No. of Recommendations: 13
This method also allows you to manage tax considerations. For the S&P, purchase 500 stocks. If you are withdrawing 4% per year, each 3 months sell 5 stocks with the greatest loss,
or smallest gain since purchase. Carry losses forward to balance gains when there are no longer any losing stocks. Will work for 25 years if you sell 20 stocks per year.
Hmm, might this not work out to a portfolio that always did a lot more "sell low" than "buy low, sell high"?
You'd also gradually have a lot less diversification as time goes on, and you'd end up holding stocks for a very long time, which isn't always good.
I would be tempted to sell whichever positions have had (say) the lowest EPS growth since purchase date, or perhaps in most recent five years.
The average quality of firm would in theory gradually improve, presumably outweighing the tax increase.
For self indexing, I prefer an approach sort of half way between an equal weight index and a big MI screen.
e.g.,
Decide how much typing you are willing to do for the sake of diversification, e.g., a 80 stock portfolio or whatever number you like.
Buy equal amounts of that many stocks, chosen at random from among the 15% of Russell 1000 stocks with the highest ROE.
Since it's only typing, I'd buy all 150 of them.
Hold a few months, a year, maybe two years if you like. Depends on your tax situation, among other things: presumably Americans with taxable accounts would pick a year.
Rebalance and repeat. i.e., sell any stocks that no longer meet the criteria and replace them with ones that do.
You'll always be holding only those things that meet the index eligibility criteria (or did fairly recently), no short term cap gains tax for Americans except for unforeseeable buyouts.
For people who like to overcomplicate, like me, I'd split it into 4 equal sub-portfolios of different stocks staggered 3 months apart and do a "dozens".
That way you aren't replacing all your stocks the same day of the year, which can cause some issues.
FWIW, a proxy of this using as its universe the 1000 biggest market cap from among the VL 1700 universe beat SPY by 3.12%/year before friction in the last 37.4 years.
Beat the S&P in about 64% of rolling years. Worst rolling year about the same as SPY. The best rolling year at 87% was better than SPY's best year of 63%.
It is a fair bit of typing, and reality is not the same as a backtest, but it probably won't be all that bad.
The underlying reasoning:
Not all high ROE firms are good investments by a long shot--there are lots of duds, bad data traps, and overleveraged bombs.
But all very good very long term investments have high ROE*, so you're getting all of the great ones and eliminating some of the bad ones.
The net result is to improve the batting average a little.
Jim
* except Amazon. Every rule has its exception : )
