Stock-to-Flow: Strengths and Limits of the Model

Stock-to-flow measures scarcity: the existing stock divided by one year's new supply. As a metric it is useful, and for bitcoin it can be calculated to the block. As a price model, however, stock-to-flow has failed. It was briefly right in 2021, bitcoin has not closed a single day above the model price since April 2021, and in November 2026 the price stood at less than a tenth of it.

The idea: measuring scarcity as with gold

"Stock" is the existing supply, everything of a good that already exists. "Flow" is the inflow, what is newly added in a year. Divide the stock by the flow and you get a number in years: how long today's production would need to produce the whole stock once more.

For most commodities this number is small. Oil, copper or wheat are consumed; the stock is hardly larger than a year's production. Gold is different. Almost all the gold ever mined still exists: as bars, coins, jewellery. The mines add only a small share each year.

In his 2019 article PlanB put gold's stock-to-flow at about 62 and silver's at about 22. That means even if the mines doubled their output, the total stock would hardly change. This inertia is seen as the reason gold has served as a store of value for millennia. Saifedean Ammous made the argument popular for bitcoin in 2018 in "The Bitcoin Standard": a good store of value is one whose quantity nobody can increase quickly. More on this idea in What is good money?.

How to calculate bitcoin's stock-to-flow

For gold, stock and production have to be estimated. For bitcoin, both are in the code. Every new block brings a fixed reward, and it halves every 210’000 blocks, roughly every four years. At one block every ten minutes there are 144 blocks a day and 52’560 a year.

That lets you calculate the SF for each epoch between two halvings:

The pattern is simple: with every halving the flow halves while the stock barely grows. So the SF roughly doubles every four years. Within an epoch it rises only slightly, because the stock grows slowly.

PlanB's 2019 model

On 22 March 2019 the pseudonymous analyst PlanB published the article "Modeling Bitcoin's Value with Scarcity". For every month since 2009 he plotted the SF against the market value of all bitcoin, both on a logarithmic scale. The points lay close to a straight line. PlanB reported the relationship with an R² of 95 % and concluded: scarcity drives value.

For the period after the 2020 halving he gave a market value of about $1 trillion, about $55’000 per bitcoin. At the time a bitcoin cost about $4’000. So the model promised roughly 14 times the price in just over two years.

With the table above, the formula gives these model prices at the end of each epoch:

The basic problem is already visible in this series: the model knows only one direction and one pace. Another zero every four years, without end.

What stock-to-flow got right

To be fair: the model had a real moment. The chart shows the daily close since 2010 on a logarithmic scale, with the stock-to-flow line and, for comparison, the Power Law. The vertical line marks the day of publication.

• The direction was right. After the halving of May 2020, bitcoin rose from about $8’600 to over $60’000. PlanB's figure of about $55’000 was actually reached and exceeded in spring 2021: on 13 April 2021 bitcoin closed at $63’573, above that day's model price of about $53’800.
• The peak almost matched. On 8 November 2021 bitcoin closed at $67’525. The model stood at about $71’200 that day, only 5 % higher.
• The metric itself holds. Bitcoin's issuance is fixed, its SF rises predictably and today is above gold's. As a description of scarcity, stock-to-flow remains correct.

Where the model failed

After 2021, price and model parted ways. The clearest evidence is the twelve months after each halving. During that time the SF doubles, and each time the model rises about tenfold. The price did not:

Only once, in 2012, did the price rise more than the model, and back then bitcoin was a niche experiment with a market value of a few million dollars. After that it fell behind every time, most recently by a wide margin.

The consequences for the ratio of price to model:

• Bear market 2022: on 21 November 2022 bitcoin closed at $15’784, 0.21 times the model.
• All-time high 2025: on 6 October 2025 it closed at $124’673. The model stood at about $828’000; the high reached 15 % of it.
• Today: on 7 November 2026 the price was $75’868 and the model about $846’000. That is 0.09 times. The ratio hit its low on 30 June 2026 at 0.07.

Of the 2’787 days since publication, bitcoin closed above the model on 608, and all of them are before 18 April 2021. PlanB's extended version "S2FX" from 2020, which expected an average price of about $288’000 for 2020 to 2024, was never reached either.

How this looks next to the Power Law, day by day and with a fair test on data up to 2019, is in Power Law vs. Stock-to-Flow. This article is about why stock-to-flow was bound to fail.

The statistical criticism

Spurious correlation: when two curves simply both rise

Bitcoin's SF has risen almost steadily since 2009. Bitcoin's price has also risen strongly over the same time. Two quantities that both grow with time almost always lie on a straight line on a logarithmic axis, whether or not they have anything to do with each other. The economists Clive Granger and Paul Newbold showed as early as 1974 that such regressions produce high R² values even when there is no relationship. This is called a spurious regression.

One warning sign: the deviations from the model are not randomly scattered but stay on the same side for months. If the price is above the model today, it is very likely above it tomorrow too. The data up to 2019 show exactly this, very clearly.

The cross-check makes it tangible: any other steadily rising quantity fits about as well. A model that uses only the time since the genesis block explains the prices up to 2019 even slightly better. So the good fit up to 2019 does not prove that it is scarcity in particular that sets the price.

Demand is missing

A price comes from supply and demand. Stock-to-flow knows only supply. Whether more people want to buy bitcoin or fewer, whether interest rates rise or exchanges go bust, does not appear in the model. It therefore cannot explain a bear market in which the SF barely rises and the price falls by more than 75 %, as from November 2021 to November 2022.

On top of that, new supply in bitcoin has long been small compared with trading. Since 2024 about 450 bitcoin are created per day. On an ordinary trading day a multiple of that changes hands. Whether the price rises is decided mainly by the people who hold or sell the existing 20 million bitcoin.

The halving is no surprise

Everyone knows the date of each halving years in advance. If halving new issuance were certain to multiply the price tenfold, investors would buy beforehand and the effect would long be priced in. A known, scheduled event cannot keep triggering the same large jump. Whether there is a rhythm nonetheless is examined in The bitcoin four-year cycle.

Gold contradicts its own role model

Gold's SF has been almost unchanged for decades, yet its price swung strongly. If the same SF allows very different prices, it cannot set a price on its own.

The model runs towards infinity

With every halving the model price multiplies tenfold. Around 2140 the last bitcoin will be mined; then the flow is zero and the SF infinite, and so is the model price. As early as 2036 it would be about $89 million per bitcoin, a market value of almost $2 quadrillion. A model whose formula inevitably leads to the impossible does not describe the future.

What remains

Stock-to-flow is good as a metric and has failed as a price model. The metric describes something real: bitcoin's issuance is fixed, falls predictably and makes it ever scarcer. That is one reason it has any long-term value at all. Why the supply ends at 21 million is explained in Why there are only 21 million bitcoin.

For your planning this means: don't count on price targets that promise a tenfold rise per halving. Anyone who built partial sales, retirement or a loan on the stock-to-flow line in 2024 is planning today with eleven times the price. A model that lets growth slow over time and gives you a lower bound is better suited. How the common models compare at a glance is shown in Bitcoin price models at a glance.