The impact of AI on secondary market investment

02-25

This article is machine translated

Show original

With the release of Sora, AI has become popular again recently, and the AI sector has also gone crazy.

In the past two years, traditional financial institutions such as Bloomberg/Wind/Tiger Broker have also released their own vertical large model applications. In the Crypto industry, many products such as Dune have also integrated AI or simply released vertical GPT. I have experienced almost all of them.

Core competitiveness in the secondary market

Regarding the "impact of AI on secondary market investment", my personal opinion is: AI is useful, but the role of large language models is very limited.

The reason is: the core competitiveness of the secondary market is the crushing of other opponents in the market by methodology and cognitive level, rather than the crushing of productivity.

Good (quantitative) traders fully understand the market characteristics through long-term experience in the market, thereby forming a set of stable and quantifiable trading strategies. Traders with insufficient skills will be eliminated by the market due to stable losses. Since then, the market has completed the evolutionary process of "good money drives out bad money", and it will become increasingly difficult to make money.

Therefore, if you want to beat the market and make stable profits, you must evolve faster than the market and have a deeper understanding of other participants in the market. This is a methodological and cognitive victory, and it is also the core competitiveness of the secondary market. In most cases, the market consensus is a risk factor, so "looking for non-consensus" is the most important ability of a (quantitative) trader, and it is also a manifestation of a high level of cognition.

To go a little deeper, this ability is not directly related to whether you have studied CFA/metalworking. If you can make money by studying CFA/metalworking well, then it will be much easier to make money in the secondary market. A common sense in life is that in any industry, if you want to make money, you must have something that most people do not have. This is competitiveness. It can be resources/capital/cognition/methodology/experience, etc.

Therefore, all people/things/projects that try to help you make a lot of money directly or at a very low cost can be classified as either bad or stupid.

Let’s talk about AI. Judging from the most advanced large language model GPT4 that I can publicly experience so far, the model's understanding and cognition of the world is obviously inferior to humans. Its advantage is that it improves productivity and its level is close to that of undergraduates. Therefore, this is not in line with the core competitiveness of the secondary market we mentioned above. Therefore, when GPT’s knowledge and understanding of the world is close to or even ahead of humans, it will have a major revolutionary impact on secondary market investment.

And from the current point of view, the performance of large language models in the Crypto industry is still far behind GPT3.5, let alone GPT4. Therefore, there is a long way to go to use vertical GPT to trade Crypto or provide users with trading decisions.

What exactly is AI used for?

So is AI useless for secondary market investment? have. There are probably several directions:

1. Mining factors: A few leading hedge funds in the traditional financial market rely on the ML/DL model to mine factors. Compared with the traditional method of mining factors by relying on people, this model mainly wins in terms of the number of factors. Sacrifice factor quality, but this is not the mainstream of the industry, and it requires very high team capabilities.

2. Clean/process data: For example, use ML to optimize missing values/outliers in the data set, identify MEV Bot’s trading volume, etc.

3. Algorithmic trading: It is mainly used in the microstructure of the market such as handicap/order book.

4. Process alternative factors: analyze the content on news/social media and analyze whether it is Positive or Negative, and even rate it.

5. Use GPT to organize natural language data: For example, the financial reports of listed companies in SEC's Electronic Data Gathering, Analysis, and Retrieval are all in text format. Use GPT to organize and process them into structured data. This effect should be good.

This is probably what I can think of at the moment. If there are any that I have missed, please feel free to add them. Through the above scenarios, we can find several rules:

Scenarios 1, 2, and 3 were quite mature before GPT became popular.
4 and 5 mainly belong to the application scope of GPT. 4 I have not tried it, but the effect is estimated to be relatively limited. The first is that the proportion of alternative factors in multi-factor strategies is very low. The second is that the factors mined through this simple and crude + low-cost method will most likely not be easy to use. , if it works well, it will soon become ineffective. The role of 5 is to increase productivity.
All the above scenarios are based on the use of AI in a very, very, very specific small link, rather than directly using AI to help you with transactions/investments, because its level of understanding of the market is too low, and in addition, there is too much signal noise in the financial industry. The ratio is too low, and it is completely incomparable with fields such as intelligent driving.

Looking at AI products in the Crypto field from this perspective, Dune’s idea is relatively reasonable. It does not try to play the role of a robo-advisor to give you direct trading signals. Of course, this product is not positioned in this way. It is still We are using AI to help you improve production efficiency, because the threshold for writing SQL is indeed too high for ordinary users, although it is not very smart now...

About LUCIDA & FALCON

Lucida ( https://www.lucida.fund/ ) is an industry-leading quantitative hedge fund that entered the Crypto market in April 2018. It mainly trades CTA/statistical arbitrage/option volatility arbitrage and other strategies, with a current management scale of US$30 million. .

Falcon ( https://falcon.lucida.fund/ ) is a new generation of Web3 investment infrastructure. It is based on a multi-factor model and helps users "select", "buy", "manage" and "sell" crypto assets. Falcon was hatched by Lucida in June 2022.

More content can be found at https://linktr.ee/lucida_and_falcon

Mirror

Disclaimer: The content above is only the author's opinion which does not represent any position of Followin, and is not intended as, and shall not be understood or construed as, investment advice from Followin.

Add to Favorites

Comments

Relevant content