Evidence / Data kit / Ethereum L1 regime

Ethereum L1 regimeCertified · HIGH confidence

Full pipeline to reproduce the Ethereum L1 regime classification (tau structural, pi demand) used by Invarians in the live panel. Runs end to end on public Google BigQuery data with a GCP free tier account.

Source

Dataset
bigquery-public-data.crypto_ethereum
Requirements
Google Cloud account, free tier sufficient for one extraction window.
Period
2020-06-01 to 2024-12-31 (adjustable in the query).
Integration window
Phi = 280 blocks, about 1 hour at 12 s per block.

Files

MD
README.md
Full reproduction instructions and expected matching criteria.
SQL
extract_eth.sql
BigQuery query, one row per Phi window with rho_ts (rhythm) and rho_s (structural load).
MD
expected_thresholds.md
The TPR, FPR, M1 and per-window state expectations to match.

Reproduction steps

  1. Open BigQuery console, paste extract_eth.sql, run.
  2. Export the result as CSV, save as eth_invariants_2020_2024_phi280.csv.
  3. Clone github.com/agentnorthstar/calibration, place the CSV in scripts/.
  4. Run python backtest_eth.py.
  5. Open the generated eth_backtest_results.csv.
  6. Confront the output with expected_thresholds.md in this folder.
Internal calibration parameters (nominal construction, divergence thresholds) are embedded in backtest_eth.py at runtime and are not exposed publicly. What reproducers confront is the output (TPR, FPR, M1, per-window states on the 4 canonical events), not the parameters. Any divergence beyond numerical noise is a published falsification, which is the intended use of this kit.