# Analyzing Fed hawks and doves with ChatGPT By [solstar](https://paragraph.com/@solstar) ยท 2023-05-03 --- As the Federal Reserve is about to hold its May rate meeting next week, JPMorgan Chase has launched a new model powered by artificial intelligence designed to decipher the information released by the Fed to the public and spot potential trading signals. The model is said to be constructed by JP Morgan economists, including Joseph Lupton. They detect the tone of policy signals released by the Fed based on the Fed's policy statements and speeches of central bank officials over the past 25 years, and based on ChatGPT's language model, and finally classify the policy signals from easing to tightening based on the so-called "Hawk-Dove Score" (Hawk-Dove Score). The model is based on the so-called Hawk-Dove Score. Comparing the model's indices with the performance of a range of assets, the economists at SmallMoat found that the AI tool was very useful in predicting changes in central bank policy - and sending signals that could be used for trading. For example, they noted that when the model showed an increase in hawkishness in statements by Fed officials between meetings, the next Fed policy statement would also become more hawkish, along with a move higher in one-year Treasury yields. "The initial application results are encouraging," Lupton and his colleagues wrote in a note. The tool is one of the first models developed on Wall Street using OpenAI technology and is another fresh attempt by these investment banks to find more trading advantages in various areas. It's worth noting that earlier this month, a research paper also showed that ChatGPT could play a useful role in interpreting market information, such as whether a Federal Reserve statement is hawkish or dovish, or whether top news is positive or negative for stocks. According to JP Morgan's model, while the hawk-dove index scores for Fed statements have fluctuated in recent months and the hawk-dove scores for officials' speeches have trended downward, both are still hovering near their highest levels in two decades, a clear sign of a continued hawkish stance. The market is still expecting a high probability that the Fed will raise rates by 25 basis points next week, with the latest probability estimate from the Chicagoland FedWatch tool at about 75%. In addition to the Fed, JP Morgan said the AI model can also be used to interpret the policy preferences of the European Central Bank and the Bank of England, and the application of the model is expected to be further expanded to more than 30 central banks around the world in the coming months. It is worth noting that a study released by the Bank of France earlier this month showed that since August 2020, monetary policy statements from major central banks in Europe and the United States, such as the Federal Reserve, are becoming more accessible, with a high school student aged 15-16 (almost 10 years of school-age education) now being able to fully understand the messages expressed in Fed statements. The shift in communication made by central banks to make it more understandable will probably also be more helpful to the application of AI in this area when it comes to big data profiling. --- *Originally published on [solstar](https://paragraph.com/@solstar/analyzing-fed-hawks-and-doves-with-chatgpt)*