3 ms·
Actual paper: https://journals.aps.org/prresearch/abstract/10.1103/PhysRevResearch.3.013090 https://journals.aps.org/prresearch/abstract/10.1103/PhysRev... Mor
by Animats 4y ago
Actual paper: https://journals.aps.org/prresearch/abstract/10.1103/PhysRevResearch.3.013090 https://journals.aps.org/prresearch/abstract/10.1103/PhysRev...
More useful than Quanta Magazine hype.
The basic idea is that you've got a process with feedback that behaves like a chaotic attractor, moving around a lot but staying in a stable regime.
Where's the edge of that regime?
Here's a video of a leaky bucket waterwheel that exhibits chaotic behavior.[1] If all you had was a graph of rotational velocity, could you tell when it was about to reverse? Probably. Could you train a machine learning system to do that? Yes.
It's not clear how general a result this is, but undoubtedly someone is already trying it on financial data.
[1] https://www.youtube.com/watch?v=7A_rl-DAmUE https://www.youtube.com/watch?v=7A_rl-DAmUE
- mellavora 4y agoyes, back in 1990-2000. See https://www.econstor.eu/bitstream/10419/40278/1/338823255.pdf https://www.econstor.eu/bitstream/10419/40278/1/338823255.pd... though I was thinking of another paper which I couldn't find with a quick google. See also Mandelbrot, fractals and scaling in finance, 1997 https://link.springer.com/book/10.1007/978-1-4757-2763-0 https://link.springer.com/book/10.1007/978-1-4757-2763-0