The paper explains an array of RBC puzzles by adding to the standard RBC model external margins for both physical capital and human capital, and examining model fit with US data across business cycle (BC), low frequency (LF), and Medium Cycle (MC) windows. The model results in a goods sector productivity shock with a 7500 times smaller variance than the standard RBC model, implying greatly improved shock amplification and an enhanced explanation of a wide array of correlations, volatilities and growth persistence across the windows. The model matches the data cyclicality of the main shares of GDP and GDI such as a countercyclic consumption-output ratio, procyclic investment-output ratio, countercyclic labor share of income and countercyclic capital depreciation share of income. Also matched is a countercyclic human capital investment time, a procyclic capacity utilization rate, and the declining output growth persistence autocorrelation profile that is known as the "propagation" puzzle. Using a distance metric and a uniform grid search, measures of fit are presented by window and category. In the BC window, key correlations have an average 15% deviation from the data moments; the LF growth persistence has an average 8% deviation from the data moments.