Yes, trader practices in recent markets are probably cleaner than many decades past, and the older stock market data sets measured the most prominent companies of the time. However, manipulation is usually not sustained over the long term and individual stock or market wide corrections occur. These are decade long performance statistics across many companies. Most of the slime probably washed out to reflect the equity value of the large cap economy.

To me one of the most interesting findings is the relative stability of the long-term rate of return and the shift in its composition toward equities and away from debt/fixed income securities. On the topic of hypothetical projections, one can freely choose how far back to look for your base. Since the future is entirely unpredictible, it is all a shot in the dark anyway.

Nevertheless, do you pick only the past two very strong decades? Only the post-WWII decades? From the 1920s, which is the period used most frequently, because of widespread Ibbotson use of the Ibbotson data? Even further back?

When projecting, both the mean and variance are important. The long-term relative stability of the mean is important to consider. The variability of the composition of the asset classes is important. Then, you need to decide how many booms and busts should be in the data and whether the characteristics of human fear and greed have changed over time. There are no easy answers.

Thank you, The Skilled Investor