TLDR
Bitcoin’s Stock-to-Flow model should not be used to predict Bitcoin’s exact future price, but rather should be used to understand Bitcoin’s current price action and what narratives are truly driving demand. There’s incredible value in combining S2F with demand-side analysis to understand current price action and trends.
Stock-To-Flow
I’ve always been a fan of Bitcoin’s tokenomic model that hardcodes a future supply curve because if you fix supply then analyzing a supply-demand curve becomes all the more easy. In the same vein, I’ve also always been fascinated with Bitcoin’s stock-to-flow (S2F) model, popularized by pseudonymous analyst PlanB (https://planbtc.com/) because it attempts to take advantage of the simple supply-demand mechanisms (and their implied scarcity) in order to predict future price movements. PlanB’s model gained significant notoriety when it (more or less) predicted the average Bitcoin price of the 2019/2020 bear market and the subsequent rise in price in 2021. In recent months, the model has fallen out of favor due to the bear market, which many consider to invalidate the model.
Model Criticism
It’s not my intent to explain the model in this article (reference PlanB’s website above to learn more about it) but rather to use it as a starting point for a more meaningful analysis. I believe the S2F is one piece to a larger analysis that allows Bitcoin holders to understand the underlying demand for Bitcoin and whether the current price accurately represents that demand. Prior to that, I want to address the two main criticisms towards the model, because they dovetail well into my methodology.
Criticism 1 - The model relies too heavily on past price performance and poor statistical methodology. This criticism is especially true considering the S2F models were based heavily on early bitcoin price data, which suffered from more inefficient market conditions and thus is probably less reliable. This criticism is also bolstered because PlanB updated his statistical model later to show two more bullish models… the initial model had a price prediction of ~$55k for this time period, whereas the second model (which, in my opinion, had an inferior statistical approach) had a price prediction of $100k for this time period. A third model (S2FX) was based on other assets like silver and gold and predicted a current price of $288k.
Criticism 2 - The model doesn’t take into account current market conditions or demand drivers. The model assumes constant demand that outpaces the scarcity of supply, however its clear that black swan events (like FTX) and market conditions play a large role in demand and greatly impact the price of Bitcoin.
A Different Approach
The S2F has a lot of merit and value (that’s why it has remained popular for 4 years) however the criticisms above are accurate. But the criticisms above also reflect something else: the application of S2F (to predict future Bitcoin price) is the wrong application. I believe the true value of S2F is in a different application… and this new application is both more accurate and more meaningful if you own Bitcoin.
To understand this approach, you need to understand the Bayesian concept of “priors”. This is an intuitive concept, one you certainly already use every day but probably never stopped to think about. A prior is your default assumed probability of one event occurring just before another event occurs. For example, let’s assume your favorite sports team is playing their rival in three weeks and you assume your team has a 50% chance of beating its rival. However, between now and the rivalry game, your team plays a weaker team and loses badly. After that embarrassing loss, you now think your team only has a 25% chance to beat the rival. In this case, the 50% probability of beating the rival before the embarrassing loss is your prior.
Priors are useful concepts for understanding the impact of a specific event occurring. In the example above, if someone were to ask you what the impact of the embarrassing loss was on your impressions for the rivalry game, you could quantify the impact by saying it decreased your team’s odds in half. From a statistical point-of-view, priors are extremely useful for understanding the current situation and judging the effects of changes on that situation.
This is a really long-winded way of saying this important point: S2F should be thought of as one set of priors in the supply-demand-price dynamics of Bitcoin. When S2F is used in this context, as opposed to predicting future price action, it unlocks multiple analyses that allow you to statistically answer such questions as:
Did the recent bank runs / failures change the demand drivers for Bitcoin? Are people buying Bitcoin to “get off the traditional banking system”? Is the price rally that has occurred recently sustainable from the context of the narrative?
Is the major driver of Bitcoin demand the “digital gold” narrative or the “payments” narrative (for example, LN)?
How might additional central bank rate hikes or cuts affect Bitcoin demand and therefore price?
There’s additional sets of priors that I think are important and additional questions these priors can help answer, but I’ll get to those in a later post.
Supply-Demand Series
I will be posting multiple articles to help flesh out this approach to understanding Bitcoin’s supply-demand curve and hopefully will end up in a place of providing a model for understanding Bitcoin’s current price and whether it is likely to increase or decrease at any given time. S2F will be one component of that approach, however I will also need to develop additional ways to understand demand drivers and external shocks in order to establish answers to the questions above (and additional questions). If you are interested in learning more about approaching Bitcoin’s supply-demand-price dynamics from the Bayesian context of priors, and how it might adjust your thinking about Bitcoin, subscribe to this Substack and share it with your colleagues.
In the end, we all want to make it, Kyle