Everyone agrees you should rebalance. Almost nobody agrees on when. The two standard answers — on a schedule, or when a holding drifts past a threshold — sound like a minor scheduling detail, and turn out to be a decision about costs, taxes and how much attention you want to spend.
You pick an allocation — say 60% equities, 40% bonds — because it matches the risk you are willing to carry. Then the market moves. When equities outrun bonds, they become a larger share of the total without you buying a single unit.
This is the part worth internalising: drift is not neutral. A portfolio left alone does not stay put, it gets riskier, because the asset that grew fastest is usually the volatile one. Doing nothing is an active choice to hold something you did not choose.
Pick an interval — annually, semi-annually, quarterly — and rebalance then, whatever the market has done. That is the whole method.
Its strength is that it is a habit rather than a judgement call. You are not watching prices, so you are not tempted to act on them, and a rule you follow beats a better rule you abandon.
Its weakness is that the calendar knows nothing about your portfolio. It will have you trading in a quarter where nothing moved, and leave you badly drifted for eleven months after a sharp move in month one.
Band (or threshold) rebalancing ignores the date and watches the allocation. You rebalance when something drifts more than a set amount from its target. Trades happen when the portfolio actually needs them.
The catch is that "drifts more than 5%" is ambiguous, and the two readings behave very differently:
Absolute band — measured in percentage points. A 5-point band on a 60% target triggers below 55% or above 65%.
Relative band — measured as a share of the target itself. A 25% relative band on a 60% target triggers below 45% or above 75%.
The difference matters most for your small positions. A 5-point absolute band applied to a 5% holding never triggers until that holding has doubled or vanished — the band is as wide as the position. A relative band scales with the target, which is usually what people meant.
Hypothetical numbers, chosen to be easy to follow — not a forecast and not drawn from any real period.
Start with €60,000 in equities and €40,000 in bonds: €100,000 at 60/40. Equities then rise 30% while bonds are flat. You now hold €78,000 and €40,000, a total of €118,000, and equities are 66.1% of the portfolio.
Absolute drift: 66.1% − 60% = 6.1 percentage points. A 5-point band has been breached — rebalance.
Relative drift: 6.1 ÷ 60 = 10.2%. A 25% relative band has not been breached — do nothing.
Same portfolio, same day, opposite instructions. Neither band is wrong; they encode different tolerances. What would be wrong is not knowing which one you set.
If you do rebalance: 60% of €118,000 is €70,800, so you sell €7,200 of equities and buy €7,200 of bonds.
There is a third option that gets far less attention than it deserves, and it is the one most people adding money regularly should be using. Instead of selling the winner, direct new contributions at whatever is underweight until the target is restored.
Same example. You do not want to sell the equities, so the €78,000 has to stay put and become 60% of a larger portfolio. That means the portfolio needs to reach €78,000 ÷ 0.60 = €130,000. It is currently €118,000, so you contribute €12,000, all of it into bonds.
Bonds become €52,000, and the split is €78,000 / €130,000 = 60% and €52,000 / €130,000 = 40%. Exactly back on target, with nothing sold, no disposal to declare, and no transaction costs beyond the purchase you were making anyway.
The limit of buy-only. It can only ever add, so it cannot fix an overweight position that new money is too small to dilute. The further you have drifted and the smaller your contributions, the longer it takes — and if a holding has grown past its target by more than you can realistically outspend, only selling will correct it.
You will find plenty of confident claims that one method beats the other. Treat them carefully: the ranking depends on the period measured and the assets involved, and a result that holds over one stretch of history need not hold over the next.
The factors you can actually pin down are more mundane, and more useful:
What a trade costs you. Commissions, spreads and — the big one in most jurisdictions — the tax on realised gains. Where selling is expensive, wider bands and buy-only corrections are worth a lot.
Whether you are still contributing. If money is going in regularly, buy-only rebalancing handles most drift on its own and the calendar-versus-bands question gets much less important.
What you will genuinely keep doing. Bands require checking; a date in the diary does not. An annual review you actually perform beats a threshold rule you stop monitoring in March.
A reasonable default for most people: review on a fixed schedule, act only if a relative band has been breached, and correct with new contributions before considering a sale. That is three rules, and it removes almost every occasion to improvise.