On Sunday morning, in response to the geopolitical shock of the US bombing Iran, I turned to ChatGPT’s Deep Research tool.
I asked it to spend three hours analyzing the situation to come up with five US stock trades that would make money through this Thursday. I did this before Asian markets opened, so the tool had to make predictions about the future with little to no clues from real-time market data.
This is a real test of whether AI models and chatbots can generate valuable new insights, or whether they just regurgitate what’s already known on the internet.
Deep Research produced a well-written report predicting a global “risk-off” reaction that included a drop in US equities, a spike in oil prices, along with a move into safer securities such as US Treasuries, the dollar, and gold. It also gave five specific trades designed to make money from Monday, June 23, through Thursday, June 26:
ChatGPT
Buy Lockheed Martin (LMT). ChatGPT expected the defense sector to benefit from increased military tensions, leading to higher anticipated defense spending.Buy Exxon Mobil (XOM) or the XLE oil sector ETF. With oil prices likely to spike on concerns about Middle East supply disruptions, ChatGPT advised a long position in energy.Buy gold. Investors typically seek safety in gold during geopolitical crises, so ChatGPT recommended going long on the GLD ETF.Short Delta Air Lines (DAL). Airlines are sensitive to fuel costs and traveler sentiment. ChatGPT flagged Delta as vulnerable due to an expected increase in oil prices and consumer anxiety around flying.Short the S&P 500 via the SPY ETF or futures. As a broader market hedge, it recommended shorting the S&P 500 to profit from the likely risk-off move and overall market decline in the wake of the bombing.
Around the same time on Sunday morning, tech analyst Dan Ives emailed investors with his thoughts. He advised clients to buy leading tech and AI stocks. He also inadvertently came up with a new acronym for this suggested group of tech stocks, proposing specifically Tesla, Amazon, Microsoft, Palantir, Oracle, and Nvidia (TAMPON).
So, I took Ives’s advice as a benchmark on Sunday ahead of the market open. This set the stage for a fascinating AI-versus-human performance showdown. Could ChatGPT’s Deep Research recommendations outperform the TAMPON trade?
Here are the results, as of the close of trade on Thursday, June 26:
Lockheed fell more than 2% in the period. WRONG.Exxon dropped more than 3%. WRONG.The XLE ETF lost over 3%. WRONG.The GLD gold ETF shed about 1%. WRONG.Delta Air Lines rose almost 3%. WRONG.SPY S&P 500 ETF gained more than 2%. WRONG.
Ives’s TAMPON trade:
Tesla climbed about 1%.Amazon gained nearly 4%.Microsoft rallied about 4%.Palantir jumped 5%.Oracle was up about 4%.Nvidia surged nearly 8%.
Conclusion: A pretty damning indictment of ChatGPT’s predictive abilities.
Sign up for BI’s Tech Memo newsletter here. Reach out to me via email at abarr@businessinsider.com.
