Host: Alex Mint Ventures Research Partner

Guest: Colin Freelance Trader, On-chain Data Researcher

Recording Date: 2025.2.15

Hello everyone, welcome to WEB3 Mint To Be initiated by Mint Ventures. Here, we continuously question and deeply think, clarifying facts, exploring realities, and seeking consensus in the WEB3 world. We aim to clarify the logic behind hot topics, provide insights that penetrate the events themselves, and introduce diverse perspectives.

Disclaimer: The content discussed in this podcast does not represent the views of the institutions of the guests, and the projects mentioned do not constitute any investment advice.

Alex: This episode is a bit special because we have previously discussed many topics about specific tracks or projects and exchanged some cyclical narratives, such as memes. But today we will discuss on-chain data analysis, especially the on-chain data analysis of BTC. We will closely examine its operational principles, key indicators, and learn its methodology. In today's program, we will mention many concepts related to indicators, and we will list these concepts at the beginning of the text version for everyone's understanding.

Some data indicators and concepts mentioned in this podcast:

Glassnode: A commonly used on-chain data analysis platform that requires a subscription.

Realized Price: Calculated based on the price at which Bitcoin last moved on-chain, reflecting the on-chain historical cost of Bitcoin, suitable for assessing the overall profit/loss status of the market.

URPD: Realized Price Distribution. Used to observe the price distribution of BTC chips.

RUP (Relative Unrealized Profit): Relative unrealized profit. Used to measure the ratio of unrealized profits of all holders in the Bitcoin market to the total market capitalization.

Cointime True Market Mean Price: An on-chain average price indicator based on the Cointime Economics system, aiming to more accurately assess the long-term value of BTC by introducing Bitcoin's "time weight." Compared to BTC's current market price and Realized Price, the True Market Mean Price under the Cointime system also considers the impact of time, suitable for BTC's long-term price cycles.

Shiller ECY: A valuation indicator proposed by Nobel laureate Robert Shiller, used to assess the long-term return potential of the stock market and measure the attractiveness of stocks relative to other assets, improved from Shiller's CAPE ratio, mainly considering the impact of the interest rate environment.

Opportunities to Learn On-chain Data Analysis

Alex: Today, our guest is freelance trader and on-chain data researcher Colin. Let's have Colin greet our audience.

Colin: Hello everyone, first of all, thank you, Alex, for the invitation. I was a bit surprised when I received this invitation because I am just an unknown small retail trader without any special title, quietly doing my own trading. My name is Colin, and I run an account on Twitter called Market Beggar, where I mainly share some educational content on on-chain data, analysis of the current market situation, and some trading concepts. I see myself in three roles: first, as an event-driven trader, I think about event-driven trading strategies; second, as an on-chain data analyst, which is the main content I share on Twitter; and third, more conservatively, I call myself an index investor, choosing to allocate part of my funds to large-cap U.S. stocks to reduce the overall volatility of my asset curve while maintaining a certain defensiveness in my overall position. That’s roughly how I see myself.

Alex: Thank you for the introduction, Colin. I invited Colin to participate in the program because I found his on-chain data analysis of Bitcoin on Twitter very enlightening. This is a topic we haven't discussed much before, and it's also a part that I personally find lacking in my own discussions. After reading his series of articles, I found the logic clear and substantial, so I invited him. I want to remind everyone that today, whether it's my views or the guest's, they carry a strong subjectivity, and the information and opinions may change in the future. Different people may interpret the same data and indicators differently. The content of this episode does not constitute any investment advice. This program will mention some data analysis platforms solely as personal use examples and not as commercial recommendations. This program has not received any commercial sponsorship from any platform. Let's get into the main topic and talk about on-chain data analysis of crypto assets. Earlier, we mentioned that Colin is a trader. Under what circumstances did you start to engage with and learn about on-chain data analysis of crypto assets?

Colin: I think this question should be answered in two parts. First, I believe that anyone around me who wants to enter or has already entered the financial market, including myself, has the primary goal of making money to improve their quality of life. So my philosophy has always been consistent: I learn whatever can help my profitability. By doing this, I enhance the expected value of my overall trading system. In simple terms, I learn what can make money. The second part is that my initial exposure to on-chain data was somewhat accidental. About six or seven years ago, I had no understanding of it at all, just looking at this and that. While exploring various fields, I came across some interesting research theories that I wanted to learn about. At that time, I stumbled upon the so-called field of on-chain data analysis for Bitcoin, and I started to learn and research it. Later on, I combined the knowledge I learned from other fields, mainly from quantitative trading development, with on-chain data to develop some trading models, which I then integrated into my own trading system.

Alex: So how many years have you been systematically learning and researching on-chain data analysis since you formally started?

Colin: I find it hard to define. I have never really learned it systematically. From the beginning until now, I have encountered a problem: I have not seen any systematic teaching. When I first came across this field, it was several years ago. I noticed it but didn't delve deeply into it; I just read a couple of articles to understand it. After a while, I returned to see some more in-depth content, but at that time, I was focused on studying other things. I found this quite interesting and continued to research it. There hasn't been a specific time for systematic learning; it's more like piecing things together.

