You are making the classic mistake of confusing domains exhibiting a normal distribution of outcomes (casino games) with domains exhibiting an exponential distribution of outcomes (the market). This is the sort of thinking that traps people into believing "it went up a lot, therefore it has to revert to the mean and go down" or vice versa - there is no basis for such a belief in exponential domains.
That is a very astute point. I say it more as a broad model. And to the point of casino games - I speak to blackjack only which has a finite set of cards in a deck.
I would argue, broadly, that there is a finite value in the stock market we just don't know what it is (and it changes significantly) but I do agree with you that there are some very significant differences and is a potential flaw in the analogy.
Actually there is a basis for it in blackjack and how to card count. That said I'm not sure what casino's are doing these days ever since the card counting was figured out.
They use a decent size shoe of several decks and reshuffle more than just in between rounds. Furthermore, the dealer only deals from a subset of the shoe IIRC.
Certainly not in the sense of the gambler's fallacy, but you can be sure that someone's fortunes from playing casino games will exhibit mean reversion in the sense that the next game is always more likely to bring their cumulative winnings closer to the house edge rather than further from it. Not so with the stock market. The stuff about card counting is basically impossible to do these days but can alter the house edge, and also doesn't apply to the market.
