Bitcoin Layers and 2021 Week 5 Price Analysis

a bitcoin residual price model

The 2021 Week 5 bullet points:

• Next local top in ~3 weeks.

• My model expects a 30% correction again from possible local top of ~$70k.

• Current predicted cycle top may occur June 30th, 2021 (+/- 15 days 95% confidence interval, see disclaimer in discussion at end of post).

• Price could reach up to $500k-$800k before cycle bubble instability takes over, although sufficient FUD could pop it sooner.

• By April 2022, there seems to be a $19k price floor as a possible bottom of current cycle. Based on previous bear markets, average price during the bear market is about 2x this floor.

• We're at about 15% adoption. This seems a little high to me based on other data, but not entirely unreasonable.

• The model predicts $355k ultimate price floor (+ any speculative premium during bull runs). This is similar to gold today.

Scroll to the bottom for more of the 2021 week 5 analysis.

I would like to remind readers this newsletter is a purely educational exercise. I’m fitting models based on data and am probably wrong. Don’t construe this analysis as investment advice, I am not a financial advisor and you should seek one out before making decisions with your money.

Philosophy: The Layers explaining Bitcoin Price

There are a lot of price models for Bitcoin. Some of the most popular models include @100trillionUSD’s Stock to Flow and those hosted by @woonomic on Woobull Charts.

These models try to figure out how much Bitcoin is worth at any given moment in time, using a variety of data-driven or on-chain data. I won't cover these models now, but in overall I think they're great and useful. However, as a professional MODLer, it's important to note that even the best models never tell the whole story. By themselves, they simply don't have the expressiveness necessary to model reality. Instead, I like to explore a variety of different models.

In order to keep all of this organized, I think it's useful to think about explaining the price in terms of layers. These layers loosely correspond to decreasing timescales and impact on the current price. These are the contributing factors that lead to the actual price found on a particular exchange. Today, I'll expand on how I see these layers and their impact on the long and short term price of Bitcoin. There are numerous models for each layer, and a good final price prediction would combine some amount of these models together.

Bitcoin Layers

There are roughly 4 different layers that contribute to the overall price: Adoption, Speculation, Sentiment, and Chaos.

We know Bitcoin is highly volatile, but it's important to note that this volatility is different for each layer. When we compare a model to the actual price, the difference between them is known as the "residual error." As we add these layers together we progressively explain the residuals:

Price = Adoption + Speculation + Sentiment + Chaos

Layer 0: Adoption

The bedrock of the price

Timescale: Years
Volatility Impact: Fundamental

Ultimately, Bitcoin is valuable because people use it. What is the value proposition? It is a fairly fungible thing with a limited supply that a bunch of other people want. That makes it a good store of value. You can set it aside and wait until you want to trade it for something else. In this way, the market size is really the driver of the Bitcoin price. This is why Adoption is Layer 0. As more people use the network, it becomes more valuable because the market increases. It's the diversity of the market (in terms of what you can trade for it) that is important, and a good proxy for that is the number of individuals that use it.

Metcalfe famously showed that as the number of connected individuals grow, the value of a network is proportional to the connections between them. Bitcoin thus has a “network effect.” The number of connections in a network is roughly the square of the number of connected individuals. Therefore, we expect the price to increase by the square of the number of people who join the marketplace. If twice as many people join, the network becomes 4x more valuable!

Where are we in the total adoption of Bitcoin? My current analyses indicate we're about 5-15% of the way there. Therefore, we expect about 10x more users, indicating on the order of a 100x increase in the long-term price floor (currently around $10k). It turns out such fundamental value might be less (~50x?) since Bitcoin’s Metcalfe factor seems to be slightly less than the square. Of course, Bitcoin could trade at a significant premium from this fundamental value or the Metcalfe factor could change (via Hyperbitcoinization). More on this in later articles.

Layer 1: Speculation

Bull and Bear cycles. The Bitcoin market cycle.