Alex: Understood. How long have you been applying what you've learned about on-chain data in your actual investment practice?

Colin: It's hard to define that boundary, but I think it's close to two Bitcoin cycles… but it can't really be counted as two cycles; it depends on whether you define it from a bull market or a bear market. I started getting involved around 2019 or 2020, but at that time, I didn't apply it practically because I was hesitant; I wasn't very familiar with it yet, but I had already started learning.

The Value and Principles of On-chain Data Analysis

Alex: Got it. Next, we will discuss many specific concepts related to on-chain data analysis, including some indices. What on-chain data observation platforms do you generally use in your daily work?

Colin: I mainly use one website, which is Glassnode. To briefly explain, it requires a subscription. There are two paid tiers: one is the professional version, which is quite expensive; I remember it costs over $800 a month. The second one, I forget the exact price, is around $30 to $40 a month. There is also a free version, but the information available in the free version is quite limited. Of course, besides Glassnode, there are many others, but I chose it because it matched my needs the best during my initial filtering and research.

Alex: I see. After looking at a lot of Colin's information, I also registered for Glassnode and became a paid member. I do feel that their data is very rich, and the timeliness is quite good. Now, let's talk about the second question. You mentioned that you are a trader, and you value its help in investment practice. What is the core value of on-chain data analysis in your investments? What are the underlying principles? Please introduce it to us.

Colin: Sure. First, let’s talk about the value and principles of on-chain data analysis. I plan to combine these two points because they are quite simple. In our traditional financial markets, whether trading stocks, futures, options, real estate, or some commodities, Bitcoin has a fundamental difference from them: it uses blockchain technology. The most important and frequently mentioned value of this technology is its transparency. All transfer information of Bitcoin is public and transparent, so you can directly see, for example, 300 Bitcoins being transferred from one address to another; this can be checked on a blockchain explorer. Although I cannot know who is behind this string of addresses, that is not important because no single individual can influence the overall price trend of Bitcoin. So normally, when we study on-chain data, we look at the overall market, its trends, and the consensus and behavior of the crowd. Even if I don't know who is behind this address or that address, I can analyze the flow of chips by aggregating all addresses, seeing whether they have taken profits or stopped losses, how their profit situation is, how their loss situation is, and at which price levels they prefer to buy large amounts of Bitcoin or where they are reluctant to buy Bitcoin. This data is actually visible. I believe this is the greatest value of Bitcoin on-chain data analysis compared to other financial markets because other markets cannot do this.

Alex: Indeed, this point is very important. Just like in crypto investment, we need to analyze the fundamentals just like we do with stocks or other products. As you just mentioned, on-chain data is transparent, and everyone can observe it. If other professional investors are looking at on-chain data and you are not, it is equivalent to having one less important weapon in your investment.

Challenges of On-chain Data Analysis

Alex: When you are practically doing on-chain data analysis, what do you think are the main difficulties and challenges?

Colin: I think this question is very well asked, and I plan to answer it in two parts. The first part is relatively easier to address, which is a challenging point in learning: foundational knowledge. For most people, including myself at that time, as I mentioned earlier, it is difficult to find a truly systematic teaching. Of course, I didn't inquire offline about whether there were any paid courses available, but even if there were, I probably wouldn't have dared to buy them because I have been trading for a while now, and I generally don't pay for courses. I haven't encountered any systematic teaching courses, so all the content has to be self-explored and discovered. There are many types of on-chain data, and during my research, my philosophy is to clarify the calculation methods and principles behind each indicator I look at. This is actually a very time-consuming process because when you see a particular indicator, it gives you a calculation formula, and my idea is to figure out what this formula is really trying to convey and why it is designed that way. After I understand these indicators, the next step is to filter them. Those with experience in quantitative strategy development or those who have studied indicators will know that many indicators have very high correlations. High correlation can lead to a problem where you easily generate noise in interpretation or over-interpret the data. For example, let's say I have a topping-out system with 10 signals numbered from 1 to 10. If the correlation between signals 1 to 4 is too high, it can create a problem. For instance, if Bitcoin's price exhibits a certain behavior or change today, it might cause signals 1 to 4 to light up simultaneously, which can be troublesome. If their correlation is too high, this is a natural phenomenon. If out of 10 signals, 4 are lit up, you might say this is dangerous, but that’s not very reasonable because they were bound to light up. If you don't segment them based on correlation, this phenomenon is very likely to occur. After studying the principles of each indicator and data, I can directly see from the calculation formulas whether their correlations are high or not, and I segment them based on correlation. For example, if these 5 have high correlation, I will slightly filter them and ultimately select one or two.

