What is momentum? Separate a price trend from a ranking

Price momentum measures how much a price changed over a specified window. A simple measure divides the ending price by the starting price and subtracts one. Calling this “strength” does not specify the window or price basis.

The number describes the past. A positive reading does not guarantee a later rise. Define the characteristic, the portfolio rule using it and the orders generated by that rule separately.

A time-series view examines one security against its own past: was its price rising or falling over the chosen interval? It does not require comparison with another security.

A cross-sectional view measures several securities over the same interval and ranks them on the same date. The least negative security can rank first. Being highly ranked does not necessarily mean having a positive return.

In this fictional teaching example, EX-A moves from 100 to 110 yen, EX-B from 200 to 210 and EX-C from 100 to 98. Their changes over the same window are +10%, +5% and −2%. These are neither real prices nor investment results.

Ranking the changes gives A, B and C. A and B both rise by 10 yen, but their percentage changes differ because their starting prices differ. A ranking would still exist if all three changes were negative.

A short-term rise can coexist with a longer-term decline. Changing the window changes the question. Choosing an attractive window after seeing the results changes the meaning of the comparison.

Check adjustments around stock splits, the handling of missing dates, whether the window counts observations or calendar days, and which prices were available at the decision time. Rankings without consistent inputs are difficult to interpret.

KabuForge’s public course describes signals based on closes through the previous observed session and simulated execution at the next observed open. This is a course implementation convention, not the definition of momentum itself. Related explanations: factors, look-ahead bias, transaction costs.

A ranking does not determine actual holdings. Cash, quantities, fees and skipped orders can change the outcome. Compare inputs and order records instead of explaining execution from a score alone.

Workflow and display claims refer to KabuForge’s public GUI research course, reviewed October 6, 2026. The fictional examples illustrate calculations and are not results from the public course.

The public examples are research-only, and historical information availability is unverified. Explaining a general concept does not establish that every part is implemented by the software. This article recommends no securities, trades or settings. Read the disclaimer.

Jegadeesh and Titman’s original 1993 paper。 This original study compares securities using past price performance. Its market and assumptions differ from this lesson and do not establish future Japanese-equity returns.

Explore the strategy concepts learning path

• What is a moving average? Check its window and availability
• Value investing: separate valuation measures from future returns
• Quality factors: fix a profitability definition before comparing firms