What it is
Rebalancing means selling what has grown and buying what has shrunk to return a portfolio to its target weights. Calendar rebalancing does this every quarter or year; band rebalancing does it whenever an asset's weight strays outside a tolerance band (for example, 5 percentage points absolute, or 20 percent relative) around its target. This playbook explains the band approach, which trades less in calm periods and more in volatile ones, and is a small, real source of "free" return on a diversified portfolio, with an emphasis on small.
The logic
A rebalanced portfolio of two uncorrelated, mean-reverting-relative-to-each-other assets earns a slightly higher compound return than the weighted average of the two held separately, because rebalancing systematically sells the relatively expensive asset and buys the relatively cheap one. The size of the effect depends on the assets' volatility and how weakly correlated they are; for a stock-bond portfolio it is a fraction of a percent per year. Bands improve on the calendar by only trading when the drift is large enough to matter, which cuts costs and taxes.
The other side of a rebalancing trade is a momentum trader: you are selling what just went up and buying what just went down. Over short horizons momentum is often right, which is why band rebalancing with wide bands (which lets winners run for a while) has historically beaten frequent calendar rebalancing.
Setup rules
- Market: any multi-asset portfolio of index funds: equities, bonds, real estate, commodities, cash. Not a portfolio of correlated equity funds, where rebalancing has little to harvest.
- Timeframe: check weights monthly (or on any large market move); trade only when a band is breached.
- Target weights: set by the investor's plan; the playbook does not choose them.
- Bands: absolute band of 5 percentage points for assets with a target of 20 percent or more; relative band of 25 percent of the target for smaller allocations (a 10 percent target has a 7.5 to 12.5 percent band).
- Rebalance rule: when any asset breaches its band, rebalance that asset back to its target (or halfway back, to reduce turnover further) using new contributions first and trades second.
- Tax rule: in taxable accounts, prefer rebalancing with cash flows and dividends, and harvest losses when a sale is required; see wash-sale-rule.
Entry, stop, target
Rebalancing has no entries in the trading sense, but each band breach is a trade with a defined size. There is no stop; the stop is the allocation itself.
| Item | Value | Notes |
|---|---|---|
| Target | 60 percent equities, 40 percent bonds | Example only |
| Equity band | 55 to 65 percent | 5 points absolute |
| Trigger | Equities at 66 percent after a rally | Breach of upper band |
| Action | Sell 6 percent of portfolio in equities, buy bonds | Back to 60/40 |
| Typical frequency | 0 to 3 rebalances per year | More in volatile years |
| Expected benefit | Small; on the order of tenths of a percent per year | Plus risk control |
The rebalance also has a risk-control function that dominates the return function: it prevents a portfolio from drifting to 80 percent equities after a decade-long bull market and then taking the full bear.
Position sizing and risk
Position size is the target allocation; band width is the risk tolerance around it. A wider band lets winners run longer and reduces trades, at the cost of larger drift and a bigger bear-market exposure at the top. Read the allocation discussion in /learn/risk-management before choosing bands; /tools/position-size is not relevant to this playbook except for a satellite. One practical rule: never rebalance with borrowed money, and never widen a band in the middle of a rally to avoid selling.
What breaks it
- Trending regimes. In a decade-long equity bull market, rebalancing into bonds repeatedly gives up return; the band approach reduces but does not eliminate this.
- Correlated crashes. When every asset falls together, rebalancing harvests nothing, and the risk-control benefit is limited to whatever cash and treasuries were held.
- Costs and taxes. Each rebalance realises gains in taxable accounts; the harvested return can be smaller than the tax paid, which is why cash-flow rebalancing comes first.
- Mis-specified assets. Rebalancing into an asset in secular decline (a sector, a country) is averaging down with a rule. The band rule assumes the target allocation is sound.
- Behavioural failure. Investors skip the rebalance that sells the winner and buys the loser, because it feels wrong, and that is the one that matters.
How to test it
Simulate a two or three-asset portfolio over 40 or more years of monthly total returns under no rebalancing, annual rebalancing and several band widths. Report compound return, max-drawdown, number of trades and, if you can model it, taxes. Expect the differences in return to be small and the difference in drawdown to be noticeable. Then test on different asset pairs; the benefit should grow with the assets' volatility and shrink with their correlation. Because the effect is small, be careful not to read noise as edge; see expectancy-system-evaluation.
Variations
- Calendar plus band: check quarterly, trade only if outside the band.
- Volatility-aware bands that widen when realised volatility is high, reducing whipsaw trades in crashes.
- Trend-gated rebalancing: delay buying an asset that is below its 200-day average until it crosses above; see trend-following-200-day.
Further reading
diversification, correlation, max-drawdown, etf, wash-sale-rule, volatility, mean-reversion, disposition-effect, process-over-outcome, index.
Related playbooks: rules-based-dca, dividend-growth-core, trend-following-200-day, dual-momentum