The first part is relatively easy to solve and not the main difficulty. The second part is the real challenge: how do you prove your viewpoint to those around you or to yourself regarding on-chain data? I might give a somewhat crude example, but it's easy to understand. I previously wrote in a tweet that the quantitative field often tells you that trading cannot be done by simply following the past. I once gave an example: suppose there is a very strange trading strategy where the entry criterion is that if my dog barks twice and it is raining outside, then I go long. If I backtest this strategy 1,000 times and find a win rate of 95%, far surpassing the market, would anyone dare to use this strategy? It seems quite strange; barking dogs and rain lead to a long position, and the win rate is so high. This actually has a term called survivor bias. If you cannot provide any logical support for it, even with a sufficient sample size, this strategy cannot be used. Some might argue that it backtested 1,000 times with a 95% win rate, and the backtest results support that this strategy is usable. As I mentioned earlier, this is survivor bias. Simply put, if I flip a coin 10 times and get heads every time, the probability is actually 1/1024. In other words, on average, 1 out of 1,024 people doing this will succeed, and the situation of getting heads 4 times in a row is what we call a survivor; the other 1,023 people doing this will fail, and we won't see them. We only see the successful cases. Returning to Alex's question about where the main difficulty lies: we mainly look at large-scale consensus and trends. Reviewing Bitcoin's history, the three most obvious cycle tops are in 2013, 2017, and 2021, which gives us only 4 samples, absolutely insufficient. Since the sample size is insufficient, if we try to follow the past by looking at where a certain indicator was in 2013 and where it was in 2017, and then assume it should be at that level this year, that is unreasonable. Because the sample size is completely insufficient, if we do not provide it with logic for research, your theory is very likely to fail. A major issue is that, facing such a small sample size in history, I must use deductive reasoning rather than simply inductive reasoning for research. After my research, I draw a conclusion based on deductive reasoning and need to let time prove whether my view is correct or not. If it is correct, it indicates that my earlier deductive reasoning process may be reasonable. If it is wrong, I need to continue to correct my earlier deductive logic. However, if I only rely on inductive reasoning, most retail investors prefer to do this, thinking that past trends look very similar to current trends, so there should be a surge or a drop later; this is actually unreasonable. Returning to the first statement I made, I think the biggest challenge is proving to others or to myself that my reasoning is correct, so I must constantly revise my logic and assumptions and check for any flaws. Because Bitcoin is too young, the on-chain data analysis will always face the problem of insufficient sample size, and at this point, you have to rely solely on deductive reasoning, using logic to infer it, and then wait for time to validate your judgment. This is the biggest difficulty I currently face.

Key On-chain Indicators to Focus On

Alex: I understand, and I find this very enlightening. The question I asked you earlier was also a confusion I had when I started looking at various indicators on Glassnode. There are so many indicators; which one should I use as my trading reference? Because many indicators have various calculation logics. I tend to select indicators based on their logic, which is quite similar to what you just mentioned. First, I need to look at the computational logic behind the indicator, and I need to feel that this logic makes sense, rather than just relying on backtesting that seems to indicate the indicator is accurate and then using it to predict the future. As you said, the reference in deductive reasoning needs to be greater to be considered a primary indicator for us. Based on your insights just now, in your current daily analysis of Bitcoin, what on-chain indicators have you been focusing on for a long time or consider to be relatively important?

Colin: This question I have mentioned before; I will try to filter based on correlation. I look at many on-chain data indicators, and today I will introduce them from different dimensions, specifically breaking them down into three levels based on lower correlation.

The first indicator I will focus on long-term is definitely the URPD indicator. It is a chart presented as a series of bar graphs, with the horizontal axis representing Bitcoin's price and the vertical axis representing the quantity of Bitcoin. Suppose we see a very tall bar at the $90,000 position; we would know that a very large number of Bitcoins were accumulated at this price, which indicates their buying cost. That bar graph will show how many Bitcoins were bought at that price level. So, based on this, we can see at a glance that if there is a large accumulation above $100,000, we can know that many people bought in above $100,000. The URPD chart mainly has two key observation points. The first is the simplest chip structure. Suppose I see the current market situation is around $87,000, and there is a very large accumulation of chips above $87,000, according to last week's data, it should be 4.4 million. We know that there has been a significant turnover in this range, or that someone has bought in here. Since someone has bought in, it is very likely to form a certain consensus. In such a heavily accumulated range, it is easy to create an attractive effect on the price, meaning the price is likely to oscillate within this range. If it drops, the price may easily recover after a while and rise again. If it rises, the chips below have all turned into floating profits, making it easy for them to sell, engage in short-term trading, and push the price back down. So, it is very likely to oscillate within this range. This is the first observation point. The second observation point is that we can observe the distribution process of Bitcoin through URPD. The so-called distribution refers to the chips bought at low prices during the early bear market, which are then sold off. I define this process as distribution. Suppose today at the $100,000 price level, there are an additional 300,000 chips with a cost of $20,000, and if 300,000 are sold off, we can see that those who bought at $20,000 sold 300,000 today, with their average selling price around $100,000. We can observe whether those low-cost chips show any significant changes. Of course, the current price is $100,000 or $90,000, so any significant changes would definitely be a decrease, not an increase, because the current price range is above $90,000, and it won't go back down to $20,000, so there will only be a decrease, not an increase. Therefore, we can observe the rate of distribution based on this. That’s the general idea. This is the first indicator I will focus on long-term.

