Key takeaways
- A 5 point band on a 60% equity sleeve fires when equities beat the rest of the portfolio by 23.8%. The same band on a 5% gold sleeve needs 111.1%.
- Measured against each asset's own volatility, that 5 point band is a 1.31 standard deviation move on the equity sleeve and a 6.32 standard deviation move on the gold sleeve.
- Cboe's volatility indices averaged 15.40 for long Treasuries and 38.34 for crude oil over 3,882 shared trading days from March 2011 to August 2026, a spread of 2.5 times on one methodology.
- Holding strictness fixed at 1.31 standard deviations, the equivalent bands are 5.00 points on the 60% equity sleeve, 4.48 on long bonds, 2.33 on a 5% oil sleeve and 1.08 on a 5% gold sleeve.
- Widening a band from 1% to 5% on a 60/40 cut Vanguard's rebalancing events from 426 to 58 across 1926 to 2018, while the annual return moved from 8.20% to 8.22%.
What a 5 point rebalancing band width actually demands
You've been told to rebalance when a holding drifts 5% from its target. You hold bonds, equities, maybe a little gold. Does that one number mean the same thing in each of those places?
It doesn't, and the gap is far wider than most people expect. On a 60% equity sleeve, a 5 percentage point band is breached when equities outrun the rest of the portfolio by 23.8%. On a 5% sleeve, the same band needs the holding to gain 111.1% against everything else. That's not a subtle difference in strictness. It's the difference between a rule that fires in an ordinary year and a rule that almost never fires at all.
The arithmetic is not complicated. Hold weight w in one sleeve, let it return r while the rest of the portfolio stays flat, and the new weight is w(1+r) divided by (1+wr). Setting that equal to w plus the band b and solving gives r = b divided by w(1 minus w minus b). The band width you chose is fixed. Everything else in that expression is the sleeve's weight.
| Sleeve weight | Rise that breaches a 5 point band | Fall that breaches it |
|---|---|---|
| 60% | 23.8% | 18.5% |
| 40% | 22.7% | 19.2% |
| 20% | 33.3% | 29.4% |
| 10% | 58.8% | 52.6% |
| 5% | 111.1% | 100.0% |
Read the bottom row again. A 5% sleeve inside a 5 point band cannot breach on the downside at all until it's worth nothing. A holding can fall 90% and still sit inside its tolerance band, because 5% of a portfolio minus 5 percentage points is zero. That is not a corner case anyone designed. It's what the words produce when you apply them literally to a small position.
Volatility across the assets people hold differs by two and a half times
The weight half of the problem is arithmetic. The volatility half needs data, and there's one place to get it on a single consistent methodology: Cboe's family of volatility indices. Each one applies the VIX calculation to options on a different market. Cboe's methodology document defines the original as measuring "the market's expectation of 30-day forward looking volatility of the U.S. equity market, as conveyed by S&P 500 Index option prices". The gold version, in Cboe's contract specification filed with the SEC, is "an up-to-the-minute market estimate of expected 30-day volatility of gold prices", built from options on SPDR Gold Shares.
Averaging each index over the 3,882 trading days they all share, from 16 March 2011 to 27 August 2026, gives a ladder.
| Cboe index | Market | Mean level |
|---|---|---|
| VXTLT | 20+ year Treasuries | 15.40 |
| GVZ | Gold | 17.59 |
| VIX | S&P 500 | 18.16 |
| VXN | Nasdaq-100 | 21.35 |
| VXEEM | Emerging markets | 23.01 |
| OVX | Crude oil | 38.34 |
Top to bottom, that's a factor of 2.5. And one row is a surprise: gold at 17.59 sat marginally below the S&P 500 at 18.16 over the same fifteen years. The asset people describe as the volatile one was, on this measure, slightly the calmer of the two. Whatever justifies a special rebalancing rule for a gold sleeve, it isn't that gold moved more than equities over these fifteen years.
The long record agrees on the shape. The 2026 UBS Global Investment Returns Yearbook, built on the Dimson, Marsh and Staunton database, reports the average standard deviation of real returns across 20 foreign countries since 1900 at 23.4% a year for equities and 14.9% for bonds. From 1972 to 2025 the same measure reads 25.6% and 12.3%. Equity volatility has run somewhere between 1.6 and 2.1 times bond volatility for a century and a quarter. Nothing in that range is a factor of ten.
