What is Bitcoin’s scarcity worth?

We recently walked through two of the most common approaches to valuing bitcoin, both of which focused on potential demand.

In this piece and the next, I’ll look at supply, examining how the timing and magnitude of releases of bitcoin affect its price.

One theoretical framework for estimating bitcoin’s value is based on “stock-to-flow,” a metric borrowed from commodities analysts who use it to measure the scarcity of precious metals.

“Stock” refers to preexisting supply of a given asset, while “flow” describes how much of that asset can be produced in a given year. The higher the stock-to-flow ratio, the more difficult it is to meet existing demand with new supply.

Stock-To-Flow Formula

Bitcoin’s strictly limited supply has succeeded in creating real scarcity, and that has contributed to its performance. The stock-to-flow ratio allows us to measure the connection between the two, and that’s incredibly useful.

Although the ratio is useful, I remain unconvinced that it’s predictive. Stock-to-flow is helpful for understanding how bitcoin got to where it is, but not so much at explaining where it’s going. In this article I walk through the ins and outs of stock-to-flow, exploring what it might have to say about what makes bitcoin tick, and where it could improve.

Stock-to-Flow 101

Before we examine the case for stock-to-flow as a valuation model, it’s worth reviewing a few of the basics. Namely, how we define “stock” and “flow’ when it comes to bitcoin.

Let’s start with stock, which refers to an asset’s supply. For bitcoin, this is simple to measure. 19.3 million bitcoins have been mined and are in circulation right now. This figure will increase a little bit each day as new tokens get issued. That’s the numerator in our ratio.

Bitcoin's Stock Increases Gradually Over Time

Now let’s turn to our denominator: flow, or the rate that new bitcoins come online. Bitcoin’s supply is unique in that it’s dictated by precise instructions on how to print new tokens. No matter how much people may clamor for them, new bitcoins will come onto the market only when bitcoin’s code says so. It follows three cardinal rules:

• New tokens must automatically be issued every 10 minutes.
• The amount of tokens issued in each 10-minute window is prespecified.
• Every four years, the number of tokens issued during each 10-minute window is cut in half.

Unlike other commodities or precious metals, whose capacity can depend on the asset’s profitability, bitcoin’s rate of production is impossible for humans to influence. For the first four years after bitcoin was founded in 2008, its code issued 50 new bitcoins every 10 minutes. In 2012 the rate of issuance was halved to 25 bitcoins every 10 minutes. In 2016, the number of tokens dropped to 12.5. Most recently in 2020, issuance dropped to 6.25 bitcoins every 10 minutes. In 2024, that rate will be cut again to 3.125.

Here are two more things to keep in mind:

• By design, bitcoin’s stock-to-flow ratio will naturally rise over time. Our numerator, the stock, increases by a little bit each day as new tokens get released. Meanwhile, every four years the flow is cut in half, resulting in a smaller denominator.
• Stock-to-flow is an annual figure, which means it takes about 12 months for the ratio to fully absorb a change in flow.

Bitcoin’s code has halved its flow three times in the past 15 years: 2012, 2016, and 2020. (We only have reliable supply data for two of them: April 2016 and April 2020.) What we can see by charting stock-to-flow is the slope instantly kicks up after the flow cuts in 2016 and 2020, and that uptick persists for about a year as the stock-to-flow ratio absorbs the change. In between those two intense spikes, stock-to-flow trends flat.

Bitcoin's Stock-To-Flow Ratio Rises Sharply During Halving Cycles

Having covered the mechanics of stock-to-flow and how it applies to bitcoin, now we can evaluate it as a valuation model.

The Case for Stock-to-Flow

You can think of the stock-to-flow framework as a three-legged stool resting on three key assumptions:

• The stock-to-flow ratio drives price movements.
• This relationship is stable throughout time.
• It can therefore can be used to predict how much an asset will be worth.

