Let’s say that based on all the information that this AI can gather, be news, stocks data, market data or manifolds bet percentages. It gathers all that informations to figures out that within the next week or two the stock price will rise up strongly, lightly, or fall. This is not sentiment but predictions.
This process could be duplicated for the NASDAQ. But can an AI be this powerful that we haven’t seen before.
1,000
3.00Also, predicting "trends" is rather easy, your best bet is to aggregate the market predictions (alternatively, even very simple extrapolations can be quite reliable when everaging over enough data, e.g. the S&P 500 goes brrr over large enough time frames). The real question is whether a single agent can reliability "beat the market" (enough to offset its cost).
Assuming you mean something like picking the best stocks:
Historical data from its test set / backtesting? Maybe. Future data? I don't think so unless it performs a kind of 51% percent attack on information and intelligence, since it would also have to model the other AIs. The predictions of other AI agents would already be priced in.