The funniest part of this is that you have people with no fundamental understanding of companies attempting to perform fundamental analysis...the solution which

The funniest part of this is that you have people with no fundamental understanding of companies attempting to perform fundamental analysis...the solution which doesn't seem to have occurred here is...perform fundamental analysis. All of the stuff that is being signalled here can also be worked out by just reading the financial reports (the issue with a lot of these models is that information leaks through...ofc, companies that announced downsizing are more likely to go bust...if you need NLP to work this out, finance is not for you...the example of semis is the same, everyone knew that ppl were uncertain about supply, you didn't need NLP for that). The only reason these results are in any way remarkable is because academics convinced everyone that there is no "signal" in conference calls...right, everyone who has ever invested money knows this is false, this isn't surprising. And, just like value factors, this will go badly wrong because computers can't actually value businesses by themselves, they can't do fundamental analysis. Attempting to shortcut this is not smart.