Let’s say, you are a Data Scientist trying to model the Stock market behavior these days, using Reinforcement Learning, for example. You are in trouble.
First Corona. Try to figure out how the stock prices for newly founded companies claimed will produce the vaccine against Covid-19. Can you figure it out, based on some similar situation in the past? Not really: there is no such similar situation recorded in history. So, Data Science guys, watch closely what is going on with the stock market. You will most probably not have a similar situation in your lifetime.
After 2M deaths because of Corona, the world expects the vaccine put into massive production after only several months of development, instead of 10 years, as how much it would take in normal circumstances.
Now, yesterday, another “interesting to watch” situation on the stock market. The ordinary traders finally figured out that they can’t beat the system where social media is de-facto today’s media for distributing news and where nobody (at least it seems so) controls it. So, they finally decided to use it. Understand the game’s rules, then play them for your good.
This time, Reddit was the social media used for somebody’s purposes, and guess what: it was a legitimate move. They simply joined the game introduced by somebody else before them. The results? Watch out:
Now, let’s say you use Reinforcement Learning to train a model that would operate on the stock market? Think twice about the set of possible environment variables in the model.