The math
The portfolio you chose is not the one you own.
You picked an allocation once, probably years ago, for reasons that made sense. Markets have been rearranging it ever since, quietly and in one direction. Rebalancing is usually sold as a way to buy low and sell high. It mostly is not. It is the only way to keep owning the thing you actually chose.
The drift nobody notices
Start with 60% in stocks and 40% in bonds, and never touch it. Stocks are expected to return more than bonds, so in most years they grow faster, and the stock share of the portfolio creeps upward. Nothing announces this. No statement flags it. The allocation you chose simply stops being the allocation you have.
Worked example
$100,000 invested 60%/40% and left alone for 30 years. Stocks 7% real with 16% volatility, bonds 2% with 6%. 40,000 simulated paths
- Stock weight you chose
- 60%
- Stock weight at the median, after
- 82.6%
- In the most drifted tenth
- 94%
A moderate portfolio became an aggressive one, and in a tenth of cases an almost all-equity one, without its owner making a single decision. That is the problem rebalancing exists to solve.
The uncomfortable part: leaving it alone usually makes more money
Most articles on rebalancing skip this, because it undercuts the usual pitch. If stocks carry a higher expected return than bonds, then a portfolio drifting toward stocks is drifting toward the higher-returning asset. Left alone, it usually ends up with more money.
| Rebalance | Median | Bad case (10th) | Worst drawdown | Ending stock weight |
|---|---|---|---|---|
| Never | $401,494 | $179,489 | -24.4% | 82.6% |
| Every year | $380,735 | $192,479 | -20.0% | 60.0% |
| Only when 5 points off | $381,765 | $192,134 | -20.1% | 60.0% |
The portfolio never rebalanced finishes with a median of $401,494. Rebalanced every year it finishes at $380,735. Leaving it alone won by $20,759, or 5.2%.
So the claim that rebalancing earns a bonus by buying low and selling high is, in the ordinary case, backwards. It sells the asset that is winning to buy the one that is not. Over long periods that is a cost, not a gain.
What it actually buys
Look at the other columns. Rebalancing improves the bad case by $12,990, and it cuts the typical worst drawdown from -24.4% to -20.0%, which is 4.4 points of protection in the year it matters most.
Rebalancing does not make you money. It keeps you owning the risk you signed up for, instead of the risk the market handed you.
That is a real benefit and it is the correct way to think about it. The drifted portfolio's higher median comes packaged with a risk its owner never agreed to. An investor who chose 60% stocks because a larger drawdown would make them sell in a panic is not better off holding 83% stocks going into a crash. The higher median is irrelevant if the drawdown forces the sale that locks in the loss. That is the behavior problem, and it is the expensive one.
The one case where the bonus is real
There is a world where rebalancing genuinely adds return, and it is worth knowing so you can tell whether you are in it. If two assets have roughly equal expected returns and move independently, then repeatedly selling the one that rose and buying the one that fell does harvest a small gain from their volatility.
Give stocks and bonds the same expected return in the model and the result flips: rebalancing finishes at $329,520 against $314,014 for leaving it alone. That is the rebalancing bonus people describe, and it is real. It just requires an assumption, equal expected returns, that does not hold between stocks and bonds. It holds better between similar assets, such as two broad equity funds.
How often, and how
Use bands, not the calendar. Rebalancing only when an asset drifts 5% points from target produced almost exactly the same result as rebalancing every year, but with a median of 9 trades in 30 years instead of 30. Fewer trades means fewer taxable sales and less to think about, for no measurable loss.
Rebalance with new money first. Direct contributions toward whatever is underweight rather than selling what is overweight. It moves the allocation back toward target without realizing a single gain, and for anyone still saving it handles most of the drift on its own.
Do the selling inside tax-advantaged accounts. Swapping between funds in a 401(k) or IRA triggers no tax. The same trade in a taxable account can hand you a capital gains bill, which is a cost the simulation above does not charge and a real one. Where each asset sits is its own decision, covered in asset location.
Revisit the target, not just the weights. The allocation right for a thirty-year-old is not right for a sixty-year-old. Rebalancing restores the target you set. It does not tell you whether that target still fits, and that is a separate, less frequent question.
What the model leaves out
Returns are drawn independently each year, which ignores any tendency for markets to mean-revert. If they do, rebalancing looks somewhat better than shown, because it buys after falls. The model charges no taxes or trading costs, which flatters frequent rebalancing. And it holds the target fixed for thirty years, which nobody should. The direction of the findings is robust to all of that; the precise dollar figures are not.
The fund industry sells rebalancing as a return enhancer, because that is easier to market than risk control. It is the other thing. You are paying a small, expected cost to stay the investor you decided to be, rather than drifting into one you would not have chosen.