You picked a mix — say 60% stocks and 40% bonds — and the market has been rearranging it for you ever since. A good year for stocks turns 60/40 into 65/35; a few good years in a row turn it into something you never chose. Rebalancing is the act of putting it back, and the question everyone asks is how often: every month, every quarter, once a year, or only when the mix has drifted far enough to matter?
We tested the common rules on a 60/40 portfolio of U.S. stocks and 10-year Treasuries from 1926 to mid-2023, across 930 overlapping 20-year periods. The short answer: how often you rebalance barely changes your return — annual, quarterly and a 5-point drift band all landed within 0.12 percentage points a year of each other. What changes everything is whether you rebalance at all. Left alone, the typical 60/40 portfolio ended its 20 years at 75% stocks, and in the worst case at 96%.
The rulesFour ways to decide when to rebalance
Every rebalancing rule answers one question — when do I trade? — and they fall into two families.
Rebalance on a date
Monthly, quarterly or once a year, you sell whatever is overweight and buy whatever is underweight, however small the drift. Easy to remember and easy to automate; it trades even when nothing needs fixing.
Rebalance when the mix drifts
You set a band — for example 5 percentage points around a 60% stock target, so 55% to 65% — and trade only when the portfolio crosses it. Fewer trades, but someone has to keep checking.
The two can be combined, and that combination is what Vanguard suggests in its Principles for Investing Success: check the mix at least once a year and rebalance if it has moved meaningfully from the target.
stock value ÷ (stock value + bond value)stock weight − target weight → 66% − 60% = +6 pointsstock value − target weight × total → $66,000 − 0.60 × $100,000 = $6,000The test97 years, 930 twenty-year periods
We started a $100,000 portfolio at 60/40 at the beginning of every month from February 1926, ran it for 20 years under each rule, and recorded the annual return, the deepest fall from a peak and the number of trades. Threshold rules were checked once a month; the annual rebalance happened in December.
Monthly
- Median return a year
- 8.60%
- Median worst drop
- −24.3%
- Trades in 20 years
- 240
- Stocks at the end
- 60%
Quarterly
- Median return a year
- 8.72%
- Median worst drop
- −23.8%
- Trades in 20 years
- 80
- Stocks at the end
- 60%
5-point band (55–65%)
- Median return a year
- 8.77%
- Median worst drop
- −23.8%
- Trades in 20 years
- 11
- Stocks at the end
- 61%
10-point band (50–70%)
- Median return a year
- 8.92%
- Median worst drop
- −23.2%
- Trades in 20 years
- 4
- Stocks at the end
- 63%
Never rebalance
- Median return a year
- 9.41%
- Median worst drop
- −24.7%
- Trades in 20 years
- 0
- Stocks at the end
- 75%
Once a year
- Median return a year
- 8.84%
- Median worst drop
- −23.6%
- Trades in 20 years
- 20
- Stocks at the end
- 61%
| Rule | Median return a year | Median worst drop | Trades in 20 years | Stocks at the end |
|---|---|---|---|---|
| Monthly | 8.60% | −24.3% | 240 | 60% |
| Quarterly | 8.72% | −23.8% | 80 | 60% |
| 5-point band (55–65%) | 8.77% | −23.8% | 11 | 61% |
| 10-point band (50–70%) | 8.92% | −23.2% | 4 | 63% |
| Never rebalance | 9.41% | −24.7% | 0 | 75% |
| Once a year | 8.84% | −23.6% | 20 | 61% |
Source: CalculatorAI · calculatorai.app · CalculatorAI calculation on Robert Shiller's stock and bond return data
Three things stand out.
- Frequency is close to irrelevant to return. Once a year, quarterly and a 5-point band finished within 0.12 points a year of each other. Monthly rebalancing was the worst of the disciplined rules, trailing annual by a median 0.2 points a year — it keeps selling stocks after every good month, which costs a little in markets that tend to keep rising for a while.
- Threshold rules do the same job with far fewer trades. A 5-point band needed about 11 trades in 20 years; quarterly needed 80, monthly 240. In a world with no costs that difference is invisible. In a taxable account, where each sale of a winner can be a taxable gain, it is the whole difference.
