Worked research note
Rebalancing frequency, turnover, and drift
In one pretax 60/40 case, tighter calendar control demanded far more turnover while the unrebalanced reference carried more risk.
By ArthaPilot Research
Published 2026-10-11
Research question
A pretax comparison of turnover, return, and risk across seven frequencies for SPY and BND; implementation costs are not modeled.
Scope: One historical path cannot identify a universally optimal rebalance frequency.
Frequency changed turnover by more than sixteenfold
The fixture holds the 60% SPY and 40% BND target, $250,000 principal, total-return basis, and January 2015 through December 2024 window constant across seven calendar-frequency rows. The worked note cites three anchors: daily, annual, and never rebalance.
Daily rebalancing produced 45.7320% annual turnover. Annual rebalancing produced 2.7894%. The daily figure was more than sixteen times the annual figure; never rebalancing, by construction, recorded zero policy-driven turnover.
Exhibit 01
Content-pinned result
The turnover price of tighter calendar control
The three cited rows from the recorded sweep. Turnover is an implementation quantity, not a modeled dollar cost in this fixture.
How much annual turnover accompanied daily, annual, and no rebalancing in the recorded 60/40 case?
Construction. Hold assets, target weights, principal, dates, and total-return basis constant. Read one-way annual turnover from the daily, annual, and never rows. Convert the response's decimal turnover values to percentages.
| Category | Annual turnover |
|---|---|
| Daily | 45.732% |
| Annual | 2.789% |
| Never | 0% |
Source and construction: Recorded ArthaPilot Rebalancing Sensitivity fixture; 60% SPY and 40% BND; 2015-01-02 through 2024-12-31; response hash a601b4b1a0f2… .
How this figure was calculated
Inputs
- A $250,000 portfolio fixed at 60% SPY and 40% BND.
- The committed seven-frequency response over 2015-01-02 through 2024-12-31.
- The request's disabled offset setting, so no tax offset enters any row.
- No inflation adjustment, so every displayed figure is nominal.
Calculation
- Hold assets, target weights, principal, dates, and total-return basis constant.
- Read one-way annual turnover from the daily, annual, and never rows.
- Convert the response's decimal turnover values to percentages.
What this step cannot establish
Turnover measures trading activity. This fixture does not convert activity into transaction costs or tax drag.
Exact values behind the figure
| Series | Calendar policy | Annual turnover (%) |
|---|---|---|
| Annual turnover | Daily | 45.732% |
| Annual turnover | Annual | 2.789% |
| Annual turnover | Never | 0% |
The recorded sweep makes the implementation gradient visible: a more frequent rule traded much more of the portfolio. It does not establish that daily or annual rebalancing was economically costly after spreads, commissions, or taxes because those frictions were disabled.
Turnover should therefore be read as the amount of activity a later cost model would need to price. It is not tax drag or an all-in expense estimate.
The highest return belonged to the drifting reference
Daily, annual, and never-rebalance CAGR were 8.59%, 8.50%, and 9.57%, respectively. A return-only ranking would put the never-rebalance row first.
But the same row also had the highest recorded volatility, 12.39%, and maximum drawdown, 23.96%. The fixture therefore does not support the claim that avoiding rebalancing improved the strategy while holding risk constant. Risk was allowed to move with the weights.
Exhibit 02
Content-pinned result
Return did not move independently of risk
CAGR for the three cited policies, with the recorded risk statistics attached to the no-rebalance reference.
Did the return ordering stay separate from the risk measures in the same recorded path?
Construction. Read the three response metrics for each selected frequency. Place return, volatility, and drawdown on the same policy categories without forming a composite score.
| Category | CAGR | Volatility | Maximum drawdown |
|---|---|---|---|
| Daily | 8.59% | 11.02% | 21.8% |
| Annual | 8.5% | 10.84% | 21.02% |
| Never | 9.57% | 12.39% | 23.96% |
Source and construction: Same recorded ArthaPilot response; total returns, no inflation adjustment, no transaction-cost model, and no investor-tax model.
How this figure was calculated
Inputs
- Pretax CAGR, volatility, and maximum drawdown from the daily, annual, and never rows.
- Identical dates, principal, holdings, and total-return basis across rows.
Calculation
- Read the three response metrics for each selected frequency.
- Place return, volatility, and drawdown on the same policy categories without forming a composite score.
What this step cannot establish
The chart describes one realized path. It does not hold the drifting portfolio's risk exposure constant or identify a winning policy.
Exact values behind the figure
| Series | Calendar policy | Annualized or peak-to-trough value (%) |
|---|---|---|
| CAGR | Daily | 8.59% |
| CAGR | Annual | 8.5% |
| CAGR | Never | 9.57% |
| Volatility | Daily | 11.02% |
| Volatility | Annual | 10.84% |
| Volatility | Never | 12.39% |
| Maximum drawdown | Daily | 21.8% |
| Maximum drawdown | Annual | 21.02% |
| Maximum drawdown | Never | 23.96% |
The no-rebalance row is a drifting-risk reference, not a winning policy. Its higher realized return may reflect the exposures it was permitted to accumulate along this particular path.
A policy comparison becomes decision-useful only after the desired allocation control is stated. Without that target, the exercise silently rewards whichever drift happened to be favorable.
What this worked case leaves unanswered
The recorded response is deliberately narrower than the full research question. It does not include tolerance bands, nearby execution-date offsets, transaction costs, or investor taxes. It also does not include the proposed four-fund portfolio.
Each can change the result: bands alter when a trade occurs, offsets change the prices encountered, and taxes convert sell-side turnover into a household-specific ledger.
Exhibit 03
Method note
Measured here and outside this note
A scope boundary that separates the verified calendar-frequency case from excluded comparisons.
