Why We Refuse to Publish Lag-Offset Money Supply Charts

One chart circulates more than any other in Bitcoin commentary: Bitcoin's price laid over global money supply, with the money-supply line shifted forward by some number of weeks until the two appear to move together. It is the single most persuasive-looking chart in the asset class, and we will not publish it.

This is not a small methodological preference. It is the clearest example of a practice this whole site is organised against, so it is worth setting out exactly what is wrong with it.

What the chart does

The construction has three steps. Take a money-supply series — usually global M2, sometimes US M2. Shift it forward in time by a chosen lag. Plot it against Bitcoin's price on a second y-axis with an independently chosen scale. The result is two lines that track each other closely enough to look like a mechanism.

Every one of those steps involves a free parameter, and every free parameter is a knob that can be turned until the picture looks right.

The lag is the tell

If the lag reflected something real — the time it takes for new money to reach risk assets — publishers would converge on it. They have not. Widely circulated versions use offsets anywhere from ten weeks to fifteen and a half, and the number tends to be revised whenever the fit degrades.

A parameter that is re-tuned whenever it stops working is not measuring anything. It is absorbing error. And with a decade of data and a free lag parameter, finding an offset that produces visual agreement between two generally-rising series is close to guaranteed — you would find one between Bitcoin and almost any long upward-trending series if you were allowed to slide it around.

The second axis does the rest

The dual axis is the quieter half of the trick. Two series on independently scaled axes can be made to appear tightly coupled or completely unrelated purely by choosing the ranges, and nothing in the chart discloses that choice. A viewer sees two lines and reads the visual gap between them as meaningful, when the gap is a decision made by whoever drew it.

Put a series that has grown 93% since 2014 and a series that has grown 27,000% on the same linear dual-axis chart and you have to compress one of them enormously. The compression is exactly where the apparent correlation comes from.

What we do instead

Every comparison on this site follows the same three rules, stated on each page:

Both series are indexed to 100 at their common start, so the comparison is percentage growth and neither asset's absolute units do any work.
One shared logarithmic axis, so equal vertical distance is equal percentage change for both lines, and no scaling choice is available to flatter either.
No offset. Both series are plotted at the dates they actually occurred.

The result is less dramatic and more honest. Indexed from September 2014, US M2 reached 193 by the October 2025 peak while Bitcoin reached 27,280. The two lines do not track each other at all, and the chart makes no claim that they do. What it shows is a straightforward contrast: one supply is a policy decision that has moved almost in a straight line, the other is a market price that has risen enormously and fallen 77% twice along the way.

The claim we are not making

None of this says monetary expansion is irrelevant to Bitcoin's price. It probably is relevant. Money-supply growth and Bitcoin's price have both trended up for a decade, and it would be strange if liquidity conditions had no effect.

The point is narrower: a chart cannot establish that. Two series that rise together over the same period are consistent with a causal link, with a common cause, and with coincidence, and no amount of alignment distinguishes between them. Adding a fitted lag does not turn correlation into causation — it just makes the correlation look tighter than the raw data supports.

There is a related reason to be cautious about the global version specifically: the underlying data is patchy. Several countries' M2 series were discontinued years ago — the euro area's in 2017, Japan's in 2017, China's in 2019 — so a "global M2" line for recent years is reconstructed from sources that no longer report on a common basis. We chart US M2, which is live and consistently defined, and say so.

Why this is worth a whole page

Because the same instinct shows up everywhere else on this site, and we would rather state the rule once. The rainbow bands are drawn from a fit, so of course price has stayed inside them — we say that on the page. Pi Cycle Top's parameters were chosen because they fit past tops, which is curve fitting by definition, and the page lists the two cycle peaks it has since missed.

A chart that has been tuned until it agrees with the past tells you about the tuning. The useful ones are the ones that were allowed to disagree.

Last updated 2026-08-10.