Alright. Finally published it 👀

Is This Cycle Over?

1/ Bitcoin reached an all-time high of 126K right around the time the peak of a four-year cycle would predict, and crashed hard after that, all the way down to 103K. The momentum on the price chart has also slowed down considerably and looks like it wants to roll over. Onchain data shows relentless selling from long-term holders and weak demand. Trump’s trade war also threatens a painful revamp of global trade. All this raises the question for investors: “Has Bitcoin topped already?”

In this article, I will be providing my thoughts on this issue after considering and consolidating many data points and viewpoints.

Given the importance of the question we address here, we are publishing this openly for all users on an exceptional basis. If you like our reports, please subscribe (link in my bio) and enjoy several deeply researched reports every single week to stay on top of the market trends. 2/ I will organize this write-up into a few points with the aid of this chart.

1- Market internals

The first observation is that Bitcoin really has not had a parabolic run this cycle. In a video on July 28th, I proactively addressed the question of what will end this cycle. A clip from this video is posted here on the Subtack version of the article.

The V-Shaped Market

I finally got around to explaining my entire 2025 playbook.

Bookmark this and mark my words. 🧵/1 2/ The core of my thesis has been that Trump's policies are overall very bullish, with bearish components happening first.

You see a lot of red and yellow to the left of the image and a lot of green to the right.

An Economic Primer on Tariffs

Tariffs have nuanced positive and negative effects on the economy which are often misunderstood. 1/ Tariffs could serve positive strategic goals in bringing jobs onshore and keeping strategic industries competitive.

So they can serve political and strategic goals, but they at a substantial cost.

Plan C and have been developing the Quantile Model since mid last year.

A few days ago, we decided to start a research project and publish additional analyses on it.

The amount of work we've been able to do in a few days just amazes me. Let's recap. 🧵 0/ 1/ First what is the model?

Nothing shows the model better than Plan C's visualization.

Simply put, it is the most reliable Bitcoin model we have as of now with a lot of advantages.

- It has strong predictive value
- It captures the Bitcoin price channel from bottom to top.
- It is a regression that uses all the data (not dot-connecting TA).
- It is robust to outliers and resilient to noise
- It also models cycle-to-cycle decay in tops, and more

x.com/TheRealPlanC/s…

How does the Bitcoin Quantile Model compare to the standard deviation bands of the Power Law Model?

Click on the post to see the full article with properly placed charts.

The Quantile model optimizes around finding the best-fitting lines for each portion of the distribution.

The Q1 and Q99 lines fit the top & bottom 1% of the data to provide a channel without any data manipulation by the analyst (key feature).

The PL model gives an average, and when adding the standard deviation bands (2 SDs up & down), it deviates from the price. Why? Because Bitcoin is both assymetric and has diminishing volatility.

The quantile model has two advantages:

- It captures the tops because it takes into account diminishing volatility.
- It captures the bottoms because it takes into account the asymmetric nature of bitcoin (i.e., the skewed distribution with most data points concentrated at the lower range).

And it does all these with a unified model, using all the at once, without the analyst manually chopping up the dataset, which introduces bias.

Below are the two models overlaid.

Our previous analysis

https://x.com/Sina_21st/status/1885382320313553010

Comparing the Quantile Model and Power Law Model

This is part of a series of investigations that Plan C and I are doing on the quantile model. The findings will be aggregated in several articles.

Question: Does the quantile model fail the cross-validation test?

A good way to test a model's predictive accuracy is a cross-validation test, where we train the model on partial data and test it on fresh, unused data.

In this analysis, I will compare:
- The quantile model's median prediction
- The power law model's mean prediction

Now, let's use partial data up to the end of 2020, and compare the results with the full data.

The Quantile Model: the median prediction does not change even a bit when it is trained on partial data versus full data The Power Law model: it is similarly robust. The full-data and partial-data versions are similar.

Will nation-state adoption break the power law?

Many people ask me whether the PL is too bearish and the SBR or a geopolitical game theoretic Bitcoin rush will break it to the upside.

In this 🧵, I will address what it takes to break the model.

0/n 1/ The power law model predicts us getting to $1M in the next two cycles. The PL average will reach $1M by 2033.