How did turnover vary across every calendar frequency actually recorded in the fixture?
Construction. Convert each annual-turnover decimal to percentage points by multiplying by 100. Order the points from daily through never to preserve the requested frequency sequence.
| Category | Recorded turnover |
|---|---|
| Daily | 45.732% |
| Weekly | 18.863% |
| Biweekly | 13.605% |
| Monthly | 8.967% |
| Quarterly | 5.526% |
| Annually | 2.789% |
| Never | 0% |
Note: The adjacent scope ledger identifies comparisons absent from this pretax calendar-frequency fixture.
Source and construction: Scope of the committed ArthaPilot Rebalancing Sensitivity fixture.
How this figure was calculated
Inputs
- The seven requested calendar frequencies and their response turnover_per_year values.
- The response's null after-tax fields and no-cost effective assumptions, retained as scope boundaries.
Calculation
- Convert each annual-turnover decimal to percentage points by multiplying by 100.
- Order the points from daily through never to preserve the requested frequency sequence.
What this step cannot establish
The curve describes calendar-frequency turnover before costs and taxes. It contains no band, offset, after-tax, or implementation-cost result.
Exact values behind the figure
| Series | Calendar frequency | Annual turnover |
|---|---|---|
| Recorded turnover | Daily | 45.732% |
| Recorded turnover | Weekly | 18.863% |
| Recorded turnover | Biweekly | 13.605% |
| Recorded turnover | Monthly | 8.967% |
| Recorded turnover | Quarterly | 5.526% |
| Recorded turnover | Annually | 2.789% |
| Recorded turnover | Never | 0% |
The present case is publishable as a worked note about a real implementation trade-off: tighter calendar control coincided with much higher turnover, while no rebalancing changed the return-and-risk profile.
This note does not estimate tax drag or identify an optimal rebalance rule. Those questions are outside its recorded evidence.
Method and constraints
Method
The displayed pretax calendar-frequency results reconcile to one committed request and response fixture.
Verification: Preserve the fixture path, source commit, engine version, and request/response SHA-256 hashes with the note.
Verification: Reconcile the displayed CAGR and turnover values to the committed response and retain the decimal-versus-percent unit convention.
Verification: Keep the no-rebalance volatility and maximum drawdown attached to its CAGR whenever the row is compared.
Assumptions held fixed
- The recorded portfolio is fixed at 60% SPY and 40% BND with a $250,000 starting principal.
- The common market window is 2015-01-02 through 2024-12-31 on a total-return basis without inflation adjustment.
- The fixture contains seven calendar-frequency rows; the note quotes the daily, annual, and never-rebalance anchors recorded in the worked artifact.
- Transaction costs and investor taxes are disabled, so turnover is not converted into dollars or tax drag.
What this study establishes
Rebalancing does not have a single cost. It trades implementation activity for control of a target allocation. In the recorded 60% SPY and 40% BND case, daily rebalancing generated 45.7320% annual turnover, annual rebalancing generated 2.7894%, and never rebalancing generated none.
Never rebalancing recorded the highest CAGR of the three cited rows, 9.57%, and also the highest volatility and maximum drawdown, 12.39% and 23.96%. Return and risk changed together as the portfolio drifted. This fixture contains no tax or transaction-cost model, so it cannot answer the after-tax question.
Limitations
- One historical path cannot identify a universally optimal rebalance frequency.
- The fixture does not measure allocation drift in the facts used here, so the report does not quantify how much drift each policy removed.
- It does not model transaction costs, taxes, tolerance bands, execution-date offsets, or cash-flow-assisted rebalancing.
- Results for one SPY-and-BND portfolio need not carry to other assets, weights, or correlation regimes.
Sources and prior research
- Vanguard Investment Strategy Group, Rational rebalancing: an analytical approach to multiasset portfolio rebalancing decisions and insights (2022)
Primary firm research on calendar, threshold, and combination policies and the trade-off between allocation control and transaction costs.
- Vanguard, Rebalancing your portfolio (2026)
Authoritative explanation that rebalancing is an allocation-control process rather than a market-timing rule, with tax-aware implementation considerations.
- ArthaPilot, Rebalancing Sensitivity guide (2026)
Calculation contract for the recorded calendar-frequency sweep and its excluded policy comparisons.
Evidence and reproducibility
cost-of-rebalancing-recorded-demo-fixture
A descriptive worked note comparing recorded pretax calendar-frequency rows for one 60% SPY and 40% BND portfolio.
Evidence date: 2026-08-17. Basis and units are stated in each recorded fact.
note=434808c8914438bd7e232fe34abb676e97d5c7c5bd1e058b4869c5cb3d3a9509; request=ce0b0ef1e34876faface54e7cb3fdad7e7ff2da4ffaaa4e1d481db67e643f85d; response=a601b4b1a0f2c3ec3d692eb87850ff308b3924c9f65e201c4111a7a58144e6ef
Source: web/src/data/research/portfolio/fixtures/rebalancing.record.json. Versions: fixture1=69daa3e2f3fd5efe2faa78e6b08748eee1bc250c@1.0.0.
- An after-tax or transaction-cost ranking
- A result for tolerance bands, execution-date offsets, or a second portfolio
- A universal optimal rebalance rule
Inspect the same question with your own inputs
The linked tools expose the assumptions, dates, and result tables behind this report.
- Compare rebalance frequencies Hold the portfolio fixed while inspecting turnover and outcomes across calendar policies.
- Backtest one policy Inspect the full wealth path, drawdown, trade history, and assumptions for a selected rule.
- Model tax-aware allocation Move from pretax turnover to a household-specific allocation and realization context.