Yes, all of those things aren't representative. I have no idea why people use the last fifty years, that is roughly your investing life-time, and you are using

Yes, all of those things aren't representative. I have no idea why people use the last fifty years, that is roughly your investing life-time, and you are using that as your only sample. That isn't investing, that is rolling the dice and hoping to get lucky. Dimson, Marsh, Staunton is the only representative long-term study of global returns. One, that sample still has selection bias (a couple of years ago, they found an issue with calculations for a market that went to zero, that knocked 10% off their estimates). Two, very bad events happen infrequently so you have a sample of a few thousand years across many countries: German hyperinflation is one year, Austro-Hungary is two years, China is a few years, etc. So there isn't a huge sample of very bad things given that capital markets have existed inconsistently for two centuries only. Three, there is a discontinuity between how finance professors look at the past, how investors today percieve the past, and how the past actually was. As an example, after WW2 equities did very well...one problem, the US had to issue a tons of govt debt, and forced everyone to own bonds. The sample doesn't really show that most people couldn't own equities. Equally, transaction costs are often not well understood (i.e. the assumption is that you can produce a market-cap weighted portfolio, and these effects are correlated to returns: when you can't produce a cap-weighted portfolio, returns are high because participation is low). But yes, their number is ~300bps real...I would emphasize though that you should actually read about market history, and understand the risks.