This implies a market cap of $21 Trillion (21M*$1M)

For context, the US GDP is currently at $29T.

Let's kick-start this year's Bitcoin analytics! 📚

Here are three of the most important Bitcoin valuation models you need to watch. I will be discussing these models and more frequently.

Follow for regular updates.🤝

Bookmark 🔖

1) Power Law trend line

this model provides a fair (average) price estimate. Right now, the Power Law price is at $84K, and the actual price at $93K. Bitcoin is trading near fair value. 2) Volatility-Adjusted Power Law Index (VPLI)

I developed this metric to put the Power Law price in perspective and account for the changing volatility and market structure across cycles. The results is a valuation metric between 0-100 with 50 being neutral.

Right now, VPLI is at 54, which is the neutral valuation zone.

The market is not cool or hot. Just about the right temperature.

#Bitcoin ETFs Update – 2024-08-26

A live thread.

1/ Blackrock's $IBIT: $224M 2/ Bitwise's $BITB: -$16M

Part II. Continuing on yesterday's article, commenting on @Adrian_R_Morris's paper on the Power Law

Autocorrelation and the Power Law: What it is and Why it is Misunderstood!

I will begin with an introduction to autocorrelation and then run an estimator that corrects for autocorrelation in the data and demonstrate that its coefficients remain the same as the traditional OLS. The attached plot also summarizes everything compactly.

Introduction
Many have said that the log-log regression of price over time is invalid because of autocorrelation. However, this comes from a misunderstanding of autocorrelation and the problem it causes.

Quoting the article: "The presence of autocorrelation in Bitcoin's price data invalidates the model's predictions."

I have encountered the autoregression argument so many times that it is high time I wrote about it once and for all.

1/8 What is Autocorrelation?

Autocorrelation (aka serial correlation) refers to the correlation of a time series with its own past values. In the context of financial data, autocorrelation suggests that past price movements can influence future prices.

2/ What Problem Does it Cause?

Ordinary Least Squares (OLS) regression is a widely used method for estimating relationships between variables. However, OLS assumes that the error terms (the difference between the observed and predicted values) are independent. When autocorrelation is present, this assumption is violated.

Although the predictions remain unbiased—meaning the average prediction is still correct—the standard errors, confidence intervals, and p-values may become unreliable.

So, while autocorrelation is something an econometrician needs to deal with, if the purpose is only getting an average prediction and not estimating the standard errors, it does not invalidate the results.

In fact, it is mathematically shown that autocorrelation only matters for standard errors and significance tests, not the average value.

In “Introductory Econometrics,” the defacto undergrad econometrics textbook, Jeff Wooldridge, explains this as follows: ... 3/ ... In the “Introductory Econometrics,” the defacto undergrad econometrics textbook, Jeff Wooldridge, explains this as follows:

“What will happen if we violate the assumption that the errors are not serially correlated or autocorrelated? We demonstrated that the OLS estimators are unbiased, even in the presence of autocorrelated errors [… this] alone does not cause bias nor inconsistency in the OLS point estimates." (Chapter 12 – Serial correlation and heteroskedasticity in time series regressions)

What he means by not being BLUE is that the estimator will no longer have the lowest possible variance, and it will be less efficient (requiring more data) but not biased.

#Bitcoin ETFs Update

A thread 🧵

1/ GBTC sells $32M.

The moving average shows how consistently lowe its outflows have been. In the last 40 days, it has had an average outflow of -$29M.
2/ Franklin $0

Research Report – Every $200M flow to #Bitcoin ETFs Increases The Price by 1% 🔥

I just finished an analysis of the relationship between the price of #Bitcoin and the ETF flows. Let’s dive in.🧵

1/ Regressing % change in price on ETF net flow produces the following prediction. 2/ It has an R2 of .42, meaning this model was able to predict 42% of the variations in price (%) just using the ETF flow. Given everything else that is going on in the market, 42% solely from ETF data is astounding.

1/ On path dependence. @nntaleb seems to think Bitcoin's physical existence relies on miners. Therefore, it is a fragile and if no miner is interested it stays zere forever. 2/ When in fact, the blockchain holds all the prior transactions and ownerships in its 400Gb data.
A single user that has this ledger on their computer can provably show that this is in fact the real ledger and other nodes will accept it.