The second indicator I want to introduce is called RUP, which stands for Relative Unrealized Profit. The purpose of this indicator is to help us measure the overall market's profit situation, which reflects the market's profitability concerning the current Bitcoin price. For example, how much you are earning, whether it is not much or a lot, is the general concept. The principle of this indicator is very simple; through the transparent mechanism of the blockchain, we can track the buying prices of most chips. We can compare these buying prices with the current price. For instance, if someone bought at $50,000 and the current price is $100,000, we know that this Bitcoin is currently profitable, and we can calculate how much profit it has made. For example, if there are 10 Bitcoins bought at $50,000, and now the price is $100,000, one would earn $50,000, and ten would earn $500,000. We sum up all these floating profits and losses, then standardize this number based on the current market capitalization, and we can get a number between 0 and 1. This range between 0 and 1 is easy to observe. If today the RUP is high, for example, 0.7, 0.68, or 0.75, we know that the overall profit situation in the market is high, which may lead more people to want to take profits. Therefore, a high RUP is usually seen as a relative warning signal.

The third dimension I want to discuss is a fair valuation model for the market. There are actually many different Bitcoin valuation models available, each using different methods to assess the fair value of Bitcoin. The so-called fair value is essentially how much one Bitcoin is worth. After reviewing so many models, I believe the most robust one is the Cointime Price model. I haven't seen its Chinese translation elsewhere. Simply put, we often hear the name Cathie Wood, who leads ARK Invest, and this concept is mentioned in a document produced in collaboration with the on-chain data website I just mentioned, Glassnode. The main feature of this model is that it introduces the concept of time-weighting to calculate the fair value of Bitcoin. The resulting number has two main applications. The first is quite simple: bottom fishing. Suppose during a bear market, the price keeps falling and eventually drops below the valuation given by the Cointime Price. As I mentioned earlier, this number indicates how much one Bitcoin should be worth. If it falls below this level, it means you are buying at a very advantageous position. Historical backtesting and its logic show that whenever the price drops below the Cointime Price, it is actually a very good bottom-fishing opportunity. The second application is to identify tops. We can monitor the current price and see how far it is from the Cointime Price. If it deviates too much from the Cointime Price, we can assess whether this deviation indicates that the market may be approaching a top. The three dimensions I want to share are chip structure, profit status, and fair valuation models.

How to Handle Conflicting Data

Alex: Okay, that was very clear. Many users might ask a question: the three indicators you just listed represent different aspects and align with what you mentioned about their lower correlation, so they can be used together as reference indicators. Now, suppose these indicators show a divergence in practical application. For example, indicator one suggests that we are currently in a distribution phase, while indicators two and three may indicate that we are not yet close to the top from a cyclical perspective. How would you handle the situation of conflicting data?

Colin: I think this situation is not only present in on-chain data analysis but can also occur in other fields, such as technical analysis or macroeconomic analysis. In the on-chain space, my personal approach is quite simple: I assign different weights to different aspects. The aspect I value the most is the chip structure, which is the progress of distribution. The profit status helps me observe whether the low-cost chips in the market, such as those bought at $15,000 or $16,000 during the bear market, have completed their distribution. A particularly interesting phenomenon is that in every Bitcoin cycle, there are usually two very obvious large-scale distributions. For example, in 2024, the most notable case was from March to April last year, where you could definitely see large-scale distribution from a profit status perspective. However, if I only see large-scale distribution, my next question would be: have they finished distributing? All judgment criteria stem from this question. If there is large-scale distribution but it hasn't finished, I can confidently tell myself that the bull market is not over yet. For instance, during March to April last year, when Bitcoin surged above $70,000, I was quite excited because the bull market had finally arrived and reached new highs. However, it then started to oscillate for over half a year. At that time, I couldn't conclude that we had reached a bottom based on the data; at most, it was just the first distribution. Many data points also indicated that, based on the average cost of short-term holders, the situation was quite different from when a true bull market ends. So, I felt quite reassured at that time. When you mention conflicting data, if it indicates distribution, do I need to escape the top? Actually, no, because the main issue is still the one I mentioned earlier: has the distribution ended? Using this question as the standard for filtering each indicator and making judgments can easily lead to the conclusion that even if distribution has occurred and is large-scale, I just need to determine whether it has ended. Using this as a criterion can effectively address the so-called conflicting data issue.

Alex: Now let's set a scenario. Suppose we look at URPD, and this indicator has shown two distributions, similar to what you mentioned earlier, one in March and April last year, and another peak from December to January. If it shows this distribution, but the other two valuation indicators are not as high, when this situation arises, you mentioned assigning different weights. Would you reduce your position based on the weight proportions, or would you consider all three indicators together without adjusting positions based on weight, making one or two important decisions at critical moments?

