Someone trained a DQN agent to trade Bitcoin and all the AI did was buy, buy, buy! Reminded me of my JGB experience. https://medium.com/@danielzakrisson/the-self-learning-quant-d3329fcc9915#.95nas48qy …pic.twitter.com/fHZsCvDCHk
RL doesn't make any sense for trading unless you are big enough to impact the market and you have the ability to simulate that.
-
-
otherwise supervised learning techniques will work much better and are far easier to implement
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
A potential advantage of RL over supervised is that it could learn short or mid-term policies (such as day trading).
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
The biggest thing missing from most RL trading simulation is that you're trading doesn't affect the way others trade.
-
you need to use generative models to create the market that the reinforcement learning algorithms learns from
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.