 8 | 0.91x | 2.34x | +157% |
| UK | 4 | 0.70x | 1.89x | +170% |
| India | 15 | 0.92x | 1.71x | +86% |
| Pakistan | 4 | 0.56x | 1.62x | +189% |
| China | 4 | 0.65x | 1.44x | +121% |
These are not large sample sizes. We say this explicitly. Russia's 5.46x lift derives from four historical events — statistically suggestive, not definitive. The US has eight events and a more robust result. India has fifteen events and the strongest calibration in the dataset. We are honest about what the numbers can and cannot support.
What they can support: the hypothesis that chart-specific astronomical signals correlate with historical event clusters at rates meaningfully above chance is worth taking seriously and investigating further. That is the claim. No more, no less.
Backtesting has a fundamental problem: when you calibrate a model against historical data, you can always find patterns after the fact. The real test of any predictive framework is its forward performance — predictions made before the event, publicly recorded, with specific dates and falsifiable criteria.
This is what Tempora's prediction register does. Using the calibrated engine, we have generated 72 specific, dated predictions across all countries we track, running through 2030. Each prediction has:
The most significant near-term predictions from the calibrated engine:
By 2030, we will have a meaningful forward performance record across 72 predictions in 9+ countries. If the model's calibrated signals have genuine predictive power, the hit rate should exceed the 35% random baseline substantially. If it does not, we will say so, publish the analysis, and revise the framework accordingly.
India's independence chart mapped against 78 years of history. 21 dated, falsifiable predictions across politics, economy, and security.
Two independent founding charts overlaid on 30 years of Nifty 50 data. Every major crash and rally identified. Sector calls for 2026–2028.
Our first multi-country analysis. 92 historical events backtested. 15 forward predictions. The 2028–2029 global convergence identified.
The methodology paper. How we derived empirically validated weights from historical data. Why generic scoring fails. Full calibration results per country.
The calibrated engine applied forward. 12 specific prediction windows. The global convergence of 2028–2029 mapped in detail.
Three papers applying multi-chart convergence analysis to the world's most significant geopolitical fault lines. Stability corridors, fighting windows, and destruction windows mapped through 2030.
The research is the foundation. The product is Kaala — a temporal intelligence engine that makes these signals accessible in real time.
For individuals: your personal chart analysed against the same engine, showing your current operating period, active signals, and upcoming windows. For institutions: country-level prediction windows, confluence scores, and API access for analysts who want to incorporate temporal risk signals into geopolitical and investment frameworks.
The research is open. The code is open. The predictions are public. The scoring is transparent. If you find a flaw in the methodology, we want to know. If the predictions fail, we will publish that too.
That is what separates research from fortune-telling.
All papers are available at tempora.ltd. The full codebase is on GitHub. The prediction register is publicly tracked. If you are a statistician, a geopolitical analyst, an astronomer, or simply a rigorous sceptic — we are specifically interested in your challenges. The framework improves when it is tested. That is the entire point.