For U.S. two-way tactical asset allocation (TAA) managers, performance is a
function of at least two factors: forecasting ability and tilt size. Asses