Timescale: Months
Volatility Impact: High: 1-10x multiple of Layer 0

Bitcoin has a limited supply. As people adopt it, there is a rising demand for it, leading to a steady increase in price. However, when there are supply shocks, either coins flood the market or are HODLed more tightly than normal, the price can move pretty dramatically. These price fluctuations then get people excited, they begin speculating that the price might continue to rise as the shock fails to meet demand. Bitcoin has a programmed supply shock every 4 years: the number of mined coins halves until the maximum of 21 million is achieved in 2140.

A sharp supply decrease like the pre-preprogrammed halving can trigger a "bubble" where people buy in with the hope the price goes up and up. Bubbles typically attract additional participants until their expectations for price increases become unsustainable, leading to a collapse in price. Recently, there's been quite a bit of work modeling these bubbles, as they seem to follow a very predictable pattern and are ubiquitous throughout financial markets. Bitcoin has an inelastic supply, which seems to make it particularly well explained by bubble dynamics. My favorite bubble model is the Log-period Power Law (LPPL) model.

The LPPL model predicts that as a bubble forms it becomes more fragile. As the price increases, FUD can trigger a temporary dip, until price begins to rise again. But as the price goes higher, the bubble is more fragile and a dip is easier to trigger. This forms the typical “staircase pattern” seen during most financial bubbles, and the basis of the “BTFD (NSFW)” war cry. Eventually a singularity is reached and even the most minor FUD freaks everyone out too much and the bubble pops. Importantly, the top can not be predicted because the FUD is unpredictable, but we can estimate how fragile the bubble looks at any given time.

Should you just wait until a bubble pops? In Bitcoin, temporary dips typically cause a 30% correction during a bull run. Once the bubble bursts, the price can collapse by 80% or more. Some of the new network participants stay around because they realize Bitcoin is actually pretty awesome. The increased size and diversity of the marketplace then raises the price floor via Layer 0. So far, the price of Layer 0 has been higher than the previous bull-run peak, so rather than wait around, it's been better to DCA when you can.

Layer 2: Sentiment

Leveraged trading plays, FUD and Hopium

Timescale: Days/Weeks
Volatility Impact: Medium: +/- 30%

Bubbles are fascinating. But models like LPPL only explain the ebb and flow like speculative frenzy. Real prices are influenced by FUD and Hopium, various fluctuating narratives that drive people to buy or sell in the short term. Many people are taking on leverage to “buy the rumor, sell the news.” They aren't selling because they don't believe in the market cycle or fundamental value of bitcoin, they're using leverage to stack sats and trying to ride volatility. This is the domain of whales.

If you know this, you might bet on when sentiment triggers fluctuations. During a bear cycle, Layer 1 is less impactful, and most of the short term price movements can be explained almost entirely by sentiment changing in Layer 2. In both bull and bear cycles, leverage can amplify price movements as traders are liquidated. Some exchanges have been accused of using their bitcoin reserves to manipulate the market and trigger margin calls. It appears to only have a short term effect on the price and does not change price models due to Layer 0 or Layer 1.

Importantly, good and bad news cannot be predicted. Recently, Elon Musk declared his "approval" of Bitcoin, and that Hopium moved the price by over 20% overnight. We're currently coming out of a bull cycle dip and such Hopium can be the catalyst that triggers renewed speculative frenzy.

Layer 3: Chaos

Random fluctuations, momentum, bots and swing traders

Timescale: Seconds/Minutes/Hours
Volatility Impact: Medium: +/- 5%

The final layer consists of the fundamental randomness of markets. There is a complex ecosystem of traders, market makers, and bots that are all essentially gambling on very short term price movements. Technical analysis (TA) and other charting techniques fall into this category.

Financial markets seem to look like random walks in the short term. In addition to random walks, auto-regression induced mean reversion and momentum can make these movements look like something more sophisticated. However, any predictions tend to be much more about gambling in the short term. I personally find these predictions super entertaining, but treat it like playing blackjack. Sometimes you can gain a slim advantage, but eventually the house wins and you get rekt. Layer 2 FUD and Hopium can completely wipe any “pattern” forming in the chart. If you're charting on the weekly, you might actually be dealing with Layer 1, and I believe there are better tools to use (such as LPPL).