- "Never rebalance" earns more — because it stops being a 60/40 portfolio. It finished with the highest return because it ended up holding mostly stocks. That is not a rebalancing strategy beating the others; it is a different, riskier portfolio that nobody chose.
The driftWhat "never rebalancing" really means
The returns above hide what happens to the portfolio itself. Take a 60/40 portfolio started at the bottom of the 2009 crash and never touched again:
By mid-2023 the 'balanced' portfolio was 89% stocks.
Show these figures as a table
| Value (% in stocks) | |
|---|---|
| Start, March 2009 | 60 |
| End of 2010 | 70.5 |
| End of 2013 | 76.8 |
| End of 2016 | 79.8 |
| End of 2019 | 83.8 |
| June 2023 | 89.4 |
Source: CalculatorAI · calculatorai.app · CalculatorAI calculation on Robert Shiller's stock and bond return data
Across all 930 twenty-year periods, the never-rebalanced portfolio ended at a median 75% stocks, and in the most extreme period at 96%. Vanguard's own illustration with global data tells the same story: a 60/40 portfolio left alone from 2002 was already 71% stocks by 2006 and 76% by 2022.
Drift does its damage at the worst moment — when stocks finally fall. Here is the same $100,000 started in January 1990 under three rules, through the two bear markets that followed:
Never rebalance
- Stocks, Aug 2000
- 77.5%
- Aug 2000 → Sep 2002
- −24.5%
- Stocks, Oct 2007
- 71.9%
- Oct 2007 → Feb 2009
- −28.3%
5-point band
- Stocks, Aug 2000
- 63.0%
- Aug 2000 → Sep 2002
- −17.1%
- Stocks, Oct 2007
- 58.3%
- Oct 2007 → Feb 2009
- −24.2%
Once a year
- Stocks, Aug 2000
- 58.6%
- Aug 2000 → Sep 2002
- −16.2%
- Stocks, Oct 2007
- 60.3%
- Oct 2007 → Feb 2009
- −22.6%
| Rule | Stocks, Aug 2000 | Aug 2000 → Sep 2002 | Stocks, Oct 2007 | Oct 2007 → Feb 2009 |
|---|---|---|---|---|
| Never rebalance | 77.5% | −24.5% | 71.9% | −28.3% |
| 5-point band | 63.0% | −17.1% | 58.3% | −24.2% |
| Once a year | 58.6% | −16.2% | 60.3% | −22.6% |
Source: CalculatorAI · calculatorai.app · CalculatorAI calculation on Robert Shiller's stock and bond return data
The investor who never rebalanced entered the dot-com crash with more than three-quarters of the money in stocks and lost a quarter of it — roughly what a 75/25 portfolio would lose, because that is what it had become. Rebalancing does not mainly exist to raise returns. It exists so the risk you live through is the risk you signed up for.
The choiceWhich rule to use
The data supports a simple setup for most long-term investors:
- 01Pick a band, not a date. A 5-percentage-point band around each target (55–65% for a 60% stock target) caught the important drifts with about one trade every two years.
- 02Look at least once a year. Put a date in the calendar — a birthday, tax season, the first weekend of January — and check the mix against the band. If it is inside, do nothing.
- 03Rebalance with new money first. Send new contributions and dividends to whatever is underweight before selling anything. In a taxable account this is the cheapest rebalance there is, because buying creates no taxable gain; selling a winner does. Our guide to capital gains tax shows what that sale would cost.
- 04Trade inside tax-advantaged accounts when you must sell. Selling and buying inside a 401(k) or IRA triggers no tax, so that is where a bigger rebalance belongs.
- 05Rebalance asset classes, not every fund. The rule applies to the split you actually chose — stocks versus bonds, U.S. versus international. If two funds hold the same stocks, as our ETF overlap guide shows can easily happen, rebalancing between them only moves money between twins.
Write the target down
One line per asset class, adding to 100%. A target you only carry in your head drifts with your mood.
Set the band
Five percentage points either side is a good default for a two- or three-fund portfolio.
