The very essence of the EMH (which is frequently oversold, but holds to some degree) is that we can't know which stocks are going to go down or up just by looking at the current state of the market. The expected value of any given stock is the same as that of an index fund, assuming it's no transaction cost to obtain it. So ignoring fees, picking stocks randomly does have the same positive expected return as holding all a small bit of all stocks at once (which is what owning an index is), just a much higher variance and thus people choose to own indexes.

If you want to be sure of positive expected value, just pick randomly, with probability proportional to market cap. That gives the same expected value as an index fund, with much higher variance.

I see the error in my thinking here. My suggestion of rebalancing loses the high end component of variance while avoiding the 0 value case.

I guess im just too cynical to believe a company can operate indefinitely with indefinite growth!

if you rebalance / repick, yes. If it's just buy and hold, I don't agree. The benefit of index funds is the failing companies are dropped before they hit $0.

That seems like a strange perspective to me. All of the theoretical justification for passive investing (you get the average return without the fees, maximum diversification, etc.) implies that you'll get the best result from holding every possible security in some proportion. You're saying that index funds work not because of any of that, but because they've observed a market inefficiency (low market-cap or recently-fallen stocks underperform the market), and make an active decision to deviate from the market portfolio in order to exploit it. Why believe that there's exactly one easily-exploitable market inefficiency, but no others?