Demand growth in Bitcoin: a quantitative model

“The credibility of Bitcoin as a store of value increases due to stabilizing demand growth”

The end result of the quantitave model for demand growth in Bitcoin

Abstract: Demand for Bitcoin has been growing wildly since its inception. At the same time, it seems to come and go. This large amounts of price booms and busts scare off large amounts of investors. The volatility seems to contradict the promise of being a store of value. This investigation attempts to explain and nuance this apparent contradiction. It connects logical explanations with quantifications for each of the different aspects of ‘demand growth’. By comparing demand growth over a 4 year timeframe with that of a 1 year timeframe, the demand during a hype cycle can be viewed separately from the long term demand growth. With this method, the long term demand growth becomes much clearer. The method is explained, where the separation of hype cycle related demand and long term demand growth is shown step-by-step. Finally, the stable demand growth underneath the (diminishing) influence of the 4 year halving-hype-cycle is re-combined in a price chart with an indication of the uncertainty range.

Introduction

If Bitcoin proponents would have to explain growth in demand, the main answer would resemble something like the following. Bitcoin has superior properties of an asset class with a predictable, fixed supply. It can’t be confiscated, is the only truly decentralized permissionless peer to peer digital asset. It has a maximum supply which can’t be changed. It can’t be devalued by third parties. It’s highly liquid, divisible, durable, verifiable and won’t go bankrupt. It’s a medium for storing and transferring wealth over time and space. As more people realize Bitcoin’s value proposition, network effects and hype cycles increase adoption, because Bitcoin solves problems and protects wealth throughout the entire world. (see link and link)

Although it sounds very attractive, some major concerns remain unresolved. One of the many concerns is the contradiction of high volatility versus the promise of a store of value.

On-chain data analysis

Because of the unique property of Bitcoin’s publicly available ledger ánd it’s predictable and hard-coded supply schedule, a unique dataset is available to study market dynamics and human economic behavior. In this study, the ledger data is used, as well as price data from coinmetrics and coindesk.

Introduction for Hodlers

Since Bitcoin‘s’ inception, people have been discussing as to what trajectory Bitcoin pricing will follow. Will its trajectory show diminishing returns. If so, when? Although the Stock-to-Flow model is practically unfalsifiable, it shows high predictive power up to now. Lots of projections are reasoning diminishing influence from the four year cycle. 1, 2 and 3. Also, diminishing returns have been prognosed.

This investigation attempts to quantify concepts of network effects and the four year hype cycle and combines it into one model.

Bitcoin’s Booms and Busts casting doubt on Bitcoin value proposition

Assuming acceptation of the above stated benefits, doubts could remain for newcomers. Is it really such a great innovation when price crashes are this fierce? So:

- Why is the growth in demand accompanied by booms and busts? What’s happening here? Certainly the dot.com boom-and-bust in 2001 was one real bubble, but why does Bitcoin exhibit so many price crashes? This is bad for adoption!

• Shouldn’t adoption follow the all-known smooth adoption curve?

Figure: supply & demand seems to be extremely volatile, undermining the promise of Bitcoin as a ‘store of value’

Good articles about Bitcoin volatility are available and show volatility trends, which was checked by the writer. In the figure below, it is shown in the red fit, that after 2013, volatility seems to have been stabilized. Conclusion of the article on volatility: The volatility “makes it prohibitively costly to use as a medium of exchange and a unit of account”. However, in the analysis, no separation of the 4 year hype cycle and long term demand growth was taken into account. All volatility data is analysed using short time spans.

Figure: analysis of ‘stabilizing’ volatility. Baur & Dimpfl

Figure: writers analysis

A Demand model to separate hype cycles from stable demand growth

After data analysis and an approximate fit on the historical data, the demand model can be constructed. Some details will be explained in the following sections:

• How is a stable demand growth trend and hype cycle demand separated;
• How is the function of a damped sine wave applied;
• What averaging is used.

Building the model

Step 1: Supply growth equals demand growth

Both supply and demand growth can be calculated by multiplying production (or flow) in ₿ with the price($) of that day. The Bitcoin price represents the equilibrium between supply and demand. One of the unique things about Bitcoin is that the production is constant, giving the possibility to quantify supply and demand growth in $ or in ₿.

