There are lots of blog posts out there claiming to teach you how to use deep learning (typically RNNs) to predict stock prices, FX rates, BTC price, from past price data. Is it actually possible? Well, mostly not. A thread.
Show this thread
-
That doesn't mean that price data doesn't contain information. Only that this information doesn't convert to successful strategies for an average player. It's possible to show models that look like they would be successful, in theory. But you can't deploy them.
Show this thread
Remember, you need an advantage no one else has (usually a new dataset). It's purely a game of information arbitrage. You can't recycle the same information and models everyone has access to. Unless you know you're special, don't even try. And if you have to ask, you're not.
-
-
The last line is
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Thank you for the detailed take. Also, there is the whole set of predicting returns and not prices, one being stationary and other not. AI in trading has been a market wrapper for a while.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Simply put, Markets are Efficient. Also, almost all algorithms were not developed for use in "adversarial" environments. Finally, noise to signal is far to high compared to applications for which these algos were developed in the first place.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
This Tweet is unavailable.
-
This Tweet is unavailable.
- Show replies
-
-
-
Last line is
Crypto markets are so inefficient that you could make money with logistic regression. Not so long ago and in period of volatility you can even do triangular arbitrage strat. on the same exchange.. the peak of market inefficiencies.Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
People routinely underestimate the sophistication of the markets
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
I'm always suprised by how many people think they are very special and can beat professional players with very deep pockets in markets. And often they are not very open to reason, so I guess they have to make the "experience" for themselves to learn.
Thanks. Twitter will use this to make your timeline better. UndoUndo
-
-
-
Price prediction is not possible, after 10.000 iterations of various DL models on the problem I can confirm this. Though it is possible to classify trends with around 80% acc. But this has some kind of inherent uncertainty (similar to Heisenbergs) where the higher the acc. on one
-
variable, the lower the acc. on anotherone (e.g. price vs. time). Furthermore noone will go public if they found a working model, bc. that will destroy its advantage. At least that work made me pretty proficient with Keras, TF, CUDA and Python...
End of conversation
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.