Now is a good time to be an early accumulator

Discuss all general (i.e. non-personal) investing questions and issues, investing news, and theory.
Post Reply
Topic Author
Posts: 260
Joined: Tue Mar 26, 2019 10:21 pm

Now is a good time to be an early accumulator

Post by Hydromod » Tue Oct 08, 2019 11:16 am

I ran across a predictive model for the future market returns based on the average investor portfolio allocation to equities the-single-greatest-predictor-of-future-stock-market-returns. In essence, the idea is that a substantial fraction of the market response is based on the investor desire to maintain a portfolio percentage as equities, as opposed to bonds and cash, which drives the overall pricing. There is a follow-up post showing why caution is needed in the interpretation valuation-and-returns-adventures-in-curve-fitting.

Additional discussion includes a link to an online chart updating the prediction.

This was discussed here before (The Single Greatest Predictor of Future Stock Market Returns, Here's what the boglehead mind said about it when it first came out). A mix of comments ranging from bah! to interesting!

Nick Maggiuli (why-the-best-predictor-of-future-stock-market-returns-is-useless has a pessimistic viewpoint of how actionable this information would be for an investor.

The original 2013 post points to data from FRED that is used to calculate equity fraction, and uses 10-year S&P 500 total return. I decided to check out the idea a bit more.

I used the online Schiller monthly database for the S&P and dividends and the FRED data, which starts October 1945. I assign the values to the middle of their respective periods (the middle of the month for the S&P, middle of the quarter for FRED data), and linearly interpolate the FRED data to the S&P dates. Then I calculate the future returns for a fixed duration in years (4 years, 6 years, etc.) for each valid month. I calculate the present equity fraction is approximately 0.44.


The figure shows scatter plots of historical returns given starting equity fraction for various durations. The dots are color coded according to whether the equity fraction is decreasing (blue), steady (gray), or increasing (red). The range in CAGR is where the present equity fraction contacts the convex hull surrounding all of the points (the contact of the amber line with the light gray polygon), which represents the extremes that have been observed.

The big square is a regressed value (labeled E[CAGR]) based on the set of data points, with the regression fit indicated with the R-squared value.

The d[CAGR] label indicates the CAGR for the current period assuming that the expected CAGR values are realized. For example, the plot with nyr = 6 has a two-year window (4 years to 6 years); the d[CAGR] value is what is needed to get E[CAGR] = 0.9 percent after six years given that CAGR was 1.5 percent for the first four years. Actually getting the d[CAGR] value would be highly unexpected, given the possible range of returns.

What the numbers suggest is a swoon for the next ten years, with the worst decline around five years from now, and some gradual recovery afterwards. Presumably this is more likely to manifest as bouncing around rather than a smooth blah.

I don’t get exactly the same forward prediction as others; my E[CAGR] for ten years is 1.1 percent, compared with 3.6 percent by several others. The R-squared value is identical, so I may generated a different regression with three times as many points or may have botched the return calculation somehow.

Regardless, these scatterplots suggest that most of the next decade may be very forgettable for Mr. Market. I'm not sure how this would influence those of us considering Hedgefundie's scheme... may be hard to stay the course.

Post Reply