Graveyard / Flows / ETFs
Leveraged ETF rebalancing trades
Can you front-run the end-of-day buying and selling of leveraged ETFs, or short their decay?
The idea
A leveraged ETF has to buy after the market rises and sell after it falls, near the close, so the underlying may overshoot late in the day and revert the next morning. A second idea is to short a leveraged pair and earn the decay.
How we tested it
| Data | Test 1: leveraged ETF rebalancing reversal on 19 stocks (daily) and time-series momentum on 18 ETFs (daily). Test 2: single-stock leveraged ETF end-of-day flow for MSTR, COIN, TSLA and NVDA, on days with a move above 2% (2,115 days), entry at 15:50, follow and fade, both pre-registered. Test 3: short TQQQ and SQQQ in equal dollar size. |
|---|---|
| Period | Test 1: 2018 to 2026 for stocks, 2016 to 2026 for ETFs. Test 2: 2023 to 2026, development 2023-2024, holdout 2025-2026. Test 3: 2010 to 2026. |
| Costs | Not stated. |
| Holdout | Test 2 holdout 2025-2026, with a random twin. |
| Control | A random twin for test 2. |
Result
| Test 1: rebalancing fade | about 0 bp, no change around 2022 launches |
|---|---|
| Test 1: time-series momentum Sharpe | 0.35 or less |
| Test 2: development follow, to next open | +36 bp (twin p below 0.001) |
| Test 2: holdout follow, gross | −5.4 bp (p 0.51) |
| Test 2: holdout fade, net | +1.4 bp (p 0.50) |
| Test 2: 15:50 to 16:00 move | −1.1 bp |
| Test 3: short TQQQ and SQQQ | CAGR 1.5%: rebalancing pays back the decay |
Why it failed
The effect in the development years did not carry into the holdout. A rebalanced short pair pays back the decay it collects. Test 2 could not apply an assets-under-management filter because the history of fund size was not available.
What would make it worth retesting
A market where leveraged ETF assets are much larger than the daily volume of the underlying stock, or real fund-size history showing a flow of more than 20% of the close and a holdout-surviving sign.
Past results of a backtest say nothing certain about the future.