Historical Predictions
Overview
This project evaluates the accuracy of 39 famous astrological predictions made between 1555 and 2022. The dataset includes a mix of Mundane (world events), Financial, and Electional predictions to determine if historical success rates exceed random chance.
Data Source
A curated list of well-documented predictions from astrologers such as:
- Nostradamus
- William Lilly
- Evangeline Adams
- Jeane Dixon
- Modern Financial Astrologers (Raymond Merriman, Arch Crawford)
Methodology
- Success Criteria: Rated based on historical consensus of the outcome.
- Statistical Test: Binomial Test (Success Rate vs 50% chance).
- Segmentation: Analysis by Category and Time Horizon.
Results
- Overall Accuracy: 56.4% (22 Correct, 17 Incorrect)
- Significance: P = 0.2612 (Not significant). The results are indistinguishable from random chance.
Performance by Category
- Financial: 80% (Small sample, selection bias likely)
- Mundane: 60%
- Electional: 25% (Significantly worse than chance - likely due to strong partisanship bias in modern predictions)
Performance by Time Horizon
- Short Term (<2 years): 45% Accuracy
- Long Term (>10 years): 85% Accuracy
- Note: This high long-term accuracy likely reflects "Survivorship Bias" — we only remember 400-year-old predictions if they came true (like Nostradamus on the French Revolution).
Files
predictions_data.csv: The dataset.analysis.py: Python processing script.RESULTS.md: Detailed statistical report.accuracy_by_category.png: Visualization of success rates.horizon_vs_outcome.png: Scatter plot of Time vs Success.
Data Provenance
Historical Text
- Source: Project Gutenberg / Internet Archive.
- Content: Late Medieval/Renaissance Almanacs.