Colin: My approach is the former because no one can truly know whether we are at a real top. No one can escape at the highest point; if someone could, that would be impressive, and I would definitely want to meet them. Personally, I interpret the top as a gradual process. Although it may seem quick when looking at daily charts, if you are in the moment, for example, at $69,000 during the last cycle's top, you wouldn't feel that it is the top. We can only make a judgment based on data that the conditions for forming a top may be present. So, based on this premise, I would adopt a segmented stance. For instance, when I believe the conditions for a top are gradually maturing, if I see an indicator giving me a warning during this period, such as a divergence in RUP that I previously shared on Twitter, I would correspondingly reduce my position. Of course, the extent of this reduction should be predetermined; it wouldn't be appropriate to randomly reduce without knowing how much to cut. I would first outline a rough plan, for example, dividing my position into four parts. Once a certain type of warning appears, I would reduce one part, and when the second warning comes out, I would reduce another part. I would also plan that no matter what, the last portion of funds must exit. For instance, if the bear market has been confirmed to end, but other warnings have not yet appeared, we need to devise an extreme strategy for the final exit.

Alex: Understood, we gradually exit and reduce positions based on different warning signals.

Colin: Yes.

Judging BTC's Position in This Cycle and the Basis for It

Alex: I understand. I have been following your Twitter account recently, where you regularly apply the indicators we just discussed, including the underlying concepts, in your trading practice. Now, looking at Bitcoin, it has been oscillating in the range of $91,000 to $109,000 for almost three months. Currently, there is significant divergence in the market regarding this price range, unlike in December and January when everyone felt that this bull market was far from over and would surge to $150,000, $200,000, or even $300,000, with many optimistic views. The current market is quite divided; some believe the top for BTC is around $100,000, while others think BTC has not yet peaked in this cycle and that there will still be a major upward wave in 2025. Based on your current comprehensive judgment, what is your view? Where does BTC stand in this large cycle? What data sources support your judgment?

Colin: Before answering this question, I should probably give a warning: I am actually very bearish about 2025. I believe BTC is currently in a position that has the conditions for forming a top. I know that many people, including some participants around me, have not had good returns during the so-called special bull market of 2024 because the overall market behavior in 2024 is quite different from previous cycles. The most obvious point is the absence of an altcoin season. This has hurt many people, including some non-professional traders among my friends who have entered this market and suffered significant losses in altcoins. Why is this the case? Looking back at 2024, there was an altcoin rally at the beginning of the year, and the second occurred in November last year when Trump was elected president. Compared to previous cycles, these two altcoin rallies had a significant and obvious point: their sustainability was quite poor. Even during the rally in November and December last year, altcoins did not rise comprehensively; it was a very clear sector rotation. At that time, there was a DeFi sector that rose, followed by older coins like XRP and Litecoin, and that sector rotation was very evident. From this, we can see that if people consider the current bull market in 2024 as a bull market, this cycle is actually very different from previous ones. There is also a theory that a bull market must have an altcoin season before it ends, but I personally believe that you cannot say that the appearance of an altcoin season is a prerequisite for the end of a bull market; this is clearly not strongly correlated. We cannot use this as a judgment for whether the bull market has ended. As mentioned earlier, on-chain data analysis has an inherent shortcoming: the sample size is always insufficient. Simply using historical conditions to extrapolate today's market is akin to following the past blindly, which is not advisable. If you were to follow the past blindly, the tops in 2013, 2017, and 2021 should have appeared around the end of the year based on the timeline.

I personally believe that we are now in a position that has the conditions for forming a top. The reasons are quite complex, and I use many indicators and data to make this judgment. Let me briefly mention a few core points.

First, the chip structure we just talked about, which is the URPD chart. We can see that in 2022 and 2023, a large amount of low-cost chips were accumulated when they bought a lot of BTC at low prices. To date, many of these chips have already been distributed. To put it simply, they have sold off and are no longer participating. Some listeners might wonder, "What does it matter to me if they sold?" There is a concept that needs to be explained: at the end of every bull market, it is almost always because those low-cost chips have finished distributing, and then the bull market ends. A less intuitive point here is that it is not because they dumped the market that the bull market ends; rather, it is because the price has been rising, and they sell all the way until they are done selling, at which point the price stops, and the bull market is about to end. This is not just a random thought; there is a logic behind it.

Assuming that every BTC chip participating in the market is a high-cost chip, for example, bought above $90,000, while the chips bought at $50,000, $20,000, or $30,000 have already exited. At this point, as long as there is no obvious or strong upward trend in the price, even if it is just a wide range of oscillation, such as the oscillation between $70,000 and $50,000 last year, or the current oscillation between approximately $90,000 and $109,000, it will put significant pressure on these high-cost chips. High holding pressure can lead to a problem: if the price is around $95,000 or $96,000, and it drops to $89,000, which is less than 10%, the pressure on these chips becomes very high. Many of them are short-term traders, and once the pressure builds up, they may choose to sell. This selling can lead to further price declines, which in turn causes other high-cost chips to be unable to bear the pressure, leading them to sell as well, creating a chain reaction. This is what I see from the URPD chart: many low-cost chips have already been distributed.

