Prof. Gilbert and Prof. Liabson - How do you think that the ability to leverage herd-mentality will change with the advent of ML? For example, say we have a learning algorithm that predicts that a stock will increase in value, thus a lot of people buy it…which actually winds up driving the price lower instead. Everyone would freak out that the learning algorithm was wrong, but the algo was right - what actually changed was the environment it was being applied to.
To extrapolate this into the long term: let this example occur at some moment in time, denoted as Y. Using my own intuition, I’d think that over time Y+1 would be the result of inverting the prediction from Y, and Y+2 would be the result of inverting the prediction from Y+1, and so forth. Can we even continue to reliably use past results? Especially if herd mentality makes algorithmic predictions immediately fluctuate into the inverse?
Gilbert Video Link: https://www.labxchange.org/library/pathway/lx-pathway:53ffe9d1-bc3b-4730-abb3-d95f5ab5f954/items/lx-pb:53ffe9d1-bc3b-4730-abb3-d95f5ab5f954:lx_simulation:5e3f229f?source=%2Flibrary%2Fclusters%2Flx-cluster%3AModernPrediction&fullscreen=true