Momentum, one of the strongest and most comprehensively studied trading strategies, bets on the persistence of past trend. The trend is typically measured by the past cumulative returns of individual stocks and therefore the strategy is implemented by buying stocks with highest past returns and selling stocks with lowest past returns. However, the past cumulative return can be a noisy proxy for the past trend, especially for those stocks with high volatility. We apply a test to detect the presence of drift in the stock returns in the formation period eliminating the impact of volatility. We form our pure momentum portfolios based on the strength of the drift, which we argue is a more precise measure of trend. We find that our pure momentum strategies deliver significantly higher Sharpe ratio, higher Sortino ratio, lower volatility, less negative skewness, and lower kurtosis. Moreover, the strategy return is not fully spanned by the Fama-French five-factor model plus their momentum factor with abnormal returns significantly different from zero.