If you ask a large language model (LLM) like ChatGPT how to choose an LLM, it will give you an answer such as this from GPT-4:
In the world of finance, the principle of supply and demand serves as a foundational mechanism for determining the fair price of any asset class at any point in time. This economic concept holds that the equilibrium price of an asset is established when the quantity demanded by buyers matches the quantity supplied by sellers.
Fundamental factors have long played a crucial role in assessing traditional equity markets, where investors analyze a company's financial health, industry position and overall economic climate to determine its intrinsic value. Key metrics such as earnings, revenue and debt-to-equity ratios provide a clear picture of a company's performance, enabling investors to make buy/sell decisions. However, such metrics are not available yet in the rapidly evolving world of cryptocurrencies.
The absence of financial statements and difficulty of estimating the impact of emerging technologies makes it hard to value cryptocurrencies by traditional pricing methods. Moreover, the extreme price volatility further challenges the efficiency of fundamental analysis in the crypto space.
In the absence of traditional valuation methods, the price often seems to be determined by the sentiment around the overall crypto market and/or a particular token. The perception and emotional reactions of market participants often play a more prominent role in driving price fluctuations and shaping investment decisions.
For a rational trader, such irrationality presents an opportunity in the market – if only she could quickly and accurately capture the mood (aka sentiment) of the market. For many years, working with sentiment seemed like an insurmountable challenge. Day traders mostly relied on crypto news headlines, Discord insider chats and announcements. And systematic traders had to invest considerable effort into development of just average-quality sentiment analysis tools. The limitations of technology at the time made it difficult to efficiently process and understand the vast amounts of data generated by global media.
The revolution in transformers and LLMs, specifically, allowed traders to approach sentiment at scale, delivering an incredible improvement over traditional methods that relied on manual scoring and Word2Vec models.
The competitive landscape of software-based technology companies vying to create the best LLM is rapidly evolving now. The table below provides an impressive illustration of this ongoing race, showcasing some of the key players and their respective contributions to the field:
These LLMs continue to increase in size and improve performance, surprising even their creators. And while people debate about whether LLMs are the first signs of artificial general intelligence (AGI) or just mindless parrots, their use in different industries and finance in particular will only accelerate.
The potential revolution brought about by transformers and LLMs could significantly transform the crypto trading landscape. With the capability to assess market sentiment on a larger scale, traders might be able to capitalize more effectively on market irrationalities.
You can learn more about LLMs, their types and applications in our most recent white paper, available here.