Bitcoin Price Temperature (Bands)

An indicator for the price bandwidth of Bitcoin’s 4-year cycle

Dilution-proof
Dec 15, 2020 · 11 min read

This article follows up on a previous one in which a metric was introduced that is being rebranded and expanded upon as the Bitcoin Price Temperature (BPT). The BPT metric will first be described in a more detailed manner, including the rationale behind the proposed causal mechanism of its 4-year market cycles, the interpretation of the BPT metric and its limitations. Thereafter, the BPT Bands concept is introduced and supported by several charts in which the indicator and its potential application are visualized.

Bitcoin’s 4-year cycle

When Bitcoin was launched, early network participants were rewarded 50 BTC for each new block that was created. Every 210.000 blocks, this ‘block reward’ is halved, gradually slowing down the supply issuance. Due to this built-in disinflationary monetary policy, Bitcoin has a strictly controlled and finite supply. After 33 halvings (~2140), the block reward will become lower than the smallest unit in the system (1 satoshi or ‘sat’, which is 0.00000001 BTC), topping off Bitcoin’s supply at a total of 20,999,999.97690000 BTC.

To ensure that blocks are created roughly every 10 minutes so that Bitcoin’s transaction capacity and supply issuance are relatively stable, a difficulty adjustment mechanism was built in. Every 2016 blocks (~2 weeks), the Bitcoin software checks to what extent new blocks have been created every ~10 minutes. If the average block time is lower, it increases the complexity of the random number that miners are trying to guess in order to win the rights to create the next block. Likewise, if the average block time is more than 10 minutes, the difficulty is adjusted downwards, making it easier for miners to create new blocks. If blocks are mined exactly every 10 minutes, the duration of one halving cycle is 4 years (210.000 blocks *10 minutes per block = 2.100.000 minutes, 2.100.000/60/24/7/52 = 4.00641 years). In practice, the average halving cycle duration was 3.8 years (1381 days) so far, as a result of the continuously growing network capacity (‘hash rate’).

A side effect of these halvings is that once every ~4 years, a supply shock is introduced to the market, abruptly lowering the amount of new supply that becomes available via mining. If the net demand for Bitcoin stays the same (e.g., people periodically buying Bitcoin as a savings vehicle or investment), this abrupt decrease in newly minted Bitcoin means that the only other way to acquire Bitcoin is by buying them from current holders. Since holding Bitcoin as a long-term store of value (also known as ‘hodling’) is such a popular theme under market participants, those Bitcoin may only become available at increased prices. The price increase on its turn leads to increased awareness that is typically accompanied with an increase in demand, throwing oil on the fire and creating manic market circumstances that lead to overheated prices (a feedback-loop known as ‘reflexivity’), often followed by a rapid decline and cool-down period. These mechanics have been eloquently and graphically described by @Croesus_BTC in this Twitter thread.

The chart of the Bitcoin price over the past ~10 years (figure 1) shows how the halvings so far were indeed always followed-up by exponential price growth and a subsequent cool-down period. The white line illustrates the daily price, whereas the black line depicts its moving average (using the to-date-available data the first four years), which is the mean price during a 4-year window. The 4-year moving average price is continuously going up, showing that on a four year time-frame, dollar cost averaging into Bitcoin has historically been beneficial at any time.

Figure 1: A logarithmic chart with the Bitcoin price in USD and its 4-year Moving Average

It is impossible to formally conclude that the halvings caused these overheated market conditions based on a sample of just 2 observations. However, the combination of the clear rationale for the potential causal mechanism and the matching price action make a compelling case that there might be more truth to this hypothesis than can currently be proven.

The most well-known attempt to more formally test the hypothesis that Bitcoin’s market value can be modeled based on its halving-induced scarcity is the Stock-to-Flow (S2F) model. In a previous article, I summarized all developments in the S2F and later introduced S2F cross-asset (S2FX) models, as well as their most prominent critiques. The models provide an interesting outlook for Bitcoin’s future price if the hypothesis that Bitcoin’s exponential growth is indeed a function of its gradually increasing scarcity. However, problems with the assumptions made by the statistical tests that were used in the S2F model analyses and the small sample size of the S2FX model prevented the emergence of broad consensus about the outcomes.

