Does gold lead Bitcoin?

What does each study find about gold and Bitcoin?

Three answers about whether gold leads Bitcoin, from three credible sources, including a change of direction between two pieces by the same author eight months apart.

André Dragosch, head of research at Bitwise Europe, is reported by CryptoSlate on 14 January 2026 as finding through Granger causality tests that gold tends to lead Bitcoin by approximately four to seven months, and not the other way round. That is the cleanest supporting result in the record, and it reaches this page second-hand: the figure comes from that write-up rather than from a paper BlockPhi has read. It is a model estimate and past results do not indicate future results.

Michael Howell's vector autoregression, published 8 October 2025, runs the other way at short horizons. A positive shock to gold prices is followed by a fall in Bitcoin prices around four to six weeks later, before the long-run positive relationship reasserts itself. That is a model estimate as of 8 October 2025 and past results do not indicate future results.

Howell's later piece, published 28 June 2026, describes Bitcoin leading gold rather than following it: Bitcoin peaked on 6 October 2025 and gold's peak came later, around early March 2026, and he attributes the two moves to different central banks. Same author, same index, different conclusion about which asset moved first. That is a model estimate as of 28 June 2026 and past results do not indicate future results.

The peer-reviewed treatment finds less than either. A 2026 paper in the Journal of Risk and Financial Management, testing daily data from 2013 to 2025 across quantiles and frequencies, reports that full-sample directional dominance between Bitcoin and gold is generally weak once multiple testing is accounted for. That result is partially verified: the journal page returned 403 to automated fetch and the abstract was read from a search summary.

Why do the gold results disagree?

Different frequencies, different horizons, different tests. A monthly Granger test and a weekly impulse response are not asking the same question.

A Granger test on monthly data asks whether past gold moves help explain future Bitcoin moves over months. An impulse response from a vector autoregression asks what happens to Bitcoin in the weeks after a shock to gold. A quantile causality test with a false-discovery-rate correction asks whether any of the apparent relationships survive the fact that many were tested at once. Those three questions can honestly produce three different answers on the same data.

There is also a common cause underneath both assets. Gold and Bitcoin both respond to the liquidity cycle, and two assets driven by a shared third factor produce lead-lag results that move with the sample rather than holding still.

What does BlockPhi publish about gold and Bitcoin?

BlockPhi publishes a rule-based system named Gold Leads BTC, and no gold lead time. The record does not support publishing one as settled.

BlockPhi runs a rule-based system named Gold Leads BTC. The name states the premise the system tests. It does not state a result, no lead time for gold is published on this domain, and none should be inferred from the name.

That is a deliberate asymmetry with how these pages treat liquidity. Where the published record clusters, as it does for the liquidity relationship, a figure with its measure and its date beside it can be published and defended. Where the published record contains a change of direction between two pieces by the same author, it cannot.