Crypto traders are using artificial intelligence (AI) to help them turn a profit. Projects are rapidly building AI tools that can execute trades with a particular focus on permissionless, ostensibly decentralized exchanges (DEX).
Some of these tools are black boxes that use enterprise AI services, such as GPT-4 or Bard. Others use a combination of AI plus discretionary parameters.
Once set up with access to funds, AI agents can automatically search for the best place to file trades. Some limit their search to DEXs; others offer integrations with certain centralized exchanges.
Caveat emptor. Sometimes, trading bots can be disguises for Ponzi schemes. In particular, trading bots that require users to deposit capital in order to ‘autotrade’ on their behalf are often non-functional Ponzis.
AI trading bots are nothing new
Autotrading bots have been around for decades. The Bloomberg Terminal was one of the first platforms to enable computer-assisted trading in the 1980s. Also in the 1980s, the more affordable Tradestation became a popular trade bot platform for retail users.
Day traders often use bots for strategies like arbitrage trading. Foreign exchange (FX) markets popularized ‘pips per day‘ trading bots that used various forms of technical analysis to aim for a high win rate. FX bots made easy transitions into substantially similar crypto markets.
Of course, trading strategies with a high win rate are often exposed to enormous losses during their infrequent drawdowns.
To their credit, AI trading bots can spot opportunities far faster than humans, giving users a potential edge in a trading environment where prices can move fast. Nevertheless, computers are typically unable to process qualitative risks, such as fraud or incorrect data.
Similarly, computers are typically unreliable during periods of high volatility or when fundamental news is moving markets.
Previously, Protos covered some of the ways in which bots wreak havoc on DEXs through front-running or sandwich trading. Solutions like the mostly-automated Flashbots software promise to help, but sometimes wind up exploiting traders’ desire to outwit bots.
AI-powered trading tools could also ‘randomly’ inject a lot of money into one side of a market, throwing other traders’ strategies off balance. One Motley Fool writer found it easy to create one using ChatGPT’s code-writing ability. He could even optimize it for a specific exchange — Binance — and throw historical price data at it for backtesting.
Add a Twitter list and other data feeds for real-time sentiment about specific digital assets, and one could build a powerful AI trading tool today for nearly free.
AI-powered trading bots are already arriving on DEXs around the world. Some are enhanced with real-time data feeds and combine the capabilities of multiple AI services. Many of them will be unable to outsmart human traders over the long term, but some will produce promising results and attract large amounts of users and capital.