In the absence of such irrefutable proof, the predictions of these models are accompanied with a relatively broad uncertainty (margins). Therefore, there is still a need for indicators that more flexibly reflect to what extent current Bitcoin market prices are (ab)normal in the context of its own price history and volatility, particularly related to the 4-year market cycles. The Bitcoin Price Temperature (BPT) metric does exactly this.

Bitcoin Price Temperature (BPT)

The Bitcoin Price Temperature (BPT) is a measure for the distance between the current Bitcoin price and its 4-year moving average. The BPT is calculated by first calculating the difference between the daily price and its 4-year moving average, and then dividing that number by the standard deviation of that 4-year window (using the to-date-available data during the first four years). In R, the BPT can therefore calculated using the following formula:

btp[i] = (price[i] — mean(price[ifelse(i<1460, 0, (i-1460)):i]) / sd(price[ifelse(i<1460, 0, (i-1460)):i]

The BPT metric therefore reflects the number of standard deviations that a point deviates from the mean, which can technically be called a ‘Z-Score’ and is a common standardization method in multiple scientific disciplines.

Since the 4-year moving average represents the ‘normal’ price during a four year window, the BPT metric therefore reflects how (ab)normal the current price is in the context of its own 4-year price history. The BPT metric can therefore be seen as a temperature-check, where higher values represent potentially (over)heated price levels, and lower or even negative values suggest that those prices are relatively low based on a 4-year window. Figure 2 illustrates the BPT over time.

Figure 2: The Bitcoin Price Temperature (BPT)

In figure 2, the blue (BPT=0) line represents the 4-year moving average. As is evident after comparing figures 1 and 2, the BPT makes it much easier to compare to what extent the different market cycles were similar than on the regular price chart. One of the more interesting findings is that the four most prominent market cycle tops (2011, 2x 2013 and 2017) peaked out soon after the Bitcoin price hit a temperature of 8 (red line), temporarily overshooting to temperatures of up to 12 before starting their steep descend back to ‘cool down’ all the way back to 0–1 . The orange (BPT=6) and green (BPT=2) lines also represent key price temperatures where the price trend changed its course on multiple occasions. It is good to note that there are no statistical reasons why these levels were highlighted in the figure above, but these were identified based on technical analysis, which has its limitations.

The similarity between the relative price action of the cycles becomes even more apparent in figure 3. The first period of Bitcoin’s existence (blue line) was special because the existing supply was relatively heavily diluted on a daily basis and Bitcoin didn’t have a formal market price in the used data (by Coinmetrics) during the first 561 days. Nonetheless, this period had its own manic market cycle and blow-off period right after Bitcoin received a market price, a pattern that so far has been repeating during each halving cycle.

Figure 3: A comparison of the Bitcoin Price Temperature (BPT) of each halving cycle

It is important to note that the BPT is solely a backwards-looking metric and does not hold predictive powers. The visual representation of (the similarity of) the 4-year market cycles may make a compelling case that we are witnessing cyclically repeating market cycles here. However, there are no guarantees that these cycles will repeat, nor that they will be exactly similar to the previous cycles if the cyclicality itself does continue and future market tops reach similar relative price levels. Particularly the degree of euphoria during the manic part of the market psychology cycle is hard (if not impossible) to predict, since there are many more variables at play other than the time and halvings that are depicted here.

A fundamental argument against the hypothesis that these cycles are to repeat over-and-over again in the future, can be based on the notion that while the relative impact of each halving is the same (block reward / 2), the absolute impact of each halving on the block reward is decreasing (4-annual inflation rate dropping by -50%, -33.3%, -9.6%, -3.8%, -1.7%, -0.8%, etc.). Therefore, the impact of future halvings on the described supply-side liquidity shocks may gradually decrease. It is still possible that other factors (e.g., policy changes related to elections or cycles in traditional financial markets) may cyclically influence Bitcoin’s market cycle, but that those cycles don’t necessarily follow the same 4-year periods that we saw before.

However, the relative price action of the current 220-day old halving cycle (2020~2024) again shows remarkable similarities to the previous one (2016–2020). Figure 4 depicts the correlation between the BPT of these two cycles over time, where the color-overlay represents the statistical significance (green = statistically significant at p<0.05, red = not). Using the to-date-available data of the current and previous cycle, there is a high (r=0.77) correlation between the relative price movements of these two cycles.

