Is AI taking over trading more than software engineering?

We all have heard that algorithms dominate the markets. Some statistics say it used to be around 70% 1. And that was even before ChatGPT became available. AI existed long before Chat GPT came out. It existed in finance but also in many startups and internet services as well as gaming, in some healthcare tech and virtual assistants like Siri.

AI was already working in Google, Amazon and Netflix recommendation engines. But all this existed without much ability for users to prompt it. It was directed by the companies. What we could do was tell if we liked the suggestions they provided or not and our viewing or shopping habits enforced their learning algoritmhs to suggest us what to watch or buy next.

Now with Chat GPT and other AI platforms, it’s evident that AI is growing in finance. And considering that even before that, AI in financial markets was huge, it’s about to get worse. Worse for who though? Well, everyone - starting from beginner traders, to retail traders. While the best proprietary algos still outperform AI algos, the competition is getting more fierce.

We know that for many jobs, either in software, customers service or content creators, its difficult for both beginners or even people with intermediary competence to get jobs. In trading, it’s probably the worst, even though it’s not that talked about. Why? The field is already competitive and dominated by algos and big institutions.

Pre-GTP and Post-GTP AI differences in financial markets

Ok, so we know there was AI before and after. How are they different?

Before hedge funds or institutions were using machine learning and data analytics. Infrastructure needs were very specific. All data used was structured (e.g historical price data).

After, companies and individuals started using news, social media, and other data for making decisions on the go, fast. Large Language Models (LLMs) were able to take on multiple tasks, but introduced “hallucinations”, where you cannot always be 100% sure the AI suggested move is right.

So before the players were hedge funds and quants. Now there are also analysts, retail platforms, fintech.

Sometimes lose, always win

These are just few areas where AI has an edge:

• Prop algos are harder to invent.

• AI trading means more than just algos, it’s also about speed, working with huge datasets, physical server locations.

• Common trading strategies are arbitraged away - everyone running the same simple algos will lose the edge - in trading it’s all about having an edge. Read our article on Mastering the psychology of trading.

Most what is claimed to be AI trading is already behind - startups that offer AI trading, use simplistic models and settle with random performance, just to say it’s AI. We have seen it a lot in AI startups where businesses create a product claiming to use AI, where in reality they are either GPT wrappers or in worst cases, use actualy human workers instead.

That’s not to say all AI is garbage, it’s about separating the good seeds from the wild oats. The best AI traders and institutions have a huge edge over everyone else. With competition this tough, everyone who doesn’t know what they are doing, are being pushed out.