macrocephalopod

I would say returns/vol ~ t(mu, sigma, v) but otherwise more or less agree!

I'm not sure what you're getting at. The pages you posted are describing volatility drag (in particular how you can overcome it with diversification and proper bet sizing). What's the relevance?

that's not what those are describing. Shannon was specifically trying to find the most volatile stock so that he could rebalance the portfolio more frequency. and in that example the returns are stationary the whole portfolio is capturing vol, there's nothing to drag

Of course there is. The stock has expected arithmetic return of 75% and expected geometric return of 0. Vol drag is the difference. By using less leverage and rebalancing you get closer to the arithmetic return (because you halve the vol drag)

It looks like magic because the example uses a very contrived distribution of returns, but it’s the exact same principle behind why buying a stock index has positive returns even though buying a single stock will likely make you go broke.