ETF Rotation Strategies: Full Breakdown & Transcript

基于调整动量的A股ETF动量轮动策略复现和研究

0h 06m video Published Jun 14, 2026 Transcribed Sep 17, 2026 小木吱吱

74 views Recent velocity 0.0 views/hour View full performance history →

Intermediate 4 min read For: Quantitative finance enthusiasts and individual investors interested in ETF rotation strategies.

AI Trust Score 45/100

🚫 Clickbait / Waste of Time

"The title promises a strategy replication, but the video reveals significant flaws and limitations, making it more of a critique than a straightforward how-to."

AI Summary

This video replicates and tests three ETF rotation strategies from a Chinese public account called '小西湖的量化之路' (Xiao Xi Hu's Quant Road). The strategies use adjusted momentum, and the video analyzes their performance, revealing that while the backtests show high returns, the actual results are sensitive to start dates and candidate selection, with some strategies underperforming simple equal-weight buy-and-hold benchmarks.

Chapters

1 Introduction and Strategy Overview 00:00 2 Adjusted Momentum Calculation 00:51 3 Rebalancing Process and Backtest Setup 01:49 4 Backtest Results and Issues 02:25 5 Performance Comparison and Candidate Selection 03:47 6 Cross-Market Strategy and Source of Returns 04:49 7 Limitations and Conclusion 05:42

[00:00]

Introduction to Strategy Replication

The video replicates three ETF rotation strategies from a Chinese quant blog, converting R code to Python and standardizing execution. The strategies use momentum, which is a well-documented anomaly in finance.

[00:26]

Momentum as a Strategy

Momentum (追涨) is a proven anomaly since 1993, working across stocks, bonds, commodities, and forex. In A-shares, sector rotation is strong, so momentum-based rotation can work, but it lags in fast-moving markets.

[01:16]

Adjusted Momentum Calculation

The strategy uses adjusted momentum: calculate the average return over the last 20 days (base momentum), divide by the volatility of daily returns over the same period. This identifies stocks rising steadily, not just rising.

[01:49]

Rebalancing Process

Five steps: 1) Calculate adjusted momentum, keep only positive values. 2) Sort by momentum. 3) Select top N (or all if fewer). 4) Weight positions by adjusted momentum (not equal weight). 5) If all momentum values are negative on rebalance day, hold cash.

[02:25]

Backtest Results and Issues

Three tests with different candidate pools and rebalancing frequencies. The original backtest from 2024 shows good returns, but replication reveals issues: sensitivity to start date (e.g., 35.8% vs 23.5% annualized), and underperformance vs equal-weight buy-and-hold in some cases.

[03:47]

Performance Comparison

For A-share leader stocks, original claimed 110% annualized, but replication shows 83.7%. However, equal-weight buy-and-hold of the same 22 stocks yielded 81.3%, meaning rotation added only ~2% alpha, and the equal-weight had higher drawdown.

[04:25]

Key Insight: Candidate Selection Matters

The strategy's success depends heavily on the candidate pool, which is subjectively chosen. The original author selected stocks like 年贷时代韩武记, which may have survivorship bias.

[04:49]

Cross-Market Strategy

A three-ETF strategy (纳指, 呼声300, 红金强速) with low correlation (~0.2). Replication shows annualized returns of 47.4% (original) vs 47.7% (replicated), with lower volatility (44.4% annualized, 2.33 Sharpe, 14.3% max drawdown).

[05:17]

Source of Returns

Gold ETF was held 70% of the time, with average weight 57%, and gold rose over 100% in two years. The strategy essentially auto-buys gold, which worked well but is not a diversified edge.

[05:42]

Strategy Limitations

The rotation strategy did not beat any of the three ETFs' individual returns. Also, it underperforms in bear markets (e.g., negative 4.5% annualized from 2021-2023).

The momentum rotation strategies show strong returns in backtests, but they are highly sensitive to start dates and candidate selection. The edge comes more from the chosen assets (e.g., gold) than from the rotation logic itself, and the strategy fails in trending-down markets.

Mentioned in this Video

小西湖的量化之路 (public account)

link

Tutorial Checklist

1 01:16 Calculate adjusted momentum: average 20-day return divided by 20-day daily return volatility.

2 01:49 Keep only assets with positive adjusted momentum.

3 01:49 Sort assets by adjusted momentum in descending order.

4 01:49 Select top N assets (or all if fewer than N).

5 01:49 Weight positions proportionally to adjusted momentum (not equal weight).

6 01:49 If all momentum values are negative on rebalance day, hold cash.

💡 Key Takeaways

📊

Momentum is a proven anomaly

Establishes the academic basis for the strategy, making it credible.