The scientific evidence for Seismikon's approach is the peer-reviewed research it is built on. To help people verify that the methodology works in practice, we run a public demonstration using California - a well-catalogued seismic region - and publish every prediction, evaluation rule, and performance result openly so scientists, journalists, and commercial partners can check the outputs for themselves.

📖

How It Works

Plain-language explanations of the core concepts behind Seismikon's predictions - Natural Time, EPS, LSTM AI, confidence scores, evaluation methodology, and the distinction between prediction and early warning.

Read the guide →
📊

California Demo Track Record

Forward-issued predictions from our California public demonstration, evaluated against independent FDSN seismic catalogue data. All matching is performed in your browser so you can verify the scoring yourself.

View track record →
🔬

Retrospective Research

The foundational Bikos (2024) methodology and retrospective back-test results on California seismic data spanning 2000–2023. Retrospective figures are clearly labelled and kept separate from live prospective results.

Read methodology →
📄

Research & Publications

The peer-reviewed Bikos (2024) paper, Dr. Anastasios N. Bikos's researcher profile, the full methodology description, and validation results from retrospective back-tests.

View research →
📋

Evaluation Protocol

The precise, frozen rules used to decide whether a prediction is a true positive or false positive - spatial tolerance, magnitude window, time tolerance, aftershock handling, and metric definitions. Versioned so any change is traceable.

Read protocol →
⬇️

Data Downloads

Download the full public prediction record as CSV or JSON, matched outcome data, current performance statistics, model version history, and protocol version archives.

Download data →
⚠️

Limitations & Open Questions

Known constraints of the current model, geographic and magnitude scope, calibration caveats, and honest answers to the scientific questions most frequently raised about probabilistic earthquake prediction.

Read limitations →