⚠ AUDIT IN PROGRESS  ·  This firm's data is being independently verified across three tiers (automated · project lead · external reviewer) per SMU CGI audit framework. Not yet citable in external research until AUDIT-CLEARED.

Replication · open kit, three platforms, tolerance-based diff

Anyone can verify every number on this subsite.

Bottom line. The Coinbase replication kit ships the canonical Python pipeline (single source of truth for every number on the subsite); R (eventstudies) and Stata cross-check scripts; tolerance-based diff against expected_results.json. Tolerances are universal across the SMU CGI Reincorporation Tracker: ±0.5pp on point estimates, ±0.05 on p-values, ±0.01 on R², ±0.05 on donor weights. Kit status is pending per-firm event-study run; the file scaffolding ships with this build.

Read this first — what each file is for (when kit lands)
  • Daily-close panel — adjusted close prices for Coinbase and the crypto-exchange-industry donor pool (Robinhood, Block, PayPal, Marathon Digital, Riot Platforms, MicroStrategy, etc.), T−260 through T+10 around canonical event date. [KIT PENDING]
  • Donor weights — SLSQP-simplex optimization output; sum-to-1 at nine-decimal precision. [KIT PENDING]
  • Reference answers — expected_results.json for tolerance-based diff. [KIT PENDING]
  • Three pipeline scripts — Python (canonical), R, Stata. [KIT PENDING]
  • Walk-through README — installation, expected output, tolerance-diff checking. [KIT PENDING]
  • Data-derivation guide — rebuild the price panel from S&P Capital IQ, CRSP via WRDS, Ken French factor library. Inherits from canonical HOW_TO_PULL_DATA.md. [KIT PENDING]

Cross-platform verification tolerances

±0.5 pp
Point-estimate tolerance
±0.05
P-value tolerance
±0.01
R² tolerance
±0.05
Donor-weight tolerance

How to replicate (under 10 minutes, any one language)

  1. Download the kit from this folder once available. Files target ~ <500 KB total (excluding article PDF).
  2. Pick a language. Python is the canonical pipeline; R and Stata cross-checks ship in the same kit.
    • Python: pip install pandas numpy statsmodels scipy, then python event_study.py
    • R: install jsonlite, then Rscript event_study.R
    • Stata: open Stata, change directory to the kit folder, run do event_study.do
  3. Each script writes results.json and prints a tolerance-based diff against expected_results.json. Zero out-of-tolerance flags = successful replication.

Where the data comes from

  • Coinbase and auto-peer daily prices: S&P Capital IQ IQ_CLOSEPRICE_ADJ feed. Cross-check via CRSP (WRDS, academic-grade).
  • Factor returns: Ken French Data Library (FF5 + UMD).
  • Oil benchmark (sensitivity only): United States Brent Oil Fund (BNO) daily returns; WTI spot and Brent futures alternatives in the data-derivation guide.
  • 13F ownership snapshot: S&P Capital IQ Public Ownership Detailed, pre-event-date snapshot.
  • SEC filings: direct EDGAR accession URLs for every PRE 14A, DEF 14A, DEFA14A, 8-K cited.

Canonical primary sources

SEC EDGAR — Coinbase

  • Coinbase DEF 14A (April 29, 2024) — primary source verification pending (prior link removed: fake accession from build template; see canonical data.json for the firm's real event accession).
  • Coinbase 8-K Item 5.07 (June 14, 2024) — primary source verification pending (prior link removed: fake accession from build template; see canonical data.json for the firm's real event accession).
  • Coinbase 8-K (cover-page state change, December 15, 2025) — primary source verification pending (prior link removed: fake accession from build template; see canonical data.json for the firm's real event accession).

Daily closes, donor weights, expected results, and the Python / R / Stata scripts download directly from this page when the kit lands. The README walks through installation, expected output, and tolerance-based diff checking. Questions, corrections, or substantive discrepancies are welcome — please email Shane Goodwin (sgoodwin@smu.edu) at SMU Cox / SMU Dedman.