Key takeaways
- In Vanguard's 1926-2018 study of a 60/40 portfolio, every rebalancing rule tested returned 8.19%-8.39% a year after tax — a spread of 0.20 percentage points, with Sharpe ratios of 0.50 or 0.51 across the board.
- Annual-with-a-5%-band and monthly-with-a-0%-band held the same 60-61% average equity weight. One needed 34 rebalancing events over 92 years; the other needed 1,116.
- Vanguard's 2022 simulation found annual rebalancing optimal for a 60/40, worth 12 basis points a year over quarterly-with-a-5%-band, and 27 basis points over monthly — of which 80-90% came from market exposure, not from saving on trading costs.
- Rebalancing inside a tax-advantaged account rather than a taxable one raised after-tax returns by 44 basis points a year — more than twice the entire spread between rebalancing rules.
- A 2026 NBER working paper estimates predictable institutional rebalancing costs investors about $16 billion a year, roughly $200 per US household.
The honest answer: it barely matters, and the reason is worth 15 minutes of your time
You've heard "rebalance once a year." You've heard "rebalance when you drift 5%." You want to know which is right.
On the evidence, neither is, because the question is smaller than it looks. Vanguard's Getting Back on Track analysis ran a 60% stock / 40% bond portfolio through 92 years of returns — 1 January 1926 to 31 December 2018 — under twelve rebalancing rules: monthly, quarterly and annual checks, each paired with drift bands of 0%, 1%, 5% and 10%. Every rule landed between 8.19% and 8.39% a year on a tax-adjusted basis. Every one produced a Sharpe ratio — return per unit of risk — of 0.50 or 0.51. Every one held an average equity weight between 60% and 63%, right on target.
Ninety-two years, and the whole argument is worth 0.20 percentage points a year, with no winner in it. Vanguard's own conclusion is blunt: they "don't find a specific rebalancing threshold or frequency that consistently outperforms other forms of rebalancing."
So the interesting question isn't annual or 5%. It's what you were buying when you rebalanced at all — and it isn't return.
The gap that matters is between rebalancing and not rebalancing — and it's a risk gap
In that same 92-year run, the portfolio that was never rebalanced returned 8.74% a year — the highest number in the table.
It also held an average 85% in equities, against a 40% bond target, with annualised volatility of 14.0% versus roughly 11.4-11.8% for the rebalanced portfolios. Its Sharpe ratio was 0.46, the worst in the table. The chart above plots eight of the twelve rules alongside it; the four omitted are listed in the note beneath. What it can't show is that the tallest bar is a different portfolio wearing the same name.
That's the whole mechanism. Left alone, a 60/40 becomes an 85/15, because stocks outgrow bonds. Vanguard's 2022 study makes the same point on a shorter window: a 60/40 at the end of 1989, never rebalanced, was 80% equities by the end of 2021. Rebalancing didn't make you money. It stopped your portfolio from quietly becoming something you never agreed to own. You're not choosing between two ways to earn more, but between two ways to hold a line — and the data says both hold it.
Where the rules really differ is in how much you have to trade
Same table, same 92 years:
- Monthly, 0% band (rebalance on any drift at month-end): 1,116 events. Return 8.20%.
- Annual, 5% band: 34 events. Return 8.19%.
- Annual, 10% band: 14 events. Return 8.20%.
One basis point separates the first two. Thirty-three times the trading separates them too. That is the real trade-off: you buy a marginally tighter grip on your target weights, and you pay for it in transactions, tax events and attention.
A true threshold rule also requires daily monitoring — Vanguard's 2022 paper calls it "not practical for investors who manage their own portfolios." A 5% band you check once a year is not a threshold strategy. It's an annual strategy with a snooze button — the annual-plus-band row in the table, and it did fine.
The strongest counter-argument comes from Vanguard's own fund managers
There's a serious case against everything above, and it's Vanguard's own December 2024 paper, The Rebalancing Edge. Studying target-date funds with 10,000 simulations over 10-year horizons, it found a threshold rule beat calendar rules on nearly every measure. A 200/175 policy — drift 200 basis points from target, then trade back to 175 — returned 7.19% annualised, against 7.01% monthly and 7.08% quarterly. It cost 5.1 basis points in transaction costs, versus 22.1 and 18.3. It held allocation drift to 198 basis points a year, against 241 and 333. In March 2020, a monthly-rebalanced portfolio would have drifted up to 7% from target and a quarterly one up to 10%; the 200/175 rule never drifted more than about 2%.
That looks like a clean win for thresholds. Read the fine print and it's a win over monthly and quarterly — the two rules fund companies use, and the two the 2022 study identified as too frequent. Annual was never in that race. It's also a fund manager's problem, not a household's: daily monitoring, institutional dealing costs, no capital gains tax, and a benchmark to track.
When the 2022 study did put annual in the race, using the same simulation engine, it came out optimal for every stock/bond mix from 35% to 90% equities. Its edge at 60/40 was 12 basis points a year over quarterly-with-a-5%-band, and 27 over monthly. Not zero. Not much.
In the most volatile markets, the lazier rule won
The intuitive objection is that bands should shine when markets are wild — that's when drift is biggest. Vanguard tested it by sorting simulations into volatility quintiles. In the highest-volatility quintile, annual rebalancing beat monthly by 1.52% on their utility measure — by far its largest margin anywhere in the study.
