Timing Momentum Factor Using Risk Metrics: Beyond Realized Volatility

For context, this analysis builds upon our previous blog post: Timing the Momentum Factor Using Its Own Volatility. Be sure to check it out for foundational insights.

This blog post continues from our previous discussion "Timing Momentum Factor Using Its Own Realized Volatility," where we examined how realized volatility affects momentum returns over time. Here, we expand our analysis by including additional risk metrics to evaluate which best captures the risk states relevant for momentum investing.

Risk Metrics Considered

We analyze six risk metrics calculated on daily momentum factor returns, using a rolling 252-day window (approximately one trading year):

• Realized Volatility: The classic measure of return variability, annualized from daily standard deviation.
• Downside Volatility: Focuses on downside risk by calculating the standard deviation of negative returns only.
• Value-at-Risk (VaR 95%): The historical 5th percentile (worst 5%) return, reversed in sign to represent risk.
• Maximum Drawdown: The largest peak-to-trough loss over the rolling window.
• Skewness: Measures asymmetry in the return distribution, indicating whether returns tend to be more positively or negatively skewed.
• Kurtosis: Measures the "fat tails" or extreme events in return distribution.

Quantile Analysis and Average Momentum Returns

For each risk metric, we divided the data into quintiles from lowest risk (Q1) to highest risk (Q5) and calculated the average momentum return within each quintile.

The table below summarizes the average momentum returns (%) for the lowest (Q1) and highest (Q5) risk quintiles, along with the spread between them:

Risk Metric	Q1 Avg Momentum (%)	Q5 Avg Momentum (%)	Spread (Q1 - Q5)
Realized Volatility	0.059	-0.024	0.083
Downside Volatility	0.067	-0.026	0.093
VaR 95%	0.068	-0.030	0.098
Maximum Drawdown	0.042	-0.012	0.054
Skewness	0.062	-0.027	0.089
Kurtosis	0.031	-0.001	0.031

Understanding Q1, Q5, and the Spread

Q1 represents the lowest risk quintile — periods when the selected risk metric signals a relatively calm or less risky environment.

Q5 represents the highest risk quintile — periods when the risk metric indicates elevated risk or stress.

The spread between Q1 and Q5 is the difference in average momentum returns between these two extremes. A larger spread means that the risk metric better distinguishes periods of low risk (higher momentum returns) from periods of high risk (lower or negative momentum returns).

In other words, the bigger the spread, the more useful the metric is in timing the momentum factor: it helps identify when momentum is likely to perform well (Q1) versus poorly (Q5).

Interpretation and Conclusion

The spread between the lowest and highest risk quintiles gives us insight into how well each metric discriminates momentum returns based on risk levels.

• Value-at-Risk (VaR 95%) shows the largest spread (0.098), indicating it best captures the risk environment influencing momentum returns.
• Downside Volatility and Skewness also display strong differentiation, closely following VaR 95%.
• Realized Volatility, while a classic risk measure, performs slightly less strongly than VaR 95% and downside volatility in this context.
• Maximum Drawdown and Kurtosis show smaller spreads, suggesting they are less sensitive to the momentum factor’s risk-return profile.

Overall, this analysis suggests that while realized volatility remains an important measure, downside risk and tail risk metrics like VaR 95% and skewness provide more nuanced insights into the risk states that affect momentum returns.

Understanding these relationships better can help investors refine momentum timing strategies by focusing on the risk metrics that truly matter for performance.

If you haven’t already, check out our previous post Timing the Momentum Factor Using Its Own Volatility for background and methodology.