Colin believes that Bitcoin currently meets the conditions for forming a top.
Host: Alex, Research Partner at Mint Ventures
Guest: Colin, Independent Trader, On-chain Data Researcher
Recording Date: February 15, 2025
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 related to specific sectors 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 episode, we will mention many concepts related to indicators, and we will list these concepts at the beginning of the text version for easier 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): A measure of 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 the current market price of BTC and the Realized Price, the True Market Mean Price under the Cointime system also considers the impact of time, making it suitable for the price of BTC over a long cycle.
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. It is an improvement of 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 independent trader and on-chain data researcher Colin. Let’s have Colin say hello to 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 investor without any special title, quietly doing my own trading. My name is Colin, and I run an account on Twitter called Mr. Beggar, where I mainly share some tutorials on on-chain data, analyses of the current market situation, and some trading concepts. I see myself in three roles: first, as an event-driven trader, I often think about event-driven trading strategies; second, as an on-chain data analyst, which is also 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 level of 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 feel is lacking in my own area. I read a series of articles he wrote and 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 are highly subjective, 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 sharing and examples, 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 divided into 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 purely accidental. About six or seven years ago, I didn’t understand it at all; I was 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 study 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 this hard to define; I’ve never really learned it systematically. From the beginning until now, I’ve encountered a problem: I haven’t seen any systematic teaching. When I first discovered this field, it was several years ago. I noticed it but didn’t delve deeply; I just read a couple of articles to understand it. After some time, I came back to see more in-depth content, but at that time, I was focused on studying other things. I found this quite interesting and continued my research. There hasn’t been a specific time for systematic learning; it’s been pieced together like that.
Alex: Understood. How long have you been applying what you learned about on-chain data to 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 amount, 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 other platforms, but I ultimately chose it because it matched my preferences during the initial filtering and research.
Alex: Understood. After looking at a lot of Colin’s information, I also registered for Glassnode and became a paid member. I indeed feel that their data is very rich, and the timeliness is also 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: Okay. First, let’s talk about the value and principles of on-chain data analysis. I plan to combine these two points because it’s actually 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 on-chain, for example, 300 Bitcoins being transferred from one address to another, which can be checked on a blockchain explorer. Although I cannot know who is behind this string of addresses, that’s not important because no single individual can influence the 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, their profit status, loss status, 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’s like you are missing a very 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 very 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 been exposed to 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 certain indicator, it gives you a calculation formula, and my thought process is to figure out what this formula is really trying to convey and why it is designed this way. After I understand these indicators, the next step is to filter them. Those with experience in quantitative strategy development or who have studied indicators will know that many indicators have very high correlations. High correlation can lead to a problem where it is easy to generate noise in interpretation or over-interpret the data. For example, let’s say I have a system for escaping tops, and this system might have 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. Because 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 actually not very reasonable because it was bound to happen. 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 indicators have high correlation, I will slightly filter them and ultimately select one or two.
This first part is relatively easy to solve and is not the main difficulty. The second part is the real challenge: how do you prove your viewpoint is correct 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 a formula. 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’s 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’s 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 if the sample size is sufficient, this strategy cannot be used. Some people might argue that it has been backtested 1,000 times with a 95% win rate, and the backtest results support that this strategy can be used. 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, when 1,024 people do this, 1 person will succeed, and the situation of getting heads 4 times in a row is what we call a survivor; the other 1,023 people who tried this failed, and we don’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 a formula 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’s 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, given 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 viewpoint 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 ahead, which 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 still very young, on-chain data analysis will always face the problem of insufficient sample size, which means that in research, you have to rely solely on deductive reasoning and use logic to infer it, then wait for time to validate your judgment. This is the biggest difficulty I currently face.
Key On-chain Indicators to Focus On
Alex: Understood. 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 want 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 adopted as our main indicators. Based on your insights, what on-chain indicators do you consistently monitor or consider important in your current analysis of Bitcoin?
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 quantity of Bitcoin was 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 know 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 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, it can 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 can easily 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. 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 over $90,000, so any significant changes would indicate a decrease, not an increase, because the current price range is over $90,000, not $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 corresponding to the current Bitcoin price. For example, how much you are earning, whether it’s not much or a lot, is the general concept. The principle of this indicator is very simple; through the transparent mechanism of 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 it’s $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. For example, if today RUP is high, such as 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 was 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 straightforward: bottom fishing. Suppose during a bear market, the price keeps falling and eventually drops below the valuation given by 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 favorable price. Historical backtesting shows that whenever the price drops below Cointime Price, it is usually a very good bottom fishing opportunity. The second application is escaping 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.
How to Handle Conflicting Data
Alex: Okay, that was very clear. Many users might ask a question: the three indicators you listed may represent different aspects and align with what you just said 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 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. In this case, how would you handle the 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 profitability status also 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 cycle of Bitcoin, 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 profitability 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. 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 significant, 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 a portion of your position based on the weightings, 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 a top as a gradual process. Although it may seem quick when looking at daily charts, if you were in the moment, for example, at $69,000 during the last cycle's top, you wouldn't feel that it was the top. We can only make a judgment based on data that the conditions for forming a top may be present. Therefore, 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, 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 be over, 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 That Judgment
Alex: Understood. I have been following your Twitter account recently, where you regularly apply the indicators we just discussed, along with 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 heads-up: I am actually very bearish on 2025. I believe BTC is currently in a position where the conditions for forming a top are present. I know many people, including some participants around me, who have not performed well 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 trader friends of mine 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 noticeable difference: 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 cycle in 2024 to be a bull market, it is actually very different from previous cycles. There is also a theory that a bull market must end with an altcoin season, but I personally believe that you cannot say that an altcoin season must occur for a bull market to end; this is clearly not strongly correlated. Therefore, 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 infer today’s market is akin to following a formula without understanding the context, which is not ideal. If you were to follow a formula, the tops in 2013, 2017, and 2021 should have appeared around the end of the year based on timing.