The first leg of our stool—the assertion that stock-to-flow and bitcoin’s price are related—is structurally sound. The strongest support for it is empirical: Bitcoin’s stock-to-flow ratio has gone up significantly over the past 15 years, and so too has its price, producing an R-squared of about 0.94. We can see just how strong that relationship is when we overlay the rising stock-to-flow ratio with bitcoin’s price.

Bitcoin Price Rallies Coincide With Rising Stock-To-Flow Ratio

What’s driving the robust relationship? It appears that the most significant changes in the stock-to-flow ratio occurred during bitcoin’s strongest price rallies in 2016-17 and 2020-21. During these periods of climbing stock-to-flow, prices soared along with it.

Based on this, we can conclude that stock-to-flow appears to be correlated with bitcoin price changes, especially during the honeymoon periods in 2016-17 and 2020-21.

The Case Against

Now on to the second leg: The argument that the relationship between stock-to-flow and price is stable. This leg has some cracks in it: We can demonstrate that this argument doesn’t hold in practice.

This study was first published by an anonymous researcher in March 2019. Their model produced a slope constant (the “m” in y = mx + b, for the algebraically inclined) of 3.3. Now with four years of additional data, we can see that the original slope constant has failed to predict bitcoin’s subsequent returns.

Gap Between Projected and Actual Price Widens Out-of-Sample

In fact, it appears that the signal has eroded over time. We reran our own version of the model in 2023 with four more years of data to compare. Our iteration of the model produced a smaller coefficient and a weaker R-squared of 94% as opposed to 95%.

The third and final leg—the assertion stock-to-flow can be used to predict future returns—is still wobbly. It may be true someday, but current valuation methods need some tightening in order to get there.

The main problem is that bitcoin’s stock-to-flow ratio is always going to rise, even when prices don’t. Allow me to illustrate the logical implications. Analysts expect bitcoin will reduce its issuance rate in April or May of 2024, at which time the rate of new bitcoins printed every 10 minutes will drop from 6.25 to 3.125. Were we to plug those assumptions into our regression, it would forecast bitcoin hitting $31,500 around that time, $182,500 by May 2025, and $216,600 by May 1, 2028, which is around when bitcoin will reach the end of its next four-year cycle.

Stock-To-Flow Model Produces Unreasonably Lofty Valuations

Suffice it to say, that’s not a reasonable forecast.

Why doesn’t stock-to-flow better forecast bitcoin prices? My hunch is that both bitcoin and stock-to-flow are related to a third, causal variable that this model imperfectly captures. For example, it’s possible that bitcoin reacts more to issuance cuts than changes in the stock-to-flow ratio.

If bitcoin truly were sensitive to stock-to-flow, we’d expect the relationship between them to be just as strong during periods when stock-to-flow is pretty much constant as it is during honeymoon periods of 2016-17 and 2020-21. But that’s not the case, as shown below.

Stock-To-Flow Versus Price, 2014-16

Stock-To-Flow Versus Price, 2017-20

If we exclude the honeymoon periods from our regression, the R-squared of our formula plummets to 6.7%. That means our formula has virtually no explanatory power at all.

We need more data to figure out why this is. It could very well be that bitcoin is operating on four-year cycles, and we have good data for only two complete ones. Or it could just as easily be pure coincidence that bitcoin’s halving cycles arrived just as narratives around crypto reached their frothy peaks, and the relationship is totally spurious. For now, stock-to-flow only proves that there is a correlation between increasing scarcity and strong performance. We can’t yet say that it’s the cause.

In forecasting, patience is not only a virtue—it’s a prerequisite. Valuation models take a long time to mature, and that isn’t just lip service. Even if it’s true that scarcity is the key driver of bitcoin’s returns, valuation models built on that insight need refinement. Currently, they all have the same essential design flaw: They forecast a perpetually rising price. Future crypto research will have to eliminate this bias if stock-to-flow is ever going to have real merit.