Rebalancing Is a Risk Rule, Not a Return Strategy

Rebalancing is sold as a free lunch - trim the winner, top up the loser, collect a bonus without timing anything. The risk-control case for it is unconditional. The return case is a bet on a market regime, and the two get merged into a single claim they cannot both support.

What follows separates them, using the standard citations and the most recent implementation evidence.

The Cited Formula Describes Diversification, Not Rebalancing

The bonus is usually attributed to Booth and Fama’s 1992 diversification return - roughly half the variance reduction from combining assets that do not move in lockstep. The result is real and well-established.

It is also the wrong attribution. That arithmetic describes what diversification does; later work correcting the point establishes that any actual rebalancing bonus comes from the rebalanced assets reverting toward each other, not from the variance identity. The distinction matters because variance reduction is a property of the portfolio, available whether or not anyone rebalances, while mean reversion is a property of the period.

In a Trending Regime the Rule Costs Money

Perold and Sharpe’s classic result is the counterexample: in a sustained one-direction market, a fixed-mix rebalanced portfolio underperforms buy-and-hold, because the rule forces repeated trimming of whichever asset keeps winning.

Structurally, rebalancing is a bet on reversion whether or not it is labelled one - sell whatever has drifted furthest above target, on the expectation it gives some of the move back. Two assets that swap the lead deliver exactly that. Two assets where one simply keeps winning - not a rare regime; entire decades have looked like this - deliver the opposite, and the rule has no way to identify which regime it is in before the fact.

The Unconditional Job Is Holding Risk Where It Was Set

Left alone, winners grow as a share of the total and the risk profile drifts with them. A 60/40 untouched through a few strong equity years is no longer a 60/40 and carries more equity risk than anyone chose.

Restoring the split is the one thing rebalancing does in every regime. It is provable in advance, it does not depend on what the market subsequently does, and it is the only part of the case that survives without a forecast attached.

Threshold Beats Calendar on Implementation

Vanguard’s December 2024 work found a 200/175 basis-point threshold band beat monthly calendar rebalancing by roughly 15-22 basis points a year in accumulation and 22-25 in decumulation, at about a quarter of the transaction costs.

Read that as an implementation finding rather than a verdict on the return question. It measures the cost of acting on a schedule rather than on drift - which is a separate matter from whether rebalancing adds return at all.

What This Means for Allocators

Make the risk-control case and stop there. It is unconditional, it is provable, and it does not need the return argument to justify the policy.

Rebalance on drift bands rather than dates. The evidence on cost is clear even where the evidence on return is regime-dependent.

And do not promise a bonus the rule cannot deliver on demand. Whether rebalancing helps or costs over any stretch depends on which regime the market was in, and that is not something the rebalancing rule can answer in advance.