Sophia Don’t Fail Me Now: One Bank Worries ‘Algo Rally’ Will ‘Run Out Of Steam’
On Monday, we noted that markets are now "torn", with investors pulled in one direction by an express desire to extend the YTD risk surge catalyzed by the dovish Fed pivot and yanked in the other by ongoing signs of economic malaise (China’s December trade data), the US government shutdown and the ever present threat that someone, somewhere is going to step on a geopolitical landmine (e.g., Brexit, China vs. Canada, Turkey vs. Kurds).
Tuesday brought more in the way of conflict
“Of course algos aren’t “thinking” about any of that…”
There is no way of knowing what a neural-network machine-learning algo is “thinking” or “not thinking”, since the training happens in a black-box. We know ‘what’ the AI does, but not ‘why’.
CTA models are incredibly predictable to those who built them. You’re talking about something else entirely.
If they are not using machine learning, then these algos will be predictable. Predictable, and retrograde.