Bitcoin's Gold Price: A Falsifiable Model Through 2035

(silmonbiggs.github.io)

1 points | by FightinJim 4 hours ago ago

1 comments

  • FightinJim 4 hours ago ago

    Author here. Quick context on what this is and why I'm posting it now.

    I fitted a saturating exponential (3 free parameters) to the BTC/Gold ratio using 10 years of monthly data (2015–2024). R² = 0.91. The basic idea: gold's supply grows ~2% annually; Bitcoin's is capped. Once you model the adoption surge as a transient that decays, you get a clean fit and specific predictions.

    The model has been running out-of-sample since January 2025 — 14 months. Yes, the ratio has dropped from 35 oz to 12 oz since July. The model is being stress-tested right now. The statistical scorecard for it is still looking good (cumulative z-score +0.93), because the early months ran above prediction and the bust is pulling the sum back toward zero. If the bust continues, the test will eventually reject the model — publicly, on the dashboard. That's the point.

    Paper is on SSRN, code is on GitHub (linked at top of page). Happy to discuss methodology, assumptions, or limitations.

    *(edited for typo)