Artificial intelligence is increasingly being used by traders and investors to research companies, analyse financial results and identify potential opportunities in the stock market.
But how good is AI when it comes to actually picking winning stocks?
Having already started a live experiment to see whether ChatGPT can beat the market, we thought it would be interesting to put one of its biggest rivals to the test – Claude, the AI assistant developed by Anthropic.
Claude has built a reputation for its analytical and reasoning capabilities, but can it use those abilities to identify profitable short-term trades?
That’s exactly what we’re going to find out.
For this trial, we are asking Claude to select its best trade of the week from stocks listed on US exchanges, providing a detailed explanation of why it believes the chosen company could perform well over the following five trading days.
There are a few simple rules to keep things fair.
Firstly, the company must have a minimum market capitalisation of $500 million. We also want to avoid illiquid penny stocks or obscure companies that an ordinary retail trader might struggle to buy and sell.
Secondly, Claude must make its selection over the weekend, before the US stock market opens on Monday. It can analyse recent company news, earnings announcements, analyst upgrades, share-price momentum, upcoming catalysts and any other relevant information to arrive at its decision.
Once the stock has been selected, however, the trade is locked in.
We will record the official opening price on Monday (or the first trading day if Monday is a US public holiday) and then compare it with the closing price on Friday.
There will be no changing selections halfway through the week, introducing stop-losses retrospectively or conveniently ignoring trades that don’t work out.
Each result will be recorded in a spreadsheet, allowing us to monitor Claude’s overall profit or loss, strike rate and average weekly return as the trial progresses. We will also compare its performance against the S&P 500 over the equivalent holding periods to see whether it is genuinely outperforming the broader market.
Claude vs ChatGPT – Which AI Is the Better Stock Picker?
One of the most interesting aspects of this experiment is that we will be able to compare Claude’s selections directly with those from our ongoing ChatGPT stock-picking trial.
Both AI models will be given essentially the same task, with identical trading rules and holding periods.
Will they identify similar opportunities, or will their different approaches to analysing financial information lead them towards completely different stocks?
And, more importantly, will either demonstrate an ability to consistently outperform the market?
Of course, making a convincing argument for buying a stock and actually making money from it are two very different things. A company can report outstanding results, receive multiple analyst upgrades and still see its share price fall if investors were expecting even better news.
That’s what makes this a useful real-world test rather than another hypothetical demonstration of what AI might be capable of.
We will publish Claude’s selections and update the results each week, including the losing trades as well as the winners.
So, can Claude beat the market – and perhaps even outperform ChatGPT?
Follow our live results below to find out.
This experiment is for educational and entertainment purposes only and does not constitute financial advice. Returns are calculated using share-price movements and exclude dividends, dealing fees, slippage and taxes.
Leave a Reply