JPMorgan debuts AI model used to decipher Fed speeches for potential trading signals

• JPMorgan unveiled a new AI tool that can potentially uncover trading signals.
• The model does this by assessing 25 years worth of Federal Reserve speeches.

Ahead of the Federal Reserve's policy meeting next week, JPMorgan Chase unveiled a new artificial intelligence-powered tool that digests comments from the US central bank to uncover potential trading signals.

JPMorgan economists used a ChatGPT-based language model to assess the tone of policy signals from the remarks, according to Bloomberg, analyzing central bank speeches and Fed statements going back 25 years.

This model rates these comments on a scale from easy to restrictive, the report reads, referring to the gauge as the "Hawk-Dove Score."

The tool can also be used in predicting changes in central bank tightening as well, finding patterns, for example, between rising yields on the one-year US Treasury and the level of hawkishness from a policy statement.

"Preliminary applications are encouraging," JPMorgan economist Joseph Lupton, along with others colleagues, wrote in a recent note.

The Hawk-Dove Score, which can also be used for the Bank of England and European Central Bank, is on track to expand to 30 other central banks.

The Fed is expected to raise borrowing costs again next week, with the CME FedWatch Tool forecasting a 85% chance that the central bank will hike by another 25 basis points on May 3. This will be the 10th interest rate hike since March of 2022.