Just saw the title of this paper https://www.sciencedirect.com/scien...sleading and proposal for an efficient metric I totally agree with only the first half of the title even without reading the details inside the paper. There are a lot of problems for people to try using AI/ML/ANN (all sorts of NNs) but ended up with just failures in real trading. Why? It is not actually a mystery. The traps start from the very beginning to do that. From the selection of quality data, many data sources are skewed including almost all free data from such yahoo financial, or even paid sources, biased selection of those data, all sorts of troubles that ending up peeking into the future without notice, etc. Anyway, I still believe it is possible to use ML/AI properly in trading but only a few might success. Just want to share my thoughts here.
like for every "machine" there's a learning period. maybe they're not that accurate now, but some day they'll steal our future. like it was at the crack of down with AI