The second indicator I mentioned is called RUP, which measures the market's profit status. If you are interested in this indicator, you can look it up; it is quite interesting. If you overlay its line with the price line, you will find that their correlation is very high; they almost move together. This is quite reasonable because the higher the price, the higher the holding cost and profit status, so the shapes of the two lines are almost identical. Therefore, as the price rises, RUP will also rise; when the price falls, RUP will fall as well. This is very straightforward. However, when RUP shows a so-called divergence, it indicates that the market situation has changed. What does divergence mean? For example, if Bitcoin rises to $90,000, then pulls back and rises to $100,000, creating a new higher high, but RUP at $100,000 is not as high as it was at $90,000 and instead declines, this is the situation where RUP has decreased while the price has increased. It is strange why this situation occurs. The only reasonable explanation for this is that, as we mentioned earlier, RUP is calculated using unrealized profits, and the majority of unrealized profits in the market are contributed by those low-cost chips. For example, if you bought one Bitcoin at $16,000 and now it is at $96,000, the floating profit on this Bitcoin is $80,000. However, if you bought Bitcoin at $86,000 and now it is at $96,000, the floating profit is only $10,000. Therefore, the main contribution comes from those low-cost chips. So, when the price is higher but RUP is lower, it indicates that a significant portion of low-cost chips has already been sold off earlier. As a result, when the price rises further, these low-cost chips have exited, converting part of their unrealized profits into realized profits, which is why RUP does not reflect that, leading to a lower RUP and creating a divergence. This point helps me validate my interpretation of RUP, confirming that indeed, low-cost chips have exited.

The third aspect is that there is much more to discuss regarding on-chain data, but I would like to share another unique perspective: the U.S. stock market. If anyone has researched the stock market, they would know that there is a concept of valuation, which is the price-to-earnings ratio (P/E ratio). There are many different variations of valuation methods. The indicator I refer to is called Shiller ECY, which comes from Professor Robert Shiller at Yale University. He measures the yield of stock assets relative to bond assets. This indicator was mentioned in a paper he published after the pandemic in 2020. He believed that his previous model, known as Shiller PE, was no longer applicable due to structural changes in the global market after the pandemic, so he invented a new indicator called Shiller ECY to measure the market, and found that this indicator had better predictive power. In simple terms, this indicator currently shows that the valuation of the U.S. stock market is somewhat too high. It is important to clarify that a high valuation does not necessarily mean a decline; a high valuation can still go higher. However, it measures a concept similar to a spectrum, indicating that it is getting closer to a danger zone. I believe that the current position is relatively dangerous. The valuation of the stock market is mainly driven by the hottest topic, which is AI. Recently, there was a company called DeepSeek that emerged unexpectedly, causing a sudden adjustment in the valuation of the U.S. stock market. However, I am personally pessimistic about this in the short to medium term. Although DeepSeek is a long-term positive for the AI industry, I believe that this valuation effect will not end quickly, so I think there is still room for valuation adjustment. If the U.S. stock market does not perform well, then Bitcoin, as a smaller player, will naturally not look good either. However, these are just my personal biases for your reference.

Alex: Okay, Colin just provided a very detailed explanation. Let's summarize his points briefly. He believes that the current price range meets many conditions for previous valuation tops or price tops, including the distribution of chips, the status of unrealized profits, and he also referenced the Shiller ECY indicator from traditional financial markets, suggesting that there are many signs indicating a potential top.

How to Get Started with On-Chain Data Analysis

Alex: Today, we have already discussed a lot about the principles of on-chain data analysis, including how to observe some commonly used data and how to apply this data in practice. Many of our listeners may not have deeply studied this concept or system before. So, if a beginner were to ask you, "Colin, I find what you talked about today very intriguing, and I want to start learning this knowledge from scratch to guide my own BTC investments," what kind of learning advice would you give them to kick off this learning journey?

Colin: Alright, actually I have received dozens of private messages asking similar questions up to this point. My personal advice has always been the same. First, I have two main strengths: the first is on-chain data, and the second, which I consider my strength, is in technical analysis. Most people who come to ask me usually have a line chart in hand, drawing some patterns or indicators like MACD or RSI, and they ask me if there is a way to combine these with on-chain data perspectives. I must first give a suggestion: I personally do not recommend beginners start learning from the technical analysis field. The main reason is simple: there are too many schools of thought, and many of the viewpoints within these schools cannot withstand scientific scrutiny. They are purely inductive, lacking logic, and can easily fall back into the example I mentioned earlier about a dog barking when it rains, which could very well be a case of survivor bias. However, general beginners do not have the ability to distinguish whether something is genuinely useful or just a survivor bias. My personal suggestion is that on-chain data is a very suitable field for beginners, and I will mention how to learn it later. The reason I believe it is suitable for beginners is simple: first, most of the retail traders around us, or our traders, are not full-time traders. Most of them might be high school students, college students, or office workers who have their own primary jobs. If you cannot spend a lot of time on what is called "monitoring the market," then the trading role of on-chain data is very suitable for you. As we mentioned earlier, the level of observation for on-chain data is quite large, starting at least from the daily level. Since you are observing at the daily level, it means that the frequency of operations you make based on on-chain signals, such as buying or selling, is actually very low. You do not need to make 5 or 10 trades a day; you might only make four or five trades a year at most. Therefore, I think this aspect of observation is very compatible with the daily routines of students or office workers. You do not need to spend too much time; you can set aside half an hour to an hour each day to observe the alerts you have set up and see if there are any different changes in the data.

