Abstract A momentum trading approach is presented to examine the Dow Jones industrial components for a period of about past 10 years (1992–2002). An analogy between the classical dynamics in physics and the stock trade dynamics is used with the momentum, P=mv, where the velocity (v) is a relative price change in a period (τ) and the inertial mass (m) is a normalized trade volume. Extrema in the momentum time series, i.e., the singularities in the driving force provide the signals for executing trades, minima with negative momentum to buy and maxima with positive momentum to sell. Trades are implemented using a momentum threshold (Pc). A range of periodic cycles (τ=5–240 days) in time series and trading momentum thresholds (|Pc|=0.01–0.5) are considered and returns (maximum, minimum, accumulative, and average) are examined in detail on the historical DJI data for about a decade (1992–2002). Frequency of trade is generally higher with smaller periods with the high probability of higher returns at |Pc|=0.02–0.1 for nearly all stocks in DJI.