The relationship isn't monotone. From the lowest volatility quintile to the highest, annual's margin over monthly runs 0.04%, -0.22%, 0.24%, -0.39%, 1.52% — so in two of the five, annual slightly trailed monthly. The 1.52% is the turmoil quintile alone, and it does all the work.
The reason is unglamorous. Transaction costs rise with volatility, and Vanguard found they "can increase tenfold during periods of market turmoil." Trading more often in chaos means trading at the worst prices, sometimes reversing days later when the market swings back. The regime that seems to argue for more rebalancing argues, on this evidence, for less.
And only 10-20% of annual rebalancing's advantage over monthly came from lower transaction costs. The other 80-90% came from market exposure — from letting equities run a little longer before selling them.
Tax location moved the needle four times as much as frequency
In the same Vanguard analysis, doing the rebalancing trades inside tax-advantaged accounts rather than taxable ones raised after-tax returns by 44 basis points a year, "without increasing risk exposure." Set that against the 20-basis-point spread separating the best rebalancing rule from the worst.
A reader agonising over annual-versus-5% is optimising the smaller variable. In a taxable account, a 5% band that triggers in a strong year forces a realised capital gain that an annual review might have absorbed with new contributions instead. Vanguard's 2022 paper agrees: once taxes are modelled, they expect "a similar or even a less-frequent rebalancing strategy to be optimal." In a sheltered account — ISA, 401(k), SIPP — that constraint vanishes and the question shrinks back to the 20 basis points it was always worth.
"A 5% band" means two different things, and on a small holding the difference is enormous
Here the received wisdom quietly breaks. A 5% band can mean 5 percentage points of the portfolio (absolute), or 5% of the position (relative). On a 60% equity target, both are unremarkable: absolute triggers at 55/65, relative at 57/63.
On a 5% crypto sleeve, they are different universes. An absolute 5-point band triggers only at 10% or 0% — the holding has to double before you touch it. A relative 5% band triggers at 4.75% or 5.25%, which for an asset that moves like crypto could be a Tuesday. Same words, wildly different rule. The same trap sits under any small satellite position — a gold allocation, a single-country fund — which is why "just use 5% bands" is not portable.
Everyone rebalancing on the same schedule has a price
One recent finding cuts against calendar rules specifically. An NBER working paper by Campbell Harvey, Michele Mazzoleni and Alessandro Melone — circulated March 2025, revised January 2026 — studied US futures data from 1997 to 2023 and found predictable institutional rebalancing moves prices. When their calendar signal spikes, equity returns fall about 17 basis points the next trading day; their threshold signal is worth about 16. The pressure reverts almost entirely within two weeks, which tells you it's flow, not information.
They estimate this costs institutional investors more than 8 basis points a year — around $16 billion annually, or roughly $200 per US household — and that traders front-run it profitably, with a Sharpe ratio above 1. Their fix is not a different frequency but a randomised one: "simple randomization in rebalancing schedules — all else being equal — largely eliminates these costs." The predictability clusters at month-end and quarter-end, where fund calendars sit. For an individual trading once a year, this is a rounding error — but it is a reminder that the strongest empirical objection to calendar rebalancing has nothing to do with drift.
What this evidence cannot tell you
Backtests are not forecasts. The 8.19%-to-8.39% range is one path through history, and it contains a spectacular equity run; a different century could produce a different ranking. Vanguard's 2022 and 2024 papers aren't history at all — they're 10,000 simulations from Vanguard's own capital markets model, calibrated on the past and, in Vanguard's words, liable to "underestimating extreme negative scenarios."
The data is also overwhelmingly US-centric: the 1926-2018 series splices US indices for its early decades before going global, and the NBER study is US futures. The Harvey paper is explicitly not peer-reviewed. And Vanguard is not a disinterested referee — it manages the funds and sells the advice, and "don't overthink this" is a comfortable finding for a firm built on low-cost simplicity. The result holds across three Vanguard methodologies and one hostile academic one, but a reader who discounts it is not being unreasonable. None of it touches illiquid assets, concentrated single stocks or property, where a band can be breached and simply not actionable.
What would change the conclusion
If trading were free, instant and untaxed, the cost side of the trade-off collapses and tighter, more frequent rebalancing dominates. Zero-commission brokers have moved partway there — but tax, spreads and your own attention have not gone to zero.
If the equity risk premium disappeared — if stocks stopped systematically outgrowing bonds — then 80-90% of annual rebalancing's edge, the part Vanguard attributes to market exposure rather than cost savings, would go with it. Bands and calendars would converge on being the same rule.
If your portfolio isn't two liquid asset classes, the literature strains. Every study above is a stock/bond portfolio. Add a crypto sleeve or anything that gaps 30% in a week, and "annual versus 5% bands" stops being a small question, because the drift arithmetic that makes the two rules converge no longer holds.
The thing to watch isn't the calendar. It's drift itself — how far your weights have wandered from the ones you chose, and whether that's 2 points or 25. LedgerTouch shows that number continuously; a spreadsheet shows it once a year. Either is enough, which is rather the point.
This content is for informational purposes only and does not constitute financial advice. Always do your own research or consult a qualified financial adviser before making investment decisions.