Pick the check date
Once a year at minimum. After a 20% market move, look again even if the date has not come.
Use cash flows first
Contributions, dividends and withdrawals are free rebalancing trades. Direct them before you sell.
Sell inside tax shelters
If a sale is needed, make it in a 401(k) or IRA where possible, not in a taxable brokerage account.
Record what you did
Date, drift and trades. Next year you will see whether the rule is working or whether you overrode it.
The recordSeeing your drift without a spreadsheet
The hardest part of any rule is knowing where you stand — the funds sit in different accounts, and the percentages change every day. In the Portfolio Tracker you can give each holding a target weight; its Analytics tab then shows each holding's current share next to that target, flags anything that has drifted 5 points or more, and tells you roughly how much to buy or sell to get back. If you have not settled on a target yet, the Asset Allocation Calculator suggests a mix from your age, horizon and risk tolerance and compares it with what you hold now.
Rebalancing is the maintenance half of diversification: the mix only protects you if it stays the mix. And if you are about to put a fresh sum into the portfolio, the question of investing it at once or in stages is covered in lump sum vs dollar-cost averaging — that new money is also your cheapest rebalancing tool.
MethodWhere these numbers come from
- Data. Robert Shiller's public dataset (econ.yale.edu/~shiller/data.htm), January 1926 to June 2023, the last month with dividend data in the file. Stocks: S&P Composite, monthly total return = (price + one month of dividends) ÷ last month's price. Bonds: the dataset's monthly total return series for 10-year U.S. Treasuries.
- Rules. Start at 60/40. Calendar rules reset to 60/40 at the end of every month, every third month (March, June, September, December) or every December. Band rules reset when the stock share is 5 (or 10) percentage points or more away from 60% at a month-end. "Trades" counts rebalancing dates, not individual orders.
- Windows. 930 overlapping 20-year periods, one starting each month from February 1926 to mid-2003. We report the median across windows; the bear-market tables follow single portfolios started in January 1990 and January 2007.
- What it leaves out, and which way it leans. No taxes, fees or trading costs, which flatters the frequent rules: with real costs monthly and quarterly rebalancing would fall further behind the band, which is what Vanguard's 2024 study found. Prices are monthly averages and the band is checked only at month-end, so intra-month swings — and some band crossings — are missed. Ten-year Treasuries are longer than a typical total bond market fund, so the bond side swings more with interest rates than many investors' bonds do. U.S. data only; nominal returns, before inflation.
This is history, not a forecast: the gap between stocks and bonds could be smaller or larger over the next twenty years, and that gap is what drives the drift. Nothing here is advice about your own money.
FAQHow often to rebalance
How often should I rebalance my portfolio?
For most long-term investors, once a year is often enough. In our test of 930 twenty-year periods, annual, quarterly and 5-point-band rebalancing produced nearly identical returns for a 60/40 portfolio, while monthly rebalancing trailed slightly. A practical rule is to check once a year and rebalance only if an asset class has moved 5 percentage points or more from its target.
Is threshold rebalancing better than calendar rebalancing?
It needs far fewer trades for the same risk control — about 11 trades in 20 years for a 5-point band, against 80 for quarterly rebalancing. Before costs, returns were almost the same; after trading costs and taxes, fewer trades usually means a better result. The drawback is that you have to check the mix, so most people pair a band with a yearly review.
What is a good rebalancing threshold?
Five percentage points either side of the target is a common default and caught every major drift in our test. A 10-point band traded even less and returned slightly more, but it let the stock share reach 70% before acting, which is more risk than a 60/40 investor chose.
Does rebalancing increase returns?
Usually not, and that is not its job. A portfolio that is never rebalanced tends to drift toward stocks and earn more over long periods — while carrying more risk than intended. Rebalancing keeps the risk at the level you picked; any return boost is a side effect that appears only in some periods.
Should I rebalance in a taxable account?
Prefer not to sell for it. Direct new contributions and dividends to the underweight asset, and do any selling inside a 401(k) or IRA, where trades trigger no tax. Sell in a taxable account only when the drift is large and the other routes cannot close it.
