Bloomberg reported on this earlier in the week, as did other outlets.
- JP Morgan analysts are experimenting with a ChatGPT-style language model to divine policy signals in Federal Reserve statements and Fed officials’ speeches.
The JPM economists found that the AI tool can be useful in potentially predicting changes in policy — and give tradeable signals.
- For instance, they discovered that when the model shows a rise in hawkishness among Fed speakers between meetings, the next policy statement has gotten more hawkish and yields on one-year government bonds advanced.
-
“Preliminary applications are encouraging”
—
I’m not sure you need much AI to do this? Fed speakers, like many of those from other central banks, are like parrots a lot of the time, each one of them repeating the same mantra.
And another thing, to those who’ve ever developed technical analysis models. ya gotta be really careful of not back fitting to the data. After the GFC the ‘trading signals’ were pretty much BTFD, over and over and over again.
—
Remember this one? It was immediately after the ‘transitory’ parrot had been lain to rest.
