Stocks and bonds being inversely correlated is an old rule that hasn't held as true since the 2008 crisis. Stocks, most bonds, and most other asset classes have been moving up and down together although with different degrees of volatility. Long term US treasuries are a standout as one of the only investments that has a clear inverse correlation to the general pattern.
Having run a lot of simulations in determining my own portfolio I came to the conclusion that these things affected my returns, in decreasing order:
1. annual maintenance costs
2. tax effects
3. choice of index
4. diversifying beyond stocks/bonds
I imagine most people who aren't using a passive strategy can chop 1 - 1.5% off their costs just by switching to indexes, and the remaining optimizations will pale in comparison. But as the size of the portfolio increases the additional effort becomes worthwhile.
It's possible that over the long term the market will decouple again but keep in mind that most investment advice about "universal truths" about the "long term" are based on a little over 100 years of market data. Given that I started investing at age 20 and will hopefully live to old age, that's not a lot of training data in comparison to the amount of prediction it's generating.
Having run a lot of simulations in determining my own portfolio I came to the conclusion that these things affected my returns, in decreasing order:
1. annual maintenance costs 2. tax effects 3. choice of index 4. diversifying beyond stocks/bonds
I imagine most people who aren't using a passive strategy can chop 1 - 1.5% off their costs just by switching to indexes, and the remaining optimizations will pale in comparison. But as the size of the portfolio increases the additional effort becomes worthwhile.
It's possible that over the long term the market will decouple again but keep in mind that most investment advice about "universal truths" about the "long term" are based on a little over 100 years of market data. Given that I started investing at age 20 and will hopefully live to old age, that's not a lot of training data in comparison to the amount of prediction it's generating.