O’Doul’s Part 2

I posted the other day about fundamental indexation being the O’Doul’s of value investing, because it is just the first step away from a market cap weighted index and towards a more differentiated value portfolio.

I received lots of emails about the post, so figured I would spend more than 20 minutes on the time series (probably a good rule in general!). First thing to note: I had removed return outliers (stocks with bogus or unreasonably high returns) but did not remove valuation outliers (stocks that rate as extremely cheap with earnings yields or cash flow yields way above the rest of the market).  When I remove the valuation outliers (which carried very high weights because of their high earnings yields), the results change. Returns were about 1.5% per year lower (because some of the large weights did very well) and standard deviations 2% per year lower (i.e. less volatile, because the large weights were very volatile).  These results offer what I think are a more fair representation of the “value weighted” strategies I presented.

This time I am combining measures like a fundamental index would, instead of presenting several single factor results. I use cash flow, sales, and earnings to determine weights in both a “fundamental” style index and a “valuation” style index. Now, all valuation outliers are stripped out (stocks with a P/E or P/CF under 2 or a P/Sales under 0.1 were removed). The results are as follows (someone asked for more recent results as well, which are included below):

results oduls 2

Like before, value works better than fundamental which works better than market cap. Here are the rolling-10 year results historically. Both outperform market cap weighting roughly 93% of the time in 10-year periods, with the periods ending 1998 and 1999 being the only exceptions (no surprise as that was value investing’s worst period).

rolling 10 years for weighting schemes

Once again, the advantage of favoring cheaper stocks is clear. The issue I have with the valuation index, having spent a little more time with it, is that it is extremely diversified–too much so. Its top 10% holdings today represent just 6% of the portfolio and the top 100 40% of the portfolio. Compare that to fundamental index where the top 10 represent 22% and the top 100 69%. Value works great when you focus on the cheapest companies, but given that this method requires owning some weight in all large U.S. stocks, it restricts the size of the bet we can place on the cheapest stocks.

If our hypothetical index removed the most expensive half, or three quarters, or worst 90% of stocks entirely, and then weighted across the remaining cheaper stocks, the results would be quite a bit better. For example, the cheapest 10% of large U.S. stocks across this same period delivered a return of around 14.5%–more than 2% per year higher than the valuation-weighted index.

I’ve written a lot about value investing, and think that more concentrated applications of the discipline work best. But even in diluted form, value has provided a strong edge against cap-weighted benchmarks.