|
| |
| Chi-Sheng Lo |
| |
| ''Can NASDAQ-100 derivatives ETF portfolio beat QQQ?'' |
| ( 2025, Vol. 45 No.4 ) |
| |
| |
| Portfolio optimization in derivative ETF markets presents complex challenges in balancing competing objectives across instruments with fundamentally different risk-return profiles. This paper constructs a portfolio strategy to optimize NASDAQ-100 derivative ETF allocations by balancing tracking error minimization relative to the Invesco QQQ Trust (QQQ) with excess return maximization. The approach dynamically allocates investments across three specialized ETFs: a short-position fund (YQQQ), an income-focused covered-call fund (QYLD), and a triple-leveraged fund (TQQQ). Using a deep reinforcement learning (DRL) framework, the strategy applies anomaly detection to optimize rebalancing timing, incorporating dividend payments to enhance returns. The approach achieves positive excess returns across all evaluation periods, though risk-adjusted performance progressively deteriorates from substantial outperformance during training to underperformance during testing. This progression reveals both the potential and limitations of reinforcement learning approaches for multi-objective portfolio optimization when encountering evolving market conditions. |
| |
| |
| Keywords: Deep reinforcement learning, enhanced index tracking, QQQ, Nasdaq 100, derivatives, ETF |
JEL: G1 - General Financial Markets
C6 - Mathematical Methods and Programming: General |
| |
| Manuscript Received : Oct 17 2025 | | Manuscript Accepted : Dec 30 2025 |
|