Put weight and volatility together and weight wins
Now combine the two halves. Divide the move a 5 point band demands by the asset's annual volatility, and you get the band's strictness in standard deviations. The lower that number, the more often the band fires.
| Sleeve | Weight | Volatility | Move needed | In standard deviations |
|---|---|---|---|---|
| Global equities | 60% | 18.16 | 23.8% | 1.31 |
| Long Treasuries | 40% | 15.40 | 22.7% | 1.48 |
| Emerging markets | 20% | 23.01 | 33.3% | 1.45 |
| Crude oil | 5% | 38.34 | 111.1% | 2.90 |
| Gold | 5% | 17.59 | 111.1% | 6.32 |
The received wisdom says volatile assets need wider bands. The table says something else. Across the big sleeves the strictness barely moves at all: 1.31, 1.48 and 1.45 standard deviations for equities, long bonds and emerging markets, despite volatilities running from 15.40 to 23.01. Weight and volatility partly cancel. The bigger a sleeve is, the further a given move pushes its weight, which offsets the fact that the calm assets tend to be the big ones.
Where the table breaks apart is at the small end. The oil sleeve, the most volatile asset in the ladder, needs 2.90 standard deviations, and the gold sleeve needs 6.32. Both are far slacker than anything in the core. A 5 point band on a small holding is loose regardless of how the holding behaves, because 5 percentage points is simply a long way from a 5% position. Weight dominates volatility over this range, and by some margin.
What an equally strict band looks like on each sleeve
So what would a volatility scaled band actually be? Take the 60% equity sleeve's 1.31 standard deviations as the reference, since that's what a conventional 5 point band on a conventional core already delivers, and solve backwards for the band width that gives every other sleeve the same strictness. The chart above plots the answer.
| Sleeve | Weight | Equally strict band | Same band, as % of the position |
|---|---|---|---|
| Global equities | 60% | 5.00 points | 8.3% |
| Emerging markets | 20% | 4.55 points | 22.8% |
| Long Treasuries | 40% | 4.48 points | 11.2% |
| Nasdaq-100 | 10% | 2.45 points | 24.5% |
| Crude oil | 5% | 2.33 points | 46.6% |
| Gold | 5% | 1.08 points | 21.7% |
The bond answer is worth pausing on. A 40% long Treasury sleeve earns 4.48 points against equities' 5.00, so the conventional band is very nearly right for both. The question the topic starts from, whether bonds and equities can share one band, has a boring answer: at those weights, roughly yes. The interesting failures are elsewhere.
Both rules of thumb are half right, and the units decide which half
Look at the last two columns of that table together and the argument resolves. In percentage points, the equally strict band shrinks as the sleeve shrinks: 5.00 on the equity core, 2.33 on oil, 1.08 on gold. In percent of the position, it grows with volatility: 8.3% on equities, 21.7% on gold, 46.6% on oil. Same six rules. Opposite direction, depending only on how you write them down.
That explains why two pieces of received wisdom contradict each other and both survive. "Widen the band on the volatile sleeve" is right if you're quoting bands as a share of the position. "A 5 point band is already miles too wide for a small holding" is right if you're quoting them in percentage points. The distinction between absolute versus relative bands is the whole disagreement. We looked at what that ambiguity does to a small holding in our work on rebalancing frequency, and at how fast weights move in the first place in the piece on portfolio drift.
The same conclusion arrives from a different direction in our earlier simulation of crypto rebalancing, where the corridors that brought a small, fast-moving sleeve's trading under control were far narrower in percentage points than the conventional rule, and far wider as a share of the position. Equal strictness gives a 5% gold sleeve 1.08 points here, and a 5% oil sleeve 2.33. Two methods, two datasets, one neighbourhood.
What the rebalancing research says about band width
Vanguard's 2022 paper Rational Rebalancing ran every threshold from 1% to 15% in 1% increments through a simulation engine, on data as of June 2022. Their summary of the trade-off is one sentence: "The smaller the threshold, the lower the tracking error and the higher the transaction cost." Under a constraint of no more than 20 basis points of expected tracking error, they found "a threshold-based rebalancing of 3% is optimal for a 60/40 portfolio". Not 5%. Their widest tested threshold is 15%, and every one of them is applied to a two-asset 60/40. Nothing in that range speaks to a 5% sleeve, where equal strictness comes out at 1.08 points.
The turnover consequence is visible in Vanguard's long-run table. Monitoring a 60/40 monthly from 1926 to 2018, a 0% band produced 1,116 rebalancing events. A 1% band produced 426. A 5% band produced 58. A 10% band produced 24. Widening from 1% to 5% cut the trading by a factor of seven while the tax-adjusted return moved from 8.20% to 8.22%, and the 10% band returned 8.39%. Band width is a turnover decision that happens to leave returns alone, which is exactly why the choice can be made on other grounds.