The second part is how to learn. I mentioned earlier that throughout my learning process, I have not seen any free, systematic teaching materials up to today. There are many teachings, but they lack a systematic approach. They might provide you with a long article introducing one or two indicators in detail, and I think these articles are great. However, the problem is that you still do not have a framework from 0 to 1, so learning can be quite painful. This indicator looks impressive; should I learn it? Should I delve deeper? The next indicator also looks impressive; where should I start learning? My approach is quite straightforward; I learned everything. At the beginning, I did not know which ones were good or bad, so I learned all of them. I looked into the principles behind each one, examining the calculation principles, why the author designed such a formula, what they wanted to see, and whether this formula could really help them see what they wanted to see. This takes a lot of time. After reviewing all these indicators, you need to filter them. But for beginners, this process requires a lot of patience; you really need to look at them one by one slowly because trading is not an easy task. From what I have seen so far, whether in simplified or traditional Chinese, the resources available in the Chinese-speaking community are quite limited. So my suggestion is that if you want to study a particular indicator, it is best to find the original author's article. Try to avoid looking at others; the original author is definitely the person who understands that indicator the best. If you really cannot find it, at least make sure to read through their formula. The website I mentioned earlier, Glassnode, has a column called "Weekly Onchain," where they share the current market situation based on various indicators each week, not fixed indicators, in a report-like format explaining why they think the market is in its current state. You can see various indicators from there, and you can download each one to study, which will provide a large learning resource library. I also have some teachings on my Twitter, which cannot be called systematic, but if you are interested, you can take a look.

Alex: It is quite systematic. I have been following your updates, and it seems you have already written more than ten articles, basically discussing one indicator concept in each issue. Everyone can check it out. There is another question: you mentioned that your first identity is as a trader. Today, we have spent a lot of time discussing how on-chain data helps trading. However, in reality, when you trade, besides analyzing on-chain data indicators, do you also consider other factors? For example, macroeconomic factors, or some fundamental events related to Bitcoin, such as the progress of state and national finances in the U.S. regarding Bitcoin reserves. Besides on-chain data analysis, what weight do other indicators hold in your overall trading decision-making process?

Colin: Alright, I think this question is very profound. First, in terms of my system, the on-chain data part can be thought of as an independent system for my position allocation. I will have a relatively long tail of what is called spot allocation, and even in the bear market bottom, I will slightly leverage it, for example, around 1.5 times or 1.3 times. This is a system, and the main trading decision basis for this system is on-chain data. On-chain data provides me with a broad directional framework; I will know whether the market is in its early, mid, or late stages, whether it is a bull market or a bear market, providing a guiding benefit in a broad direction. As for other parts, I mentioned earlier that my other strength is in technical analysis. This part is actually too complex to discuss in detail because there are many schools of thought and some prerequisite assumptions that need to be clarified. If not clarified, it can easily mislead others. The role of technical analysis in my trading system is to refine the final entry point. For example, if I have confirmed that I want to take a certain opportunity, I will think about where exactly I should enter this trade. I will try to refine my entry point using technical analysis. Just to give an example, this is not financial advice, but let’s say I think Ethereum is a good entry point between $2000 and $2600, and I believe it will rise afterward. If I were God and knew it would rise, I would just buy it. But since I am not God, I will try to find a more satisfactory entry point within that range using technical analysis. As for what that number is, I have to evaluate it each time, so I cannot provide a specific figure, but I will have a set of measurement criteria.

Next, on the macro level, I pay more attention to the global market supply chain and the decisions of the U.S. Federal Reserve because the U.S. still has a significant influence on the financial market. Their expectations for interest rate hikes or cuts can have a serious impact on risk markets. For example, recently, when the CPI data came out poorly, the risk market adjusted its pricing accordingly because the market prices in advance. They trade based on expectations; they cannot wait until interest rates are actually cut to rise, nor can they wait until rates are actually raised to fall. There will always be an advance expectation, and those futures traders or options traders will price in their overall judgment of the market. So this is also something I pay attention to, but my macro analysis is not as in-depth as my technical analysis or on-chain data; this area is relatively my weak point. Finally, there is the news aspect or fundamental aspect that Alex just mentioned, such as strategic reserve news. This part actually goes back to what I mentioned at the beginning about what I enjoy doing, which is designing event-driven trading strategies. This involves making trades based on specific events that have a higher degree of certainty. For example, around late May last year, a senior ETF analyst at Bloomberg named Eric, whose posts were highly anticipated in the market, suddenly tweeted at around 3 AM East 8 Time that the probability of an Ethereum ETF passing had been adjusted to 75%. At that time, the entire market was expecting that the Ethereum ETF would not pass. Once this news came out, Ethereum rose 20% within 24 hours, surpassing Solana in terms of value increase, which was impressive. When such news appears, the first thing I think of is to start preparing to find a time to enter an event-driven trade, which means preparing to go long on Solana while shorting ETH. The background is quite simple: the whole world knows that the ETF is going to pass, which is a very positive signal, so Ethereum will immediately rally. The real question to consider is who is next? Given the market environment at that time, the support or buzz for Litecoin and Dogecoin was not as high as for Solana. So, I first targeted Solana, and about a week later, I began to set up a long-short trading opportunity for Solana against ETH. In simple terms, I used contracts to go long on Solana and short ETH to capitalize on the price increase between the two. I believed that the next speculative target would be Solana because Ethereum was already a confirmed fact. If Ethereum really passed, Solana would definitely see a related increase. Some might question whether this idea holds up under scrutiny. I cannot say 100%, but one clear example is in January 2024. I do not know how many people noticed that on the day Bitcoin ETF was approved, Ethereum surged, and the exchange rate also skyrocketed. If I remember correctly, the ETH to BTC exchange rate increased by about 30% within 24 hours. Many people wondered what Ethereum had to do with the Bitcoin ETF passing. The next speculation would be Ethereum. So this is one type of event-driven trading. Returning to Alex's question, I think focusing on news or fundamentals is too difficult to quantify, so I personally prefer to design event-driven strategies to respond to potential opportunities in the market where there may be inefficiencies in pricing.

