Due to my tax situation I'm not a fan of dividends, which are taxed MUCH more heavily than capital gains. ($1 in dividends has the same value as $0.70 in cap gains for me).
But many readers of the board are "US Persons" for tax purposes, and like dividends. Some might be retired. So sometimes I look at this issue.

Anyway, a KISS screen for dividends.
Value Line 1700 universe, since I'm a dinosaur and I have the data. And their criteria for inclusion seem to suit conservative investment styles.

* Find all the stocks with current dividend yields between 2% and 8%.
Less than 2% and it's not interesting for a dividend portfolio. More than 8% and there is probably something wrong.

* Ensure the domicile is in the US. I know Americans hate the hassle of dividends coming from outside their country, and the tax rates can be confusing.

* Buy the 30 stocks with the highest reported "Return on Shareholders Equity". (that's the annual figure, updated rarely, not the recent quarter field)

That's it!

Strongly recommended to run this quarterly. The turnover is smaller than you might guess, since the ROE field is updated only annually for each firm, and high ROE firms often stay high ROE.

Backtest figures with 0.4% round trim trading cost estimate:

The returns are not that exciting, but they are (in backtest) a hair better than the S&P 500 and of course you get a higher dividend yield.
As of last month, the picks had an average yield of 3.20%. The S&P is yielding around 1.37% these days, so it's a meaningful difference. In fact, 3.2% isn't a terrible idea for a SWR these days.

The backtest outperforming the S&P in the last 20-or-so years is actually pretty impressive, since dividend stocks in general lagged so horribly for a while. From around May 2014 to the pandemic lows around March 2020, the backtest of this returned only 0.5%/year, less than its yield, while the S&P managed 8.2%/year. But a portfolio of all VL dividend payers returned 2.3% a year, lagging by a remarkable 5.9%/year in that stretch, so any dividend screen was picking from a barrel full of very unlucky apples for a while.

The screen (in backtest of course) did beat the S&P by 8.3%/year 2003 till April 2014, and beat the S&P by 8%/year since the pandemic lows. Since 2003, overall it's an advantage of 3.6%/year. Worst rolling year quite a bit nicer than the S&P too.

Note, I quote the figures from 2003 for conservatism...because almost ANY dividend/value screen looked great in the 2000-2002 tech bear. This one beat the S&P by 24.7%/year during those three years.


Why would one bother to mention a screen that doesn't really outperform the broad market by much? Well, of course you get over twice the yield, its stated raison d'être.

But I also think it will get you a portfolio of very high quality firms, since essentially all great businesses have high ROE. (not all high ROE firms are great businesses, but it's a high enough correlation to tilt a portfolio in the right direction). And it's equally weighted, consequently it's very much lower risk than a typical dividend ETF which tends to reach for yield by buying poor businesses and tends to take on too much risk by being cap weighted.

And hey, it might just beat "the market".
Overall CAGR 1997-2023 inclusive beat the S&P by 3.35%/year in backtest, 30 stocks quarterly after friction.
As a backtest that's humdrum, but if (a big if) that could be achieved over a long period of time of real life use, it's a very good result.

Jim