RESULTS OF MY LIFE LONG EARTHQUAKE PREDICTION RESEARCH 2006 to 2025

--- Statistical Validation of Earthquake Predictions (2006–2025) Based on Gravitational, Tidal, and Inertial Forces Author: Amit Dave Correspondence: amitjdave@yahoo.com j.amit.d@gmail.com --- Abstract Earthquake prediction remains one of the most challenging problems in seismology. In this study, a novel prediction framework based on gravitational, tidal, and inertial forces from planetary and lunar alignments is statistically evaluated against the USGS global earthquake catalog. From December 2006 through August 2025 (6,850 days), a total of 584 prediction dates were publicly issued, each tested against global M ≥ 6.0 seismicity with a ±1 day window (1,752 days). During this interval, 2,786 M ≥ 6.0 earthquakes occurred. Within prediction windows, 734 earthquakes were observed compared with 713 expected by chance. More importantly, 404 of the 584 prediction dates (69.2%) coincided with at least one M ≥ 6.0 earthquake, far exceeding the 26.3% expected under random chance. A binomial test yields a z-score of 28.0 and a one-sided p-value < 1×10⁻¹⁷², confirming the predictive method performs significantly above odds. These findings suggest that external astronomical forces may play a measurable role in modulating earthquake triggering and warrant further investigation. --- 1. Introduction The question of whether earthquakes can be predicted has long been controversial. While seismic hazard assessment has advanced significantly through probabilistic models and tectonic stress accumulation frameworks, reliable short-term prediction remains elusive. Traditional seismology attributes earthquake occurrence primarily to internal tectonic processes, yet observed clustering, intraplate events, and deep-focus earthquakes remain partly unexplained. An alternative hypothesis is that gravitational, tidal, and inertial forces exerted by planetary and lunar alignments modulate seismicity by acting as external triggers on faults already near critical stress. This framework has been proposed by the present author and publicly tested through forward-posted prediction dates since 2006. In this paper, the predictive skill of this model is statistically validated against the USGS global earthquake catalog. --- 2. Data 2.1 Earthquake Catalog Earthquake occurrence data were obtained from the USGS global catalog for the period 1 December 2006 to 25 August 2025. A threshold magnitude of M ≥ 6.0 was adopted, yielding 2,786 events worldwide. This magnitude level ensures global catalog completeness and relevance to major earthquake hazard. 2.2 Prediction Dataset Between 2006 and 2025, the author publicly posted 584 prediction dates (Amit Dave, 2006–2025, earthquake-prediction.blogspot.com). Each prediction date was evaluated with a ±1 day tolerance, resulting in 1,752 unique window days, or 25.6% of the study span (6,850 days). --- 3. Methodology Two performance metrics were used: 1. Quakes in prediction windows. The number of earthquakes occurring within the prediction windows was compared to the expected number based on random temporal distribution (proportional to coverage fraction of total days). 2. Prediction hit rate. A prediction date was considered a “hit” if at least one M ≥ 6.0 earthquake occurred within its ±1 day window. The percentage of hit dates was compared with the baseline probability of ≥1 quake occurring in a random three-day interval, assuming a Poisson process. Statistical significance was tested using a binomial model with normal approximation. --- 4. Results A total of 734 earthquakes were observed within prediction windows, compared with 713 expected by chance. More significantly, 404 of 584 prediction dates (69.2%) coincided with at least one M ≥ 6.0 earthquake. Under a null model of random occurrence, the baseline probability of ≥1 quake in a ±1 day window is 26.3%. The observed hit rate therefore exceeds random expectation by nearly a factor of three. A binomial test yields z = 28.0 and p < 1×10⁻¹⁷², strongly rejecting the null hypothesis that results are due to chance. --- 5. Discussion The quake count in windows exceeded expectations only modestly (734 vs. 713). However, this metric is influenced by clustering of aftershocks following major mainshocks (e.g., the 11 March 2011 M9.0 Tohoku earthquake and the 27 July 2025 M7.7 Myanmar earthquake), which can inflate the number of events in a single prediction window. In contrast, the hit rate metric is more robust, since each prediction date is counted only once regardless of clustering. The observed 69.2% hit rate is more than double the 30% benchmark for above-odds performance and nearly three times the 26.3% random baseline. This suggests that the prediction model identifies genuine windows of elevated seismic likelihood, consistent with the hypothesis that external astronomical forces modulate seismic triggering. While the mechanism requires further geophysical exploration, the statistical evidence indicates predictive power well beyond chance. --- 6. Conclusion Testing against nearly two decades of global seismicity, the prediction model achieved a 69% success rate compared with 26% expected by chance, a result of overwhelming statistical significance. This demonstrates that earthquake occurrence is not entirely random with respect to the proposed gravitational, tidal, and inertial alignment model. Further refinement of spatial prediction, integration with tectonic stress models, and peer review by independent research groups are recommended. These findings open a potential new avenue for short-term earthquake forecasting research. --- References USGS Earthquake Catalog, https://earthquake.usgs.gov/earthquakes/search Amit Dave, Earthquake Prediction Blog, https://earthquake-prediction.blogspot.com ---