I think there is room for more rigorous approaches to TA, using things like agent based models to predict price floors, resistance, and impact of order flow. This can only explain a little bit of the price however, and is the domain of bots and traders. Sometimes there's a totally weird 5% dip that's entirely unpredictable, and is just due to randomness and chaos.

Layer 0-2 can be highly influenced by this randomness... like a butterfly flapping its wings in Australia creating a Hurricane in the Caribbean. A random 5% drop spooks a whale, who sells and triggers a cascading margin call right when the bubble is super fragile, causing the bubble to collapse, we enter a bear market and less users join the network so Layer 0 produces a price floor that's lower than it would have been otherwise. That said, Bitcoin has proven to be wildly resilient. We have to welcome the possibility of chaos into the market. Perhaps this chaos actually makes the ecosystem more robust, HODLers more disciplined, and developers more rigorous?

Additional Thoughts on Layers and Volatility

A note on volatility: Bitcoin is famous for it. It appears Layer 1 is the primary driver of it, with Layer 2 sentiment a close second. As Layer 0 progresses, however, it's possible we will be left with smaller swings due to Layer 2 and 3, and occasional bull and bear cycles triggered by Layer 1. These bull cycles triggered by speculation and supply shocks seem to be a feature of hard money, and was seen in the 19th century when the world worked primarily on the gold standard. If the market capitalization of Bitcoin converges to a big enough value, then these shocks will probably be fairly minor, making 10x swings not a thing anymore.

Analysis: Adoption Layer 0

Adoption has been looked at by a few really smart people. Recently, Willy Woo (@woonomic) has recently used on-chain data to estimate the number of individual users, as well as available data on user sign ups to exchanges.

I have recently been using the following paper as inspiration to do my own analysis: Are Bitcoin Bubbles Predictable?

The paper uses on-chain data up to 2018 to predict adoption. They enforce a parametric sigmoid model to the data fit. Adoption curves tend to look sigmoidal, and represents various phases such as early adopters, the majority, and laggards as people embrace technology.

Instead of fitting this model to on-chain data, I thought it might be interesting to simultaneously estimate adoption by fitting the parameters for sigmoid model and a generalized Metcalfe's Law to pricing data. In order to do this, I fit the joint model to manually selected price “floors” from the past 5 years. I added recent prices from this past year so my prediction is updated from the 2018 used in the paper.

With some luck, the recovered parameters can tell us something about the current trajectory of adoption and implied terminal market cap. This adoption estimate would then form the floor for Layer 0 in the price model. This approach uses only price data, although in the future it would be interesting to incorporate on-chain data as a prior to reduce uncertainty in the estimates.

Quick takeaways:

• The current Layer 0 price floor is $11,300. It should not go lower than this if bubble were to burst today. By April 2022, this rises to $19k (possible bottom of current cycle). Based on previous bear markets, average price during the bear market is about 2x this floor.

• We're at about 15% adoption. This seems a little high to me based on other data, but not entirely unreasonable.

• The model predicts $355k ultimate price floor in today's dollar value. This implies everyone would eventually store at least 3% of their wealth in Bitcoin assuming roughly $200 trillion global wealth. This is similar to gold today.

• We are approaching an inflection in the adoption curve. It starts slowly and then happens suddenly. ~54% adoption by 2028, ~88% by 2040.

• Looking at a linear chart shows how insane bull runs (guesstimated in green) look. Will future bull runs be as extreme as the past or will volatility decrease?

The short term price floors are probably more trustworthy than long term predictions. The model sensitivity suggests the long term market cap predictions should be consumed with a hefty chunk of salt. For example, the implied ultimate coin price changes by $50k if I add 1% noise to the pricing data. A proper sensitivity analysis and using more pricing data (rather than hand-picked "floors") would be interesting to look at in the future. Hilariously, the current prediction is in the ballpark of a Gold 2.0 thesis, even though I’m just looking at the current price growth rate fit to sigmoid + Metcalfe’s law.

