Rebalancing controls risk; it is not a source of return
The claim
Rebalancing keeps a portfolio's risk near its target, and across plausible calendar frequencies and tolerance bands the choice of rule changes outcomes far less than whether a rule exists at all.
Why we rate it moderate
Tested across many rules, markets and sample periods with consistent results. The consistency is partly why the finding is modest: no rule reliably beats the others, which is itself the conclusion.
The case for rebalancing is about risk, not return. Equities outgrow bonds most of the time, so an unrebalanced portfolio drifts steadily towards equities, and it drifts fastest during exactly the long calm periods that precede the drawdowns it will then be badly positioned for. Someone who chose 60/40 in 2009 and never touched it was holding something far riskier by 2020 without ever deciding to.
The return effect is small and sign-ambiguous. Dichtl, Drobetz and Wambach tested a wide range of calendar and threshold rules across stock-bond mixes and markets, and found no rule that reliably dominates on risk-adjusted return after costs. Sun and co-authors, approaching it as an optimisation problem, reached the same practical conclusion: the tolerance band matters more than the calendar, and both matter less than acting at all.
The so-called rebalancing bonus is not a free lunch. Selling what has risen and buying what has fallen profits when prices mean-revert and loses when they trend, which is why rebalancing modestly underperforms in strong sustained bull markets. Its payoff is contingent on the return process, unlike the risk control, which is mechanical.
The practical rule follows from that. Pick something simple — annually, or when an asset drifts more than five percentage points from target — write it down, and prefer to rebalance with new contributions rather than sales, which costs nothing and avoids realising gains. In a 401(k) rebalancing is free of tax consequence, which is a good reason to do the work there rather than in a taxable account.
The behavioural value may exceed the financial one. A written rule that requires buying the asset that just fell 30% is a commitment device against the impulse to do the opposite, and that impulse is expensive enough to have its own note in this library.
Where this breaks down
- In a taxable account, rebalancing by selling realises gains, and the tax can exceed the benefit. Use contributions, dividends and withdrawals first.
- Rebalancing into a genuinely impaired asset is not a virtue. The logic assumes the assets are broad and permanent, not individual securities that can go to zero.
- Very frequent rebalancing adds cost and can lower returns; very infrequent rebalancing allows the drift the practice exists to prevent. The middle is wide and flat.
- The evidence is about broad stock-bond portfolios. It does not transfer automatically to portfolios of narrow or illiquid holdings.
Sources
Follow these rather than taking our word for the summary.
Hubert Dichtl, Wolfgang Drobetz and Martin Wambach (2016). Testing rebalancing strategies for stock-bond portfolios across different asset allocations
Applied Economics, 48(9), 772-788
Finding: No rebalancing strategy consistently outperformed the others on a risk-adjusted basis after transaction costs, across allocations and markets.
Walter Sun, Ayres Fan, Li-Wei Chen, Tom Schouwenaars and Marius Albota (2006). Optimal Rebalancing for Institutional Portfolios
The Journal of Portfolio Management, 32(2), 33-43
Finding: Optimal rebalancing is a trade-off between tracking error and transaction cost, best approximated by tolerance bands rather than a fixed calendar.