Low volatility: measure historical variation without promising safety

Here low volatility means a smaller standard deviation of daily returns over the same window. The universe and window determine what “low” means.

Small variation does not prevent losses: steady declines can have little dispersion. Total volatility also differs from beta, which measures sensitivity to a market.

EX-A has daily returns −1%, +1%, −1%, +1%; EX-B has −2%, +2%, −2%, +2%. Both means are zero. Dividing squared deviations by all four observations gives standard deviations of 1% and 2%.

Using N−1 instead gives about 1.155% and 2.309%. The ranking stays the same but the values change. These are fictional observations.

Fix frequency, return definition, dividend treatment and window length before inspecting performance. One shock can dominate a short window; longer windows may combine different conditions.

Stale prices and zero-filled missing returns can make variation look artificially low. Report actual observations and corporate-action adjustments.

Selecting securities with small individual variation differs from optimizing portfolio variance using correlations. Low-volatility holdings may still concentrate in one sector.

Low-beta research also has different definitions and portfolio construction. Historical variation is not a bound on future losses.

KabuForge’s public course diagnoses momentum features. Define a low-volatility characteristic and confirm the input route separately; the course does not establish a built-in low-volatility strategy or minimum-variance optimizer.

Read factors, concentration and maximum drawdown to assess more than dispersion.

Concept references: NumPy standard deviation and ddof documentation, original low-beta research. Software workflow and limitations: KabuForge public GUI research course. Reviewed October 6, 2026. The fictional examples are teaching calculations, not course execution results.

The public course is research-only and does not certify historical information availability. A concept explanation does not establish that every method is implemented in the software. This article recommends no securities, trades or settings. Read the disclaimer.

Explore the strategy concepts learning path

• Multifactor research: check scale and overlap before combining scores
• Test Japanese Equity Momentum with a Public Sample
• What is momentum? Separate a price trend from a ranking