Analysis: Speculation Layer 1

We are definitely in the bull run of the Bitcoin market cycle. Fitting to the LPPL model, assuming the bull run started at the beginning of October 2020, we've just experienced our first major dip. Since I've written out this model using proper noise statistics, this model has 95% confidence intervals. It should be a bit more trustworthy than the long term Layer 0 predictions. It’s important to note the LPPL model is very aggressive, it usually predicts more extreme price movements than what will happen in the mid-to-long term. As more data rolls in, it will become more accurate, so even though the plot below shows $100k in ~40 days, this is unlikely in my opinion. Also, this is just a model fit to data, black swans and poor assumptions could make these predictions meaningless.

According to the model, we're on the way back up after a dip and should see a pretty hefty run for the next 3 weeks until we experience a local top and correction at around $60k (+/- $10k, 95% confidence). For the current local top, we saw the price go slightly past the 95% confidence limit (~$40k) and then crash to the lower level (~$30k). If this plays out similarly, we would see a run to $70k and then a dip to $60k with a possible wick to $50k.

Here are the takeaways:

• Next local top in ~3 weeks

• It expects a 30% correction again from possible top of ~$70k

• Current predicted bubble top will occur June 30th, 2021 (+/- 15 days 95% confidence interval, see disclaimer below)

• Longer term (months): Price could reach up to $500k-$800k before bubble instability takes over, although sufficient FUD could pop it sooner.

Date of collapse disclaimer: The predicted date of collapse will shift around as we get more data, so the confidence bound is based on current data and can not account for future price action. As we approach the true date, the predicted date will start to remain constant and the confidence interval will become more relevant. This date has ranged from April-Nov 2021 when running on earlier data, to give a flavor of the variability of this prediction based on available price data. I hope to regularly update this prediction in the form of this weekly newsletter in the future, and track this predicted date of collapse.

Analysis: Layers 2 and 3

I think it's fun to track current trends in FUD and Hopium. These have a short to mid-term effect on price, and can be seen as the interaction between Layers 1 and 2 and can influence the decisions traders make in Layer 3. These memetic sentiments are fundamentally unpredictable, and serve as speculation on what “reality” holds for Layer 0. Remember, Layer 0 drives everything.

Some current Hopium and FUD:

• FUD: Tether might be backed by fractional reserves. This has been around for a long time. If tether fails as a stable coin it might drive volatility, but most traders would probably buy bitcoin to unload their tether. It may make international (non-USD) exchanges have to reinvent themselves, and could shift sentiment against the space in general. This is is a Layer 1 risk (could trigger a bear market), not systemic.

• Hopium: Institutional adoption is coming. Elon Musk and Micheal Saylor are influential proponents. This seems to be a Layer 1 thing, and might have triggered the bull market earlier than most expected.

• FUD: Regulation. This one is kinda scary. This is potentially relevant to Layer 0, but so far adoption has been doing really well. I personally believe we are at the point of no return and technologies like Tor make it impossible to stop adoption eventually. This could slow us down though... but so far only Layer 2 seems to care, so hasn't effected the price dramatically.

• Hopium: Hyperbitcoinization. Bitcoin is a candidate for a world reserve currency. Right now, estimates of adoption indicate Bitcoin could become a Gold 2.0 and will replace that market cap. Institutions and governments may hold it like they do now, but find MMT too difficult to ever get rid of. This is a Layer 0 question, and I'm optimistic, but less confident we can overcome a 10 Trillion dollar price floor market cap in the next few decades. We may have speculative bubbles well above that mark though... and maybe that will lead to hyperbitcoinization.

For now, the numbers look like we're nowhere near the bull run end. I'm curious if a “super-cycle” takes over. Perhaps we'll have a local top end of June, and another bubble will quickly form for the later half of the year, which could take us higher.

Like this content? Please share it and comment. Currently, I plan to publish a newsletter updating the above analysis, let me know if you’d subscribe to a paid version (more regular and deep analysis) via substack.