Ross’s (1976) theory of Arbitrage Pricing requires news related to the factors affecting stock prices to be mean-zero, serially uncorrelated white-noise processes, though it specifies neither the number nor the identity of these factors. In this study we test whether six candidate factors namely, industrial production growth, inflation, movements of default premium, shifts in the slope of yield curve, policy-related uncertainty and developments in tech-sector could serve as pervasive forces affecting stock returns in the US for the period from 2005 to 2017. To isolate shocks associated with these factors we estimate the residuals of ARMA models, arguing an assumption of random walk processes implied by several previous surveys could lead to spurious results due to the presence of significant autocorrelation. Moreover, we implement a Fama-MacBeth (1973) variant procedure to estimate the risk premiums where the cross-sectional regressions are substituted by a panel data analysis with time-varying factor loadings assuming premiums remain time-invariant throughout the testing period. To support our empirical tests, we have constructed two algorithms; the first one arranges stocks into portfolios upon three different criteria (exposure to market risk, market capitalization and industry) while the second estimates exposures of returns to the factors with a rolling window technique. Considering annual returns and allowing exposures to vary per annum, we corroborate our findings with the three different techniques of portfolio formations and the model explains approximately one-third of stock prices fluctuation. When we reduce the time-horizon of returns to one month (increase the frequency of observations) and allow risk exposures to change per month, we find that the estimated premia are quite sensitive to both the subperiod we set and the technique behind the portfolio formation. Sign and magnitude of premiums are at least plausible.