In the analysis, the following assumption is made. The total available supply of about 18.6 Million Bitcoin is assumed to be absorbed by the underlying demand. The newly created supply being created is balanced with the growth (or shrinkage) in demand.

Step 2: Separating the halving cycle from demand growth

If the 4 year halving cycle is thát influential, what happens if we take it out of the equation? This was done by plotting supply & demand on a time axis, and calculating the demand growth over a time span of 4 years. Calculating it back to a 1-year demand growth yields the following graph. The orange line clearly shows a very stable rate of demand growth, especially after H2 ’15.

Figure: Annualized demand growth, based on 1 (white) and 4(orange) year time spans. An average of 20 weeks was taken.

A few things stand out in this way of visualizing demand growth:

• Only over a 4 year horizon, the average yearly demand increase is relatively stable. The reason for this can be found in a principle called ‘aliasing’. In short, a recurring signal(e.g. sine wave) is not visible, when the sample rate is lower than 2 times the frequency of the signal.
• The demand growth is roughly equal to 90%, or between 56% and 159%. The demand growth looks rather stable compared to the volatility induced by the 4 year cycle.
• Since 2015, the rate of demand growth has stabilized.
• Currently, there is no indication, yet, of diminishing demand growth.

Disregarding the 20 weeks average of the demand growth (production x price), the graph looks like this:

Figure: Annualized demand growth, based on 1 and 4 Year time spans. Weekly demand data was used, no averaging was applied.

On the background, the white striped graph shows the year-on-year (YoY) demand growth/shrinkage.

Step 3: Constructing a quantitative model of the hype cycle

Let’s investigate whether above white line has something to do the 4 year halving cycle. A separate plot of the YoY demand growth on a linear scale is shown below.

Figure: Assessing whether the halving(timing in green) was responsible for the hype cycle. Although n=2, it is a widely accepted and logical explanation.

A widely accepted preliminary conclusion can be drawn that the hype cycle is caused by the halving. Logically, this is very well explainable. An insightful explanation can be found here for it.

Next up is a model of the 4 year halving hype cycle on a linear scale.

“Damped sine waves are often used to model engineering situations where a harmonic oscillator is losing energy with each oscillation. For example: a bouncing tennis ball or a swinging clock pendulum.” (More info)

An adapted damped sine wave is used to model the halving cycle. Especially in the linear chart, the potential of this modeling is shown, although it is not flawless and no perfect match is depicted.

Figure: A model of the 4 year halving hype cycle, Exponential and linear axis.

For clarifying reasons, the model with ánd without subtracted stable demand growth are shown below, together with demand growth data of 1 years and 2 years.

Finally, we plot the resulting model together with the demand data graph.

The resulting price chart

With the demand model completed, the model can be used to plot the price chart. This ensures a proper validation of the model. The demand model data is divided by the supply and then simply plotted on a time axis.

Observations

• The obvious price influence of the halving is clearly visible. In the model, a halving of the supply simply causes the price to double. Moreover, the halving moment is clearly visible by the small jump in price.
• The underlying stable demand growth quantified at 90% price growth per year is clearly visible. The price increase because of this demand growth is more than sufficient to correct for fiat (M2) inflation by the central banks.
• Diminishing influence of the 4 year halving cycle is clearly distinguished. The price range appears to be reducing from 7.2x in ’13 to a much less volatile 3.2x for the coming halving.

Discussion

Probably 2 factors have great influence resulting in diminishing influence of the 4 year halving cycle: Increased size of the Bitcoin market capitalization. A quickly decreasing percentage of the total supply is produced by miners.

On a macro scale, this investigation is relevant. The relatively high volatility which is so detrimental for the narrative of a ‘store of value’ can be expected to subside within a few years. This in turn, will facilitate a new phase of adoption as it will be taken more seriously by large institutions, ensuring a continuation of the demand growth of about 90% per year.

Conclusion

• This investigation might help reduce skepticism about the volatility that comes with Bitcoin.
• It shows that it can be expected that Bitcoin’s value proposition as a store of value could become stronger.
• It explains the somewhat less astronomical returns after a halving.
• In short: it shows that Bitcoin as an asset class is maturing.

Disclaimers

• This investigation doesn’t intend build a statistical model. It merely attempts to clarify influences in the Bitcoin market.
• This is not financial advice.

History

A first model referencing to demand growth trends is from November 2019.