Rebalancing is the act of returning a portfolio's weights to their targets after markets have moved them. It is mechanically simple and psychologically difficult, because it requires doing the opposite of what feels right.
Why weights drift
If equities rise and bonds do not, the equity share of the portfolio grows. Left alone, a portfolio intended to be 60% equities becomes 70%, then 75%. Nothing was decided; the risk profile changed anyway.
This is the core argument for rebalancing: it restores the risk that was chosen. It is not a technique for improving returns, and presenting it as one is a common error.
Two approaches
Scheduled. Rebalance at fixed intervals — quarterly, semi-annually, annually. Simple, predictable, easy to explain, and it trades regardless of whether the drift was material.
Threshold-based. Rebalance when a weight moves beyond a band around its target — for example, 5 percentage points. It responds to actual drift rather than to the calendar, and it trades less when markets are quiet.
A hybrid is common and sensible: check on a schedule, act only if a threshold is breached. It gives a regular review without trading for its own sake.
The costs, honestly
- Transaction costs — spreads, commissions, market impact. Material for illiquid holdings.
- Tax consequences — in a taxable account, selling a appreciated position to buy a laggard realises a gain. This can exceed the other costs combined, and it is the reason many taxable portfolios rebalance using new contributions rather than sales.
- Opportunity cost — rebalancing trims what has been rising, which may continue rising. It is a deliberate choice not to chase, and it will sometimes look wrong.
Ways to reduce the cost
- Rebalance with cash flows. Direct new contributions and income to the underweight holdings. This moves the weights without selling anything.
- Use bands, not continuous adjustment. A 5% band trades far less than an exact target.
- Prioritise tax-advantaged accounts. If some accounts are sheltered, rebalancing there avoids the tax cost entirely.
- Never trade a small amount. Below a certain size, the fixed costs exceed the correction.
The harder discipline
The mechanically difficult part is not the calculation. It is the direction. Rebalancing always involves adding to what has done poorly and trimming what has done well — which, in the middle of a rally, feels like a mistake, and in a drawdown, feels like throwing money after a loss.
Two things make it feasible:
- The rule was set in advance, when the feeling was absent. A stated band is not a judgement about the market; it is an instruction to oneself.
- The purpose is restating the risk, not making a call. The position is not "the laggard will recover" — it is "the portfolio was meant to carry this much risk, and it does not."
When not to rebalance
Sometimes a drift is a decision, and the honest thing is to make it deliberately:
- The objective changed. A horizon shortened, a liability was met, a circumstance altered. Rebalancing back to a target that no longer fits is not discipline, it is inertia.
- The asset class's role changed. If the reason for holding a position no longer applies, the right action is to review the target, not to force a return to it.
- The cost is prohibitive. A small drift in an illiquid holding is often better tolerated than corrected.
The distinction between rebalancing and changing the policy is worth keeping clear. The first restores an existing decision. The second is a new one, and it deserves the same evidence that the original did.