Vanguard's 2024 target-date work adds the cost mechanism. Transaction costs there are modelled "as a function of market volatility and trade size", and are "expected to be higher when markets are more volatile and when larger trades are executed". A band that fires often on a volatile asset is buying its trades at the worst possible moments. That is an argument for wider relative bands on volatile sleeves, and it comes from the cost side rather than the drift side.
The strongest objection: equal strictness is not equal risk
Here's the serious case against everything above. Equalising how often a band fires is not the same as equalising how much risk the drift creates, and the second is what rebalancing is for.
A 5 point overweight in a 60% equity sleeve and a 5 point overweight in a 5% gold sleeve are not comparable events. The first moves a portfolio from 60/40 to 65/35. The second doubles a satellite holding. If you scale bands so that both fire at the same frequency, you have made a decision that trade count is the thing being managed. There's a defensible view that risk contribution is the thing being managed instead, and on that view the volatile sleeve should be held tighter, not looser, precisely because its mistakes cost more.
The evidence for what drift does to risk is unambiguous. In Vanguard's run from 1926 to 2018, a 60/40 left alone drifted to an average 85% equity weight and 14.0% annualised volatility, against 11.4% to 11.8% for every rebalanced version. The Yearbook makes the same point from the other direction: since 1900, equities and bonds have each lost more than 70% in real terms on several occasions, "yet a 60:40 equity:bond blend has never declined more than 50%". The blend is the protection. Anything that lets the blend wander is spending some of it.
That objection has real force, and it's why the framework above answers a narrow question. It says what band width makes a rule equally sensitive across sleeves. It doesn't say sensitivity is what a band should equalise. Those are separate arguments, and this evidence only settles the first.
What this evidence cannot tell you
The volatility ladder is option-implied, not realised. Cboe's indices measure what option prices say about the next 30 days, and that number typically sits above what actually happens, because sellers of options charge for the risk. The ratios between markets are more reliable than the levels. That the DMS record puts long-run equity volatility at 23.4% while the VIX averaged 18.16 over fifteen years is a good illustration of how much the measurement choice matters. The relationship between assets is a much sturdier finding than any single figure.
The bond rung is the shakiest. VXTLT is built on options on a 20+ year Treasury fund, and Cboe only began publishing it live on 12 August 2024, so the earlier history comes from Cboe's own reconstruction. Duration also does most of the work here. State Street's long Treasury fund carried an option adjusted duration of 14.22 at 30 June 2026, against 5.88 for its aggregate bond fund. Swap one for the other and the bond sleeve's volatility roughly halves, and its equally strict band moves with it. "Bonds" is not one asset class for this purpose.
The sample is also short and American. Fifteen years of option data covers one long equity bull run, one pandemic, and one inflation shock. It does not cover a 1970s. Every index in the ladder is priced on US-listed options in US dollars, so currency effects sit inside the numbers rather than beside them. And the six indices used here carry no crypto measure, which is why the most volatile sleeve anyone actually holds is missing from the table entirely. The relationship between volatility and the risk you feel is itself contested, which we looked at separately under volatility vs drawdown.
Finally, the whole calculation assumes the rest of the portfolio holds still. It doesn't. When the remainder moves too, the gap between a sleeve and everything else opens faster than these figures suggest, so every band above fires more often in practice than on paper. The direction of that error is at least known.
What would change the conclusion
If bands were quoted as a share of the position by default, most of this disappears. The arithmetic that makes a 5 point band absurd on a 5% sleeve is entirely a consequence of quoting the band in portfolio percentage points. A house that writes all its bands as 20% of target has already solved the problem without knowing there was one.
If correlations went to one, the volatility ladder would stop mattering. These calculations work because a sleeve can move while the rest of the portfolio doesn't. In a crisis where everything falls together, weights barely shift, bands don't fire, and the rule you chose is irrelevant to what happens next.
If a small sleeve stopped being small, the arithmetic inverts. The 6.32 standard deviation slackness of a 5 point band on a 5% gold position is a fact about 5%, not about gold. At 20% the same holding needs a 33.3% move, which is inside one bad year. Any allocation that has been allowed to run is quietly operating under a different rule from the one written down.
The figure worth watching isn't the band. It's the ratio of your band to the sleeve it governs, and whether that ratio is roughly the same across the positions you hold. LedgerTouch shows current weights against target continuously, which is the number that ratio is built from. Two sleeves with the same band and very different ratios are not being managed by the same rule, whatever the rule says.