I personally believe that we are currently in a position where the conditions for forming a top are present. 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 represented by the URPD chart. We can observe that the low-cost chips accumulated in 2022 and 2023, when a large amount of BTC was bought at low prices, have seen a significant amount of distribution up to today. To put it simply, they have sold off; they are no longer participating. Some listeners might wonder, "What does their selling have to do with me?" There is a concept that needs to be explained: at the end of every bull market, it is almost always due to the completion of distribution of those low-cost chips, which leads to the end of the bull market. A less intuitive point here is that it is not because they are dumping 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 ends. This is not just a random thought; there is a logic behind it.
Assuming that every BTC chip participating in the market today 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 fluctuation between $70,000 and $50,000 last year, or the current range of about $90,000 to $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, that is actually less than a 10% drop. However, the pressure on these chips is substantial, and many of them are short-term traders. Once the pressure builds up, they may choose to sell, which would lead to further price declines. This decline would then cause other high-cost chips to also 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 profitability status. If you are interested, 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 extremely high; they almost move together. This is quite reasonable because the higher the price, the higher the holding cost and profitability 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 conditions have 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 instance, if you bought one Bitcoin at $16,000 and it is now at $96,000, the floating profit on that Bitcoin is $80,000. However, if you bought Bitcoin at $86,000 and it is now at $96,000, the floating profit is only $10,000. Thus, the main contribution comes from those low-cost chips. Therefore, if 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 is higher, these low-cost chips have exited, converting some unrealized profits into realized profits, which is why RUP appears lower, 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 studied the stock market, they would know that there is a concept of valuation, specifically the price-to-earnings ratio (P/E ratio). There are many variations of this valuation method, and the indicator I refer to is called Shiller ECY. This indicator comes from Professor Robert Shiller at Yale University, and it 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 primarily 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 poorly, then Bitcoin, as a smaller player, will naturally not look good either. However, these are just my personal biases for your reference.
Alex: Alright, Colin just provided a very detailed explanation. Let's summarize his points briefly. He believes that the current price range meets many conditions that have historically indicated a top in valuation or price, including the distribution of chips, the status of unrealized profits, and he also referenced Professor Shiller's 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 delved deeply into this concept or system before. So, if a beginner were to ask you, "Colin, I find what you said 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, so far I have received dozens of private messages asking similar questions. 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 piece of advice: 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 views within these schools cannot withstand scientific scrutiny. They are purely inductive, lacking logic, and can easily fall 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. I believe it is suitable for beginners for a simple reason: most retail traders around us are not full-time traders; many 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 "watching the market," then the role of on-chain data trading 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 based on on-chain signals, such as buying or selling, is 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 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 significant changes in the data.
The second part is about how to learn. I mentioned earlier that throughout my learning process, I have not seen any free, systematic teaching materials to this day. 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, which I think are great, but the problem is that you still do not have a framework from 0 to 1. This makes learning quite painful because this indicator looks impressive, but 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 because I initially did not know which was good or bad. I looked at 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 genuinely help them see what they wanted. This takes a lot of time. After reviewing all these indicators, you need to filter them. However, 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. Therefore, my suggestion is that if you want to study a particular indicator, it is best to find the original author's article; try not to look 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 a weekly report based on various indicators, not fixed ones, discussing the current market situation and why they believe it is that way. You can see a variety of indicators 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 over 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 have a relatively large long-tail so-called spot allocation, and even during the bear market bottom, I might 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 can know whether the market is in the early, middle, or late stages, whether it is a bull market or a bear market, and it provides a significant directional guidance benefit.
As for other parts, I mentioned earlier that my other strength is in technical analysis. This part is quite complex, and I cannot elaborate too much 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 find a way to use technical analysis to refine where I should enter this trade. Just to give an example, this is not financial advice, but let’s say Ethereum is a buy between $2000 and $2600 because I believe it will rise afterward. If I were God and knew it would rise, I would just buy it. However, 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 have a set of measurement criteria.
Next, regarding macro aspects, 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 severe 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 expectations in advance. They do not wait until interest rates are actually cut to rise, nor do they wait until rates are actually raised to fall; there is always an anticipation. 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 deep as my technical analysis or on-chain data; this area is relatively my weak point.
Finally, there is the news aspect or fundamental news, such as strategic reserve news. This part goes back to what I mentioned at the beginning about what I enjoy doing: I design event-driven trading strategies. This involves targeting specific events to create high-certainty trading opportunities. 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, my first thought is to prepare to find a time to enter an event-driven trade, which means preparing to go long on Solana while shorting ETH. The background is simple: the whole world knows that the ETF is likely to pass, which is a significant positive, so Ethereum will surge immediately. 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 would go long on Solana using contracts 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 inevitably see a related increase. Some might question whether this idea holds up under scrutiny. I cannot say 100%, but a 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 nearly 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 find that 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 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 U.S. stock market opening hours. The reason is simple: the liquidity in the crypto market is usually 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 watching 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 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 market 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 occur; once they happen, I want to see if there are any changes in the data. Generally speaking, the fixed data I mentioned earlier includes futures market open interest, market volatility, the fear and greed index, and ETF net inflow and outflow.
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, specifically the emotions of people in the U.S. 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 something unusual, I start to think about whether it is baseless or if there are trading opportunities within it.
Aside from recording this data, I also spend time watching the market because, as I mentioned earlier, technical analysis is one of the few areas where I can boast a bit. I will 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—it's pretty much like that.
Alex: Understood. Colin just shared his daily work, and the amount of information and mental workload 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 get into 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 talent 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 am 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。
