A 2x leveraged ETF promises exactly twice the daily return of its underlying. In practice it consistently delivers less. This gap — the tracking error — is not random noise. It is a structural, predetermined daily deduction built into the product's swap mechanics, and it accrues to the short seller unconditionally regardless of market direction.
AVL, the T-Rex 2x Broadcom ETF, achieves its 2x exposure through a total return swap. The counterparty delivers 2x AVGO's daily return but charges a financing rate comprising the risk-free rate plus a spread for the difficulty of hedging a single-stock position. This cost is deducted from NAV every trading day — silently, whether AVGO rises or falls.
Daily mechanics: If AVGO rises 1%, AVL delivers approximately +1.95% not +2.00%. If AVGO falls 1%, AVL delivers approximately -2.05% not -2.00%. The drag operates in both directions. The correlation of daily strategy P&L with AVGO's return in this backtest is 0.002 — statistically indistinguishable from zero.
The position is structured as $10,000 long AVGO / $5,000 short AVL, rebalanced daily to maintain delta neutrality. The gross spread — observed tracking error of 23.12% minus borrow cost of 0.6% — is the return available before friction.
Before presenting backtest results, it is necessary to address a significant anomaly. December 2025 contributed $2,157 of the strategy's total $3,149 P&L — 68% of all profit from a single month out of 18. Any unqualified presentation of headline returns would be misleading without this disclosure.
What caused the December anomaly: AVGO fell 14.1% in December 2025. The daily compounding mechanic of a 2x leveraged ETF causes losses to compound asymmetrically over sustained directional moves — volatility decay. AVL fell substantially more than 2x over the month, generating outsized tracking error of 17.4% against a typical monthly run-rate of 0.3–1.7%. This P&L was real, but it required a specific confluence of high volatility and directional momentum that cannot be reliably expected to repeat.
All results in this paper are presented on two bases: full sample (including December) and adjusted (December P&L replaced with a structural-TE-only estimate). The adjusted figures are the author's preferred basis for evaluating forward-looking merit.
The backtest runs from AVL's inception (October 10, 2024) through March 13, 2026 across six distinct market regimes. Position: $10,000 long AVGO, $5,000 short AVL, 0.6% annual borrow, rebalanced daily.
| Metric | Full sample | Adjusted (excl. Dec windfall) | Weekly rebal (full) |
|---|---|---|---|
| Total return | +31.49% | +13.39% | +50.07% |
| CAGR | +21.45% | +15.96% | +33.40% |
| SPY CAGR (same period) | +9.93% | +9.93% | +9.93% |
| Sharpe ratio | 1.28 | 1.33 | 1.50 |
| Calmar ratio | 22.06 | 16.45 | 5.38 |
| Max drawdown | -0.97% | -0.97% | -6.21% |
| Beta vs SPY | 0.042 | 0.042 | 0.222 |
| Win rate | 60.3% | 60.3% | 55.5% |
| Win / loss ratio | 1.95x | 1.95x | 1.33x |
Why daily rebalancing dominates on a risk-adjusted basis: Weekly produces a higher headline CAGR (33.4% vs 21.5%), but 87% of that advantage came from a single month. On an adjusted basis both are comparable in CAGR, while daily carries 6x less drawdown and a beta of 0.04 vs 0.22. The weekly strategy's superior headline figures reflect a momentum windfall, not structural edge.
The strategy was profitable in all six regimes encountered. The daily P&L correlation with AVGO's return was 0.002, confirming that market direction does not determine the sign of returns.
| Event | AVGO return | Strategy P&L | Classification |
|---|---|---|---|
| AVGO bull run (Oct–Dec 2024) | +27.7% | +$115 | Normal carry |
| DeepSeek crash (Jan 2025) | -6.4% | +$42 | Normal carry |
| Liberation Day (Apr 2025) | +7.6% | +$39 | Normal carry |
| AVGO earnings spike (Sep 2025) | +22.1% | +$48 | Normal carry |
| December 2025 selloff | -17.6% | +$2,102 | Vol decay event — not repeatable |
| 2026 YTD (to Mar 13) | -7.3% | +$195 | Normal carry |
The primary risk is a prolonged unidirectional move in AVGO. Between daily rebalances the position drifts out of its 2:1 hedge ratio. Synthetic stress tests show the worst case: a +1.5%/day move over 20 consecutive days produces approximately a -4% loss on capital. This risk is symmetric — sustained down moves also hurt through the same drift mechanism.
Momentum risk is symmetric. The ideal environment is choppy and mean-reverting, where both structural carry and volatility decay accrue. A trending market in either direction gradually unwinds the hedge. Daily rebalancing substantially reduces but does not eliminate this exposure.
The 0.6% borrow cost estimate reflects current market conditions and is not fixed. Borrow costs for single-stock ETFs can spike when the underlying moves strongly — potentially forcing closure at an inopportune time. This should be monitored daily through your prime broker.
AVL launched in October 2024, giving this backtest only 17 months of live data. The sample does not include a prolonged bear market or rising-rate environment. The structural tracking error argument is robust, but its specific magnitude in different regimes remains untested.
On the adjusted basis — stripping the December volatility windfall — this strategy produced a CAGR of approximately 15.96% at near-zero market beta and a maximum drawdown under 1%. That is a genuinely attractive risk-adjusted profile, and it rests on a structural argument that is observable and persistent across all 17 months of available data.
The full-sample CAGR of 21.45% is accurate but misleading as a forward-looking estimate. The author's view is that the honest expected return is in the range of 11–16% annually — the lower bound representing conservative structural carry, the upper bound allowing for occasional volatility decay events. Investors should base any allocation decision on the adjusted figures presented here, not the headline number.