Equity Supply/Demand Indicator


I read a very interesting post from AlephBlog  which led me to another blog called Philosophical Economics. It’s a long and in depth article I had to read a few times to understand but the basic gist of it is that when investors are under allocated to equities, future returns are better than when they are over allocated. It utilizes the Fed Flow of Funds report to develop a ratio of the value of equities held by market participants over the TOTAL value of capital where capital equals equities+cash+bonds.

It’s a novel approach that basically examines positioning extremes by market participants and then delves into pitfalls of more common valuation techniques. The post was written in 2013 so I wrote some code in Python to recreate the model with fairly close results:


There is a very high negative correlation (-0.89) between the model and SPX returns 10 years forward. In other words, the lower the average allocation to equities the better stocks perform in the next 10 years.


Using a linear regression model, the current value of ~42% average allocation predicts 21.2% 10 year Total Return or about 1.95% annualized which is not so great. As you can see, 42% is also quite high historically approaching the market top in the 1970’s. Add this to the exceedingly long things to worry about.

There are a few limitations to the model that I see;

  1. With a 10 year return horizon and data only going back to 1940 we have a relatively small sample.
  2. It would also be worthwhile to check 9,8 or even 5 year forward returns and see if the model is still viable. If any models’ predictability falls apart with small changes then it raises serious doubts as to the validity of said model.
  3. I wasn’t able to exactly match the blogs’ model. For example, with an allocation level of 40% he forecasted a return of approximately 6% annualized, a very large discrepancy indeed for a minor change. It may be due to additional data (his going from 2013). He also utilized average returns on the linear regression model and I utilized TR and annualized later. I’ve contacted the author and will update accordingly.

That said, it does have an intuitive appeal which perhaps can explain future returns better than valuation based approach relying on earnings forecast or investor sentiment to determine multiples (P/E multiples for example)

The code is posted here.



Be the first to comment on "Equity Supply/Demand Indicator"

Leave a comment

Your email address will not be published.