Rebalance Frequency Study on a 9-Asset Leveraged Portfolio

Question

Given a 9-asset leveraged portfolio, what's the Sharpe-optimal rebalance cadence?

Options tested: - Daily - Weekly (Monday close) - Monthly (month-end) - Quarterly (quarter-end) - Yearly (year-end) - 30% band (rebalance when any position drifts > 30% from target)

Methodology

Portfolio: Same 9-asset weights as a sample heavy-leveraged portfolio (internal-only study)
Data: Real ETF data, 2012-02-02 → 2026-05-18 (14.3 years, common-start window)
Cost model: 1bp × turnover on each rebalance
Starting capital: $10,000

Results

Frequency	CAGR	MDD	Sharpe	Calmar	Final $	Rebals	Avg turnover/rebal
Daily	29.16%	-59.28%	0.944	0.492	$384,268	3,592	1.74%
Yearly	29.12%	-57.19%	0.949	0.509	$382,592	14	30.49%
Quarterly	28.07%	-60.78%	0.927	0.462	$340,510	57	13.88%
Monthly	27.64%	-60.15%	0.918	0.459	$324,389	171	7.98%
Weekly	27.59%	-59.36%	0.911	0.465	$322,629	745	3.71%
30% band	27.89%	-59.90%	0.911	0.466	$333,786	70	14.18%

Interpretation

Daily and yearly are tied for best Sharpe (~0.945-0.949). Middle frequencies (weekly, monthly, quarterly) cluster at 0.91-0.93. The 30% band variant is essentially equivalent to weekly. Spread across all frequencies: about 0.038 Sharpe.

The mechanism:

Daily wins by keeping the portfolio precisely at target weights at all times (small drift, small rebalances).
Yearly wins by letting the high-CAGR LETFs run a full year without being trimmed prematurely (captures momentum within each year).
Middle frequencies sit in a worse zone — trimming winners too often (giving up momentum) without the precision benefit of daily.

This is a real and somewhat counterintuitive finding. The conventional wisdom that "monthly or quarterly is the sweet spot" doesn't hold for a high-vol multi-leverage portfolio like this one.

For tax efficiency (taxable accounts):

Yearly is the clear winner. 14 trades over 14 years means ~1 rebalance per year, generating predictable annual short-term capital gains, minimizing wash-sale complexity, and keeping after-tax returns much higher.

For tax-deferred (401k/IRA):

Daily and yearly are essentially tied on Sharpe. Yearly wins on operational simplicity (14 trades vs 3,592 over 14 years).

The 30% band variant:

Middle of the pack. Not bad, not best. If you want "set and forget with a guardrail," yearly rebalancing outperforms the 30% band on Sharpe, MDD, AND turnover. The band approach offers no real advantage here.

Practical implication

For Implication for heavy-leveraged portfolios with ZROZ: yearly rebalancing is probably the right frequency. It maximizes Sharpe (tied for best), maximizes tax efficiency, minimizes operational burden, AND has the best max drawdown of any frequency tested. Strictly dominant on three of four metrics.

The only argument for daily would be "I want to be precisely at target weights at all times for psychological reasons" — which is real but expensive in tax-managed accounts.

Caveats

Window is bull-tilted. 2012-2026 was good for the high-CAGR LETF holdings. The "let winners run" benefit of yearly rebalancing wouldn't be as large in a sideways or bear regime where rebalancing captures mean-reversion gains.
9-asset portfolio at ~2x effective leverage. A different portfolio composition (fewer assets, lower leverage) would show a smaller spread between frequencies. The 0.04 Sharpe gap is partly an artifact of high-vol holdings benefiting from "let winners run."
1bp/side cost assumption may understate friction for less-liquid holdings (AGQ, SLVP, ZROZ specifically). Higher cost would push the answer slightly toward less frequent rebalancing.
Single window (2012-2026) without stress testing on dotcom or GFC. A different regime might show different optimal frequency.

Source

Saved log: /tmp/defensive_and_rebal.log (TEST B section).

Inline runner pattern:

import pandas as pd, numpy as np
import sys; sys.path.insert(0, '/Volumes/Mac External/Claudes/trader/src')
from trader.data.yfinance_src import fetch_daily

PORTFOLIO = {"AGQ":0.10,"UGL":0.10,"ZROZ":0.25,"SLVP":0.05,
             "UPRO":0.05,"QLD":0.05,"TQQQ":0.15,"USD":0.20,"SOXL":0.05}

prices = {sym: fetch_daily(sym)["close"] for sym in PORTFOLIO}
# Align on common index, compute daily returns
# Simulate: each rebalance day, snap to target weights, apply cost
# Vary rebalance_set across {daily, weekly, monthly, quarterly, yearly, band}
# Compute Sharpe / MDD / final equity for each

a sample heavy-leveraged portfolio (internal-only study) — full portfolio analysis using weekly as the default