Does Portfolio Rebalancing Improve Returns? What Research Shows

Usually not. Portfolio rebalancing is a risk control tool, and over most long periods it costs a little return rather than adding any, because it systematically sells whatever is compounding fastest. What it reliably improves is the worst drawdown and the odds that the portfolio you hold in a crisis is the one you actually chose. The narrow exceptions are portfolios of volatile, similar-return, imperfectly correlated assets, where a real rebalancing bonus can show up.

The claim that rebalancing boosts returns is one of the stickiest ideas in retail investing, and it survives mostly because it sounds like it should be true. Selling high and buying low, on a schedule, without emotion. Who could argue. The trouble is that the assets you are selling high are usually not high, they are just winning, and the ones you are buying low are usually not cheap, they are just losing. This is educational content, not investment advice.

Does portfolio rebalancing actually improve returns?

In a portfolio where one asset has a materially higher expected return than the others, rebalancing lowers expected return by definition. A 60/40 stock and bond portfolio that is rebalanced keeps taking money out of the higher returning asset and putting it into the lower returning one. Do that for thirty years and you end up behind the version that drifted, in return terms. You also end up with a portfolio that never quietly became 85/15 without telling you, which is the whole point.

Vanguard's research is fairly blunt about this framing: the case for rebalancing is risk control, not return enhancement. Their 2024 work on target date funds did find a small edge for threshold based rebalancing over calendar based rebalancing, in the range of 15 to 25 basis points a year, but that is a comparison between two rebalancing methods, not evidence that rebalancing beats leaving things alone.

Where the rebalancing bonus is real

There is a genuine effect, sometimes called the rebalancing bonus or diversification return, and it is worth understanding precisely so you know when to expect it. It shows up when three conditions hold at once:

Stocks and bonds fail the first test badly over most long windows, which is why rebalanced 60/40 portfolios usually trail unrebalanced ones on return. Two volatile assets with similar long run outcomes and low correlation, on the other hand, are the textbook case where the bonus appears. That is why the effect gets discussed most in commodity, managed futures and crypto contexts rather than in plain stock and bond portfolios.

Rebalancing versus buy and hold

Run the honest comparison and read both columns. Over a long single-asset bull run, buy and hold wins on return and loses badly on concentration: by the end, the portfolio is dominated by whatever ran, and the drawdown when it turns is much worse. Over a choppy, mean-reverting period, the rebalanced version often wins on both, because the trimming happens near local highs and the buying near local lows.

Neither of those is a prediction. They are descriptions of what happened in particular windows, and the window you pick decides the answer, which is exactly why picking the window matters more than the conclusion. The discipline is to test your own holdings across several start and end dates rather than the one that makes your preferred answer look good. The traps involved are covered in backtesting pitfalls.

What rebalancing does improve

Three things, all of which matter more than a few basis points of return.

It holds risk near the level you chose. A portfolio left alone through a long equity run does not stay at your target; it becomes progressively more aggressive precisely as valuations get more stretched. Rebalancing is the mechanism that keeps the risk you are running attached to the risk you decided you could tolerate.

It caps single position blowups. Inside an equity basket, a holding that runs for three years can become a third of the portfolio. The next drawdown in that one name is then a portfolio event rather than a position event. This is the same mechanism that turns an equal weight index back into a concentrated one if nobody resets it.

It removes the decision from the moment. A written rule made in advance, when nothing is happening, is worth more than good judgment applied during a 20% decline. Most portfolios do not fail because the math was wrong; they fail because the person changed their mind at the worst possible time.

Does rebalancing cost anything?

Yes, and in a US taxable account the tax cost usually dwarfs the trading cost. Selling an appreciated position realizes a capital gain in that tax year, and positions held a year or less are taxed at ordinary income rates rather than long term capital gains rates. A quarterly schedule realizes far more short term gains than an annual one purely as a function of the calendar. Net capital losses offset gains and then up to $3,000 of ordinary income a year, with the rest carried forward (IRS Topic 409).

That cost has a direct implication: since the return benefit of rebalancing is small or negative and the tax cost is real, the sensible response is to rebalance less often rather than more, and to do as much of it as possible inside tax advantaged accounts. The full breakdown of cadences and their costs is in portfolio rebalancing, and the cadence question specifically in how often you should rebalance your portfolio.

How to test it on your own portfolio

Stop arguing about the general case and check the specific one. Build the allocation you actually hold, run it under two rules over the same window, and compare four numbers: total return, worst drawdown, volatility and number of trades. Then repeat over a different window that includes a bear market. If rebalancing helps your mix, it will show up in the drawdown column consistently and in the return column inconsistently, which is the pattern you should expect.

Be careful about one thing while you do it. It is very easy to keep adjusting the band width and the review date until the rebalanced version wins, at which point you have not learned anything about rebalancing, you have fitted a rule to one history. If the conclusion only holds at exactly 4.5% bands checked in March, it is noise. A real effect survives being approximately right.

Indexes is educational and informational software. It never places trades, connects to a brokerage or holds assets, and nothing here is investment or tax advice. Backtests are hypothetical and past results do not predict future returns.