Figure 4: The correlation between the BPT values of the current (2020~2024) and previous (2016–2020) halving cycle

It is impossible to proof that history will indeed repeat itself. Nonetheless, many Bitcoin market participants appear to have built some conviction in the assumption that the repetitive nature of these market cycles will continue in the foreseeable future. Therefore, using the BPT to monitor these relative price movements, as well as the expected price levels if the BPT were indeed to reach the key BPT levels of prior cycles again, may be useful. This brings us to the new concept of BPT Bands.

Bitcoin Price Temperature (BPT) Bands

Since the BPT values reflect the ‘temperature’ of the Bitcoin price, the metric can be useful as a color-overlay on the regular Bitcoin price chart. Additionally, it is possible to display the key levels that were identified above (or any other BPT level) on that same price chart, by simply multiplying the standard deviation by the BPT level (so 2*SD for BPT=2, 6*SD for BPT=6, etc.) and adding this to the 4-year moving average. The result is displayed in figure 5.

Figure 5: The Bitcoin Price Temperature (BPT) Bands for BPT=0 (blue), BPT=2 (green), BPT=6 (orange) and BPT=8 (red)

This BPT Bands visualization allows for several interesting observations:

  • While the regular BPT chart was convenient for identifying potential price support and/or resistance levels (the key levels identified as the colored horizontal lines in figure 2), displaying them on the price chart (figure 5) emphasizes that if these BPT levels are reached again in a later cycle, it is done at a higher price due to the long-term upwards trend.
  • As a result of the previous point, the chart illustrates that when a certain price level is reached again at a later point, the ‘temperature’ of that price tends to have cooled down significantly. For instance, when the Bitcoin price reached a near $20,000 price for the first time late 2017, its BPT was ~8. When it recently did so again in late 2020, the BPT was ‘just’ ~3, suggesting that the ~$20,000 all time high price level is much less abnormal at the time of writing than it was during the 2017 cycle top.
  • The BPT Bands are similar to Bollinger Bands, the differences being that the BPT bands use a much larger (4 years) time window than is usually done with Bollinger Bands (20 days), as well as range up to a much larger number of standard deviations (up to 12 vs. 1–2) that is used. Furthermore, due to the combination of the large time-frame and the high volatility over that time period, the Bitcoin price is rarely in the negative BPT range, making the negative BPT bands of negligible use, unlike Bollinger Bands. Nonetheless like Bollinger Bands, the BPT Bands widen during periods with high volatility (e.g., a bull market) and contract during periods with low volatility.

This responsiveness to volatility is also one of the key differences between the BPT Bands and the Mayer Multiple bands, the 2-year MA Multiplier, the Golden Ratio Multiplier and similar moving average-based bands that are calculated by multiplying a certain moving average. The prices related to those bands gradually move up at a pace that is equal to the slope of the moving average itself. The slopes of the BPT Bands move up more steeply if the 4-year volatility increases, suggesting higher prices during volatile market conditions that may be more appropriate in that context.

Special thanks go out to Twitter user @Anoi30604540 for providing feedback on the developed charts, as well as for reviewing the draft of this article.

Follow Dilution-proof on Medium or Twitter to be notified of future updates.

The indicators that were introduced in this article are free to be replicated, used and expanded opon by others, as long as the author of and/or the link to this article is referred to. Future work could go out to developing a TradingView indicator for the BPT Bands (e.g., in a wider range from -2 to 12 so that the bands are more visible on more smaller time-frames) and/or a Python implementation of the R code that is available on Github, which can be used for more flexibly visualizing the BPT Bands on online chart platforms.

Disclaimer: This article was written for entertainment purposes only and should not be taken as investment advice.

Get Best Software Deals Directly In Your Inbox

Coinmonks

Coinmonks is a non-profit Crypto educational publication.

Coinmonks

Coinmonks is a non-profit Crypto educational publication. Follow us on Twitter @coinmonks Our other project — https://coincodecap.com

Dilution-proof

Written by

Fascinated by #Bitcoin’s on-chain data, 4-year cycle & potential as the base-layer of an optimistic, sound future. Focus on the signal, ignore the noise.

Coinmonks

Coinmonks is a non-profit Crypto educational publication. Follow us on Twitter @coinmonks Our other project — https://coincodecap.com