I think this is not rigorous enough to draw any real conclusions.
If he had done a proper job of reproducing he would have created a write-up of his work explaining his reproduction methodology. The next step would be to get his work peer reviewed.
I think only then you have come close to the amount of analysis and rigour necessary to discredit so many authors of (possibly peer reviewed scientific articles) academic research.
The fact that he mentions that he doesn't know what a meta analysis is in the comments suggests that possibly _his_ results might not be what he purports them to be.
Exactly, I also find it extremely dubious a "Tier 1 professional quant trader" implemented 130+ papers in 7 months.
This means obtaining the same data, reproducing feature engineering and hyperparameters. Implementing learning algos.
Maybe the guy is genius and god-like in NLP, finance, data science and machine learning but even then 7 months is too little time.
I was amazed at how few people call out this obvious lie here.
If he had done a proper job of reproducing he would have created a write-up of his work explaining his reproduction methodology. The next step would be to get his work peer reviewed.
I think only then you have come close to the amount of analysis and rigour necessary to discredit so many authors of (possibly peer reviewed scientific articles) academic research.
The fact that he mentions that he doesn't know what a meta analysis is in the comments suggests that possibly _his_ results might not be what he purports them to be.