“You think the rise and fall are just a game of market buying and selling power, but AI can see higher-dimensional things.”
Organized by: Peng SUN, Foresight News
“I successfully escaped from the top at the beginning of 2022, and successfully bought the bottom at the end of the year. When Bitcoin was only over 20,000 US dollars, I firmly saw more than 40,000 US dollars.” As a “dog gambler” in the cryptocurrency market that can’t predict the rise or fall every day, the author When I see these achievements, I can’t help but feel jealous.
But in fact, these judgments are not made by humans, but are decisions made by artificial intelligence based on data.
Like most practitioners involved in the Web3 industry, Everest Link Capital AI strategy fund manager Xin Liu also developed a strong interest after hearing the story of Bitcoin for the first time, but her major is more in studying the market. .
Just like “Alpha Dog” defeated the human Go master, AI can execute trading strategies without emotion through massive data analysis and learning, and can find clues in the K-line clues that humans may not be able to find. As Xin Liu said:
“You think that the rise and fall are just a game of market buying and selling power, but AI can see higher-dimensional things.”
**Foresight News: What are the backgrounds of the current team members? When did they start trading cryptocurrency? Why did you choose cryptocurrency as an investment target? **
Xin Liu: I started to join the blockchain industry in 2016, and that’s when I started developing the “Night Navigation Star” AI deep learning model. It has been 7 years so far. The model is mainly used in cryptocurrency and stocks. The market has gone through many rounds of bull and bear markets, and has also experienced many extreme black swan markets.
In 2016, cryptocurrencies, or Bitcoin, were still excluded from mainstream finance, because I came from an asset management background. During the asset allocation process, I was looking for some alternative assets in the hope of reducing the overall risk of the portfolio. In 2016, I listened to an in-depth introduction to Bitcoin for the first time at the IC Cafe at Zhangjiang Hi-Tech Subway Station. I very much recognized the two points of decentralization and incentives.
Why do I, a person who grew up in the traditional financial system, accept decentralization so quickly? Because I have a special experience when I was doing macroeconomic research at my old employer Reuters, I established China’s first decentralized community of fixed income traders from scratch, so I deeply understand the implications of decentralization. The importance of great energy and motivation.
So I wrote a research article in “Wall Street Insights” on August 14, 2017, studying the positive impact on the entire asset portfolio after investing 5% of assets in Bitcoin. Later, Andrew Ang, the great god in the asset management industry and chief strategy officer of Black Rock, elaborated on the same point of view in “Asset Asset Allocation with Crypto: Application of Preferences for Positive Skewness” in 2022.
**Foresight News: What criteria are used when selecting specific trading targets? **
Xin Liu: In fact, my choice of track is very open. The biggest feature of this market is that it innovates very fast, and it is precisely because of the huge vitality that fast innovation brings to it. Open is embodied in reducing preconceived prejudices and trying to accept more innovations under the premise of high winning rate and high safety factor. Generally, as long as a certain type (whether it is “classical” or “local dog”) appears with a certain scale Consensus, I will not let it go and will study it. There are many people who support and like something, no matter it looks like a ballet or a duet, it must have its own uniqueness that satisfies human nature, and its existence is reasonable.
In addition, it must have a sustainable and extended incentive mechanism. I value incentives very much, and the incentive mechanism should not be complicated. If it is simple and clear to ignite individual desires, I will approve it more. You can see that successful projects, whether they are public chains or applications within the ecosystem, have this characteristic: everyone adds fuel to the fire, especially in the world of web3.0, you must have a mechanism that can ignite the inner fire of each individual.
**So you can see that my compatibility is very strong. So why can we cover such a wide range of innovative varieties with high risks? Because AI deep learning provides precise control of price trends, this greatly improves the winning rate and safety factor of accepting innovation. **For example, the innovative strategy we are currently developing uses AI deep learning models to find opportunities for inscriptions and memes. Although inscription has become popular recently, practitioners also have to face 1) the risk of unsuccessful mint inscription projects 2) high gas fees during the mint process 3) the risk of mint being buried if unsuccessful 4) not being able to compete with automated mint studios, etc. question. So which promising inscription varieties mint to choose? If there is no mint, at what price can the success rate be higher? This is a problem we are solving with AI deep learning, and the same is true for meme varieties.
