Quant Investing: How Volatility-Scaled improve Momentum (I) 中文
傳統的動量因子較無法捕捉短期內市場的走勢，本研究旨在驗證利用近期波動度來動態調整部位權重是否能使時間序列動量因子整體績效的有所提昇，並透過實證發現其在各情境下的績效皆有改善。
Time-series momentum is a strategy based on absolute momentum which means the trend of an asset is independent of others in the universe. In the previous article, we have found that time-series momentum has better performance than the traditional momentum strategy. But there is still a weak point for this strategy which is the time it needs to react and change the trend. For a regular momentum strategy, the look-back period will be 6-24 months to decide the trend. It can’t respond well to the change of the recent market for the target that is in deeply bull or bear trend. Thus, we can adjust the weight according to the volatility of nearly day returns based on the result of Baltas and Kosowski(2019).
This graph shows the performance of Volatility-scaled momentum (VSMOM, thereafter) and time-series momentum (TSMOM, thereafter) based on the same universe which is SPDR industry ETFs. Since the volatility has a negative correlation with the return. It will have higher leverage in the bull market, and vice versa. Figures 2 and 3 show that the weight of each ETF held on the long side in 2008 and 2020. We can see that the leverage shrinks when the market crashes.This causes VSMOM to have a smaller max drawdown than TSMOM. Also, the leverage recovers when the market reversal and becomes larger when the market back to the bull market.
Based on the dynamics adjust to the weight of each ETF, we can greatly improve the performance whether it is the bull market or finance crashes without other indicators or information.
Contact us at info@gammaparadigm.com for the full report.
Yu-Chin Lin
Quantitative Research Intern, Quantitative Finance, Gamma Paradigm