Alex: Understood. Thank you, Colin, for your very logical and organized explanation. He clearly articulated the thought processes behind each operational strategy, including the scenarios in which they might be applicable. It is evident that he has a very rich toolbox and knows which tools to use in which scenarios, rather than making vague decisions based on feelings.

Daily Life of an On-Chain Data Researcher

Alex: So, for the final question, as a trader and an on-chain data analyst, what does a typical workday look like for you? Besides focusing on on-chain data, what other information might you look at or what tools might you use?

Colin: Alright, this question is quite interesting because my typical day is rather boring and monotonous. My schedule is not very regular, but I try to stay awake during the U.S. stock market opening hours. The reason is simple: the liquidity in the crypto market is usually the best when the U.S. stock market opens. If my energy allows, I will look for short-term trading opportunities during this time. This has actually been a habit I developed several years ago. If I am really tired during the day, I will take a short nap because the chances of missing trading opportunities during the day are relatively low, while the chances of missing them at night are higher, and monitoring the market is more valuable at night. You may notice that every weekend or during the day on weekdays, especially during Asian daytime, the market tends to be quite boring, with most cases being sideways movement, low trading volume, and poor liquidity. This is why I try to stay awake at night.

After I wake up in the morning, besides observing the on-chain data for any changes, as Alex mentioned, I will also look at and record some additional data that I want to monitor. In addition to the candlestick charts, I will regularly scan through all the trading assets I usually follow. I also manually record the net inflow and outflow of Bitcoin and Ethereum ETFs in the U.S., as well as the market's volatility and the fear and greed index, as it is another quantifiable indicator of market sentiment. Additionally, I check the open interest in the futures market. If there is an extreme surge or drop today, I might also look at the liquidation volume. I record all this data because I am quite sensitive to it. The remaining data involves checking for any additional events that may occur; once they happen, I want to see if there are any changes in the data. Generally, the fixed data I mentioned includes futures market open interest, market volatility, the fear and greed index, and the net inflow and outflow of ETFs.

Another piece of data I like to monitor is the pricing of contracts on Coinbase relative to mainstream exchanges like Binance and OKX, checking for any premiums or discounts. I believe this can also be a quantifiable sentiment indicator, reflecting the sentiment of U.S. funds. For example, if there is a significant premium on Coinbase, it indicates that their buying pressure might be stronger. This was very evident when Trump was elected. I observe these numbers daily, maintaining this sensitivity, and once I notice any anomalies, I start to think about whether it is baseless or if there are trading opportunities within it.

Aside from the time spent recording this data, I also monitor the market because, as I mentioned earlier, technical analysis is one of the few areas where I can boast a bit. I spend a small amount of time, say a few hours, monitoring the market and checking whether my daily planned and adjusted trading strategies have reached my expected positions. If they are close to or have already reached those positions, I will focus intently on the market, looking at the data I want to see or checking if my trading plan has deviated and needs adjustment. I have two screens; on the other screen, I keep Twitter open, managing my own Mr. Bagel account there.

Outside of trading, my life is quite boring. I occasionally go for a run, but not very frequently; the purpose is just to keep moving and not be sedentary all day. The rest of my time is mainly spent with family. So, my day is quite dull, with nothing particularly noteworthy because trading is essentially my job. Therefore, I am not very different from regular office workers or students; I mainly work, then have meals, and sleep. That’s about it.

Alex: Understood. Colin just shared his daily work, and the amount of information and mental effort involved is quite significant. However, he seems to have systematized and modularized it, allowing his brain to engage in a series of important tasks, including data tracking, without needing a special startup each day. He has habits for what to do during each time period and a clear arrangement that helps him enter a state more quickly. We can also observe that Colin has a strong curiosity about trading, investing, and the business world. He gains not only money from it but also a lot of enjoyment. I feel that this state is an important trait for a good trader and investor. Thank you, Colin, for coming on the show today and sharing so many thoughts and systematic explanations about on-chain data analysis, investment, and trading. I hope we can invite Colin again in future episodes to share more knowledge on other topics. Thank you, Colin.

Colin: Alex, you are too kind. I’m just sharing my personal views. Thank you.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。