Sensitivity to Rebalance Frequency

A monthly-rebalanced factor strategy shows a gross Sharpe ratio of 1.5. Switch to weekly rebalancing and gross Sharpe rises to 1.7 — more frequent rebalancing captures the signal more precisely. But turnover roughly quadruples, and after realistic transaction costs the weekly version's net Sharpe falls to 0.9 while the monthly version's net Sharpe only falls to 1.3. The "better" frequency flips entirely once costs are included. Rebalance frequency looks like an operational detail, but it interacts with both signal decay and trading costs in ways that can dominate a backtest's headline result.

Two effects pulling in opposite directions

More frequent rebalancing generally captures a signal closer to its freshest, most informative state — a monthly rebalance trades on a signal that may already be a few weeks stale by the time it's acted on and then held for weeks more before being updated again, while a daily rebalance stays closer to current information throughout. This pulls gross performance up as frequency increases, for signals that decay over time. But every rebalance generates trades, and every trade has a cost — more frequent rebalancing means more turnover, and turnover scales roughly linearly with rebalance frequency for a strategy that meaningfully changes its positions each time. This pulls net performance down as frequency increases. The frequency that maximizes gross performance and the frequency that maximizes net performance are often different points, and only the net comparison determines what a strategy can actually earn.

Worked example: sweeping rebalance frequency

Gross and net Sharpe ratios at four rebalance frequencies, assuming 15 basis points of round-trip cost per unit of turnover:

Frequency	Turnover (annualized)	Gross Sharpe	Cost drag	Net Sharpe
Quarterly	2.0×	1.1	0.03	1.07
Monthly	6.0×	1.5	0.09	1.41
Weekly	26.0×	1.7	0.39	1.31
Daily	130.0×	1.9	1.95	-0.05

Gross performance rises monotonically with frequency, as expected for a decaying signal captured more freshly each time. Net performance peaks at monthly and collapses at daily, where cost drag overwhelms the entire gross edge. A researcher who only backtests and reports the daily frequency — perhaps because it has the best-looking gross Sharpe ratio — would be recommending the single worst net choice on the table.

What this means in practice

Always report a rebalance-frequency sweep net of realistic costs, never gross-only — gross performance alone systematically favors frequencies that are too high, because it's blind to the cost side of the tradeoff entirely. The right frequency also depends on how fast the underlying signal actually decays: a slow-moving valuation signal gains little from daily rebalancing and should rebalance rarely; a fast-moving short-term reversal signal may genuinely need near-daily rebalancing despite the cost, because the alternative is trading a signal that's already stale. There is no universally correct frequency — only the frequency correct for that specific signal's decay rate and that specific strategy's cost structure.