Once a Year at Christmas — How Simple Annual Rebalancing Determines the Long-Term Performance of Leveraged ETFs
"Don't pick the flower too soon, and give it fertilizer only when there's enough."
This study backtests a simple strategy — hereafter Christmas Rebalancing Strategy — that allocates 50:50 between 3× leveraged Nasdaq-100 ETF (TQQQ) and cash, rebalancing back to 50:50 once a year at Christmas (Dec 27), over a 30-year period from 1995 to 2024. The strategy achieved a cumulative return of +617× ($100k → $61.8M), CAGR of +23.9%, and MDD of −87.9%, outperforming simple buy-and-hold of the same asset (B&H: $15.1M / +18.2% / −99.97%) by 4.1× in absolute return with 12 percentage points better MDD. Notably, applying the same strategy to QQQ (1×) results in underperformance versus simple Buy & Hold, consistent with the academic finding that the value of rebalancing scales non-linearly with asset volatility.
Keywords: leveraged ETF, annual rebalancing, lifecycle investing, volatility decay, backtest, TQQQ, QQQ
Leveraged ETFs have long received two diametrically opposed assessments. On one hand, the conventional wisdom holds that volatility decay makes them unsuitable for long-term holding. On the other, certain investors — most notably the Bogleheads' HFEA (Hedgefundie's Excellent Adventure) community — have reported empirical cases where leveraged ETFs combined with disciplined rebalancing can dramatically outperform the broader market.
This study focuses on a single decision variable: Given a 50:50 allocation between a leveraged ETF and cash, how does rebalancing frequency determine long-term performance? Specifically, we examine whether annual rebalancing holds any advantage over other frequencies (monthly, quarterly, event-driven) across 30 years of data.
We name this strategy Christmas Rebalancing Strategy because: (1) the investor executes it once a year around Christmas (Dec 27, near the last trading day of the year) — a date no one forgets; (2) it is calendar-based rather than signal-based, eliminating timing bias; and (3) simplicity is the core — the ability to follow a single rule for 30 years is what makes it viable.
The Christmas Rebalancing Strategy Rule (one line):
Every year at Christmas (Dec 27), rebalance the portfolio back to TQQQ 50% / Cash 50%.
Ayres (Yale Law) and Nalebuff (Yale SOM) proposed a 2:1 margin leverage strategy in early career that gradually de-levers over one's lifetime, using U.S. equity data from 1871. They reported +19% higher expected retirement wealth versus a 100% equity strategy, and +90% versus Target Date Funds. Christmas Rebalancing Strategy abandons the Ayres-Nalebuff lifecycle-stage adjustment in favor of a fixed 50:50 ratio with annual resets for maximum simplicity.
Proposed by an anonymous Bogleheads forum poster: UPRO (S&P 3×) 55% + TMF (Long-term Treasury 3×) 45%, rebalanced quarterly, grounded in risk parity principles. Key differences from Christmas Rebalancing Strategy: safe asset (TMF vs. cash), frequency (quarterly vs. annual), and underlying (S&P 3× vs. Nasdaq 3×). HFEA faced a severe test in 2022 when TMF plunged −70% in an inflationary environment.
Fischer Black and Robert Jones proposed Constant Proportion Portfolio Insurance in 1987, dynamically adjusting risky asset exposure relative to a defined floor. Christmas Rebalancing Strategy is mathematically very similar to CPPI with multiplier=1 and floor=0, but replaces dynamic adjustment with a fixed 50:50 ratio and annual reset.
A recent arXiv preprint, "Compounding Effects in Leveraged ETFs: Beyond the Volatility Drag Paradigm", mathematically demonstrates that long-term returns of leveraged ETFs are determined not simply by volatility drag but by return autocorrelation. In trending markets (positive autocorrelation), leverage amplifies gains; in mean-reverting markets (negative autocorrelation), it amplifies losses. Our results — where 50:50 rebalancing dominates during the dot-com bubble/bust/recovery cycle — are consistent with these theoretical predictions.
Day 0:
Capital = $100,000
Buy TQQQ = $50,000 worth (fractional shares)
Hold cash = $50,000
Day t (Dec 27 each year, ≈ Christmas):
total = TQQQ_value + cash
target_TQQQ = total × 0.5
if TQQQ_value > target_TQQQ:
sell (TQQQ_value − target_TQQQ) → cash
else:
buy (target_TQQQ − TQQQ_value) ← cash
Commission: 0.03% per trade
Cash interest: 0% (conservative)| Rank | Strategy | 30-Yr Final | 30-Yr CAGR | 30-Yr MDD | Calmar |
|---|
Period A (1995-2010: Dot-com Bubble/Bust + Financial Crisis): Christmas Rebalancing Strategy dominates all strategies. TQQQ B&H loses 18% of capital despite 16 years of holding.
Period B (2010-2024: Bull Market + Single Crash): Simple buy-and-hold dominates. Christmas Rebalancing Strategy trails in absolute returns but maintains the lowest MDD at −42%.
Applying the same strategies to unlevered QQQ yields the opposite outcome. With QQQ 1×, simple Buy & Hold dominates all strategies ($4.81M vs. Christmas Rebalancing Strategy $1.04M). The 50% cash allocation, earning 0% for 30 years, creates a massive opportunity cost.
The stark contrast between TQQQ (Christmas Rebalancing Strategy $61.8M vs. B&H $15.1M, 4.1× outperformance) and QQQ (Christmas Rebalancing Strategy $1.04M vs. B&H $4.8M, 0.22×) suggests the following relationship:
The value of rebalancing scales non-linearly with asset volatility.
This is consistent with the autocorrelation theory from Section 2.4. In highly volatile assets with both strong trending and mean-reverting phases, rebalancing captures a virtuous cycle: automatic profit-taking at bubble peaks and automatic buying after crashes. In low-volatility assets, simple buy-and-hold prevails.
Comparing 30-year CAGR across rebalancing frequencies, the principle "less frequent = better" holds consistently across both eras. During the 1995-2000 dot-com bubble (a 5-year trend), monthly rebalancing repeatedly trimmed gains, resulting in cumulative underperformance. Annual rebalancing captured the bulk of the bubble's upside.
From the autocorrelation perspective: short-term (daily to weekly) returns are dominated by mean reversion → frequent rebalancing hurts; long-term (quarterly to annual) returns are dominated by trends → frequent rebalancing hurts. Cycle-level periods (1-3 years) capture the bubble ↔ crash oscillation, making annual rebalancing the optimal capture frequency. Once per year may not be accidental — it may coincide with the natural unit of financial market cycles.
The central findings of this study:
This study also evaluated two strategies popular in Korean investment communities:
Both strategies worked in a single era (2010-2024), but over the full 30-year horizon, they fall far short of the simple Christmas Rebalancing Strategy approach.
This study reports a 30-year backtest of Christmas Rebalancing Strategy — rebalancing once a year at Christmas to maintain TQQQ 50% / Cash 50%. Key findings:
The strategy's appeal lies in its simplicity — one rule, one action per year, no market signals, the possibility of unwavering execution for 30 years. Yet that simplicity is also its greatest challenge. Maintaining the rule when 80% of your capital has evaporated is simple on paper but nearly impossible in practice. This behavioral risk is the strategy's true cost — a dimension that neither academic nor practical research has been able to quantify.
Future research directions: performance comparison across different assets (SOXL, SPXL, FNGU), lifecycle variants (dynamically adjusting the 50:50 ratio with age), additional bear-market protection mechanisms, multi-country validation, and statistical sensitivity analysis on start-date dependence.