In the first case, I agree. One can cherry pick past data and find some basket of stocks that outperforms the market. But in practice the funds that do this ten

In the first case, I agree. One can cherry pick past data and find some basket of stocks that outperforms the market. But in practice the funds that do this tend to underperform, so it seems like something that is unlikely to be successful in practice. The second point is more interesting. The author is arguing the current simple heuristic for "value indexes" is over-inclusive. He cite two specific classes of stock for this: cyclical stocks (like gun manufacturers) and those in long-term decline (like malls). If the author is right, there probably is a way to revise these indexing heuristics (still not specific stock picking). I think the smart-beta wave a few year back tried some things like this.