**Foresight News: Can you give us a detailed introduction to the trading strategy based on AI deep learning? What parameters are considered in the deep learning model? What indicators were consulted? What kind of logic is the trading direction essentially based on? **
Xin Liu: AI deep learning is simply a way of using artificial intelligence to imitate human neurons to think. The picture is the simplest model. The inputs layer represents the input information, and the output layer refers to the final output information. , the middle layer is collectively called the hidden layer, and the hidden layer is extremely complex and exquisite. Instead of humans manually defining features, AI deep learning can automatically learn and extract features from raw data. Common areas of study include price trends, trading volume changes, market sentiment, etc. Therefore, the nonlinear and adaptive characteristics of AI deep learning, coupled with the ability of computers to process large-scale data, enable the neuron system simulated by AI to have the ability to surpass human understanding of the nature of things in some fields.
As for the specific parameter design, it is inconvenient to disclose too much because it involves commercial confidentiality.
**Foresight News: The rise and fall of a financial product is essentially the result of the game of buying and selling power in the market. Why is this theoretically disordered event predictable? Is it a kind of “metaphysics” for traders to predict future rising and falling trends? What is the scientific basis for it? **
Xin Liu: I have read a paper “Observing Schrödinger’s Cat with Artificial Intelligence: Emergent Classicality from Information Bottleneck”. I used artificial intelligence to observe the famous Schrödinger’s cat. The conclusion is that the essence of gods, humans and lower creatures is The difference lies in the ability to process information. This is the best answer to both of your questions.
For example, you mentioned that “the rise and fall of a financial product is essentially the result of the game of buying and selling power in the market.” From this perspective and logic, the market cannot be predicted in advance. This is the normal causal perception of ordinary people, right? But there is another possibility that the ups and downs game is not like this at the level of higher-dimensional information processing capabilities. In fact, in the dimension of AI deep learning, what I can see in May is that the price of Bitcoin will rise to more than 40,000, and subsequent events that are conducive to this conclusion will come out to catalyze the completion of this conclusion. I don’t know the specifics in May. What events will lead to the completion of this conclusion, but there must be corresponding events. Can you see the difference in the order of cause and effect? **This is because AI’s ability to process information is much higher than that of humans, and it can see things from higher dimensions that ordinary people cannot see, thus forming prophecies with a “metaphysical” nature. **
**Foresight News: In the communication, you mentioned that you felt that the “four-year bull-bear transition rule” that appeared based on the halving cycle in the past may become invalid in this round. Why did you make this judgment? **
Xin Liu: First of all, let me correct you. What I mean is that Bitcoin has had halving cycles in the past and now, so the bull and bear cycles are very regular, so it makes it easier for ordinary people to judge bull and bear. There is a bull market before and after the halving, and there are two years of bear market left. ,Right. But the fourth halving will end soon. A large proportion of Bitcoins will enter circulation, and the remaining mineable parts will be very small. Based on common sense, it must be the large circulating part that affects the small mineable part, so The bulls and bears of the future will be very complex and not as clear-cut as they are now. So after this round of halving market is over, what logic do people use to judge bulls and bears or big cycles? I think AI deep learning will become an important basis for judging long-term bulls and bears.
**Foresight News: As the number of participants in the cryptocurrency market increases, the market trend begins to become complex. It can be seen from this year’s trend that it is no longer a relatively simple cyclical trend like in the past. In the future, it may In more complex market conditions, how can the AI deep learning model help the market judgment? Compared with some “theories” used to judge market trends, what are the specific advantages of using AI? **
Xin Liu: You mentioned that the market trend is becoming more and more complex. Cyclical trends may exist, but they are by no means simple. Bitcoin bulls who have experienced the market in March 2020 should be deeply touched by this, plus the subsequent halving. In conclusion, it is foreseeable that the threshold for judging the bull and bear market will be higher in the future. The advantage of AI is to discover trends that are not visible to mortals from high dimensions through its powerful ability to process data. For example, do you know when we judged the bottom price of this bull market? The bottom was first proposed at 16000 in August 2022. We reminded the bottom a total of 4 times in August, September and November 2022. But when 16,000 is near, anyone who has experienced it knows what it is like. It is very pessimistic. I have been told that large institutions have seen 11,000 several times, but what our own model saw at the time was that there was no room for decline below 16,000. . Just in October and November this year, AI discovered two stocks, Estee Lauder and Dollar General. At that time, the market was very pessimistic about these two stocks due to negative financial reports. Until December, when Goldman Sachs released its 2024 stock market outlook, these two stocks were the top two stocks on the to buy list. We discovered them 2 months earlier than Goldman Sachs. **In short, it is to practice what Buffett said: “If others are greedy, I will be afraid of others and I will be greedy.” **
**Foresight News: Some people in the market believe that trading and investment are two different things. Trading may ignore fundamentals to some extent and only focus on the information given on the market to judge the market trend within a certain period of time. , while investment may be more inclined to long-termism and focus on the development of industries, tracks, and projects. What are your views on “trading” and “investing”? **
Xin Liu: I think “investment” and “trading” are not contradictory things, especially since the mid- and long-term trading strategies and investment strategies like ours are consistent. If the trading you are talking about refers to strategies such as high frequency and arbitrage, then I admit that they are very different.
Investment and trading are both important abilities for us to survive in the market. The reason why some people think they are contradictory is because it is difficult to possess these two important abilities at the same time. But with the help of AI, we have achieved both investment and trading capabilities, so in the future you will see that AI will greatly improve individual productivity.
As we all know, the risks of crypto assets are very high. On the one hand, the risks come from the fact that projects are faced with the public at a very early stage, or even when there is only a story blueprint, so we must face the risk that the project may fail. Second, innovation and iteration in the encryption world are very fast. Long-termism in the encryption world needs to be compatible with rapid innovation. Third, even on an excellent track, token prices will fluctuate greatly. Therefore, as asset managers, we are charged with the mission of creating profits for LPs, which is to balance long-termism and flexibility in investment, so as to ensure that we can create positive and stable returns for LPs in the long term. So I don’t think there’s any need to debate whether it’s better to go the investment route or the trading route, it’s best to do both.
**Foresight News: What is your basic judgment on the trend of the cryptocurrency market in the next few years? **
Xin Liu: Let me briefly talk about my four judgments:
- Bitcoin’s trend will be more complicated
The reason is that after the fourth halving, the long-term law of the Bitcoin market will no longer be as easy to judge as it is now. The halving is like a magic needle that dominates the current market trend, but after the halving is over, there will be more factors that crypto practitioners have never encountered before to dominate the market. In layman’s terms, this cycle is the last opportunity for ordinary people to make money easily, and it all depends on strong and in-depth research capabilities.
- Cryptocurrency investment will become more decentralized, and super individuals in the AI era will replace institutions or coexist with them.
The world today has moved towards disintermediation, especially in the crypto world, where information asymmetry will become less and less common. Investors are becoming more and more mature, have their own independent judgments, and are increasingly unwilling to take over VCs. Investment, whether it is a super individual or an institution, is composed of people. In essence, the core is to provide superb cognition and judgment. Especially in the AI era, individual productivity will be extremely amplified. Perhaps what one person can do is now What an institution does, so the key to judging credibility in the future is whether it can output correct investment judgments and results in the long term, and it does not stick to whether it is given by an institution or an individual.
- The trends of mainstream currencies and “small varieties of currencies” diverge
With the Binance incident settled, it means that regulation is on the fast track for compliance requirements for centralized exchanges. For many fledgling products, the cost of compliance is very high, and the cost of dealing with potential lawsuits It is also very high. This also means that the risks of the project sponsors themselves increase, which will squeeze the living space of small varieties in centralized exchanges. On average, small varieties will get less traffic, and mainstream coins will get more traffic. many. In the future, it is unlikely that altcoin chickens and dogs will rise to the sky in 2017, but funds will wander through several small varieties in stages.
- Decentralized exchanges and wallets are booming
The charm of the encryption industry is that every new participant, no matter when you join, as long as you have a unique vision, you will definitely have your own stage and right to speak, instead of latecomers having to take over for the old vested interests. Therefore, the encryption world is absolutely attractive to young people around the world. This all depends on the continuous innovation of the encryption world. And if small varieties that represent innovation are squeezed on centralized exchanges, they will inevitably gather in a new location, and decentralized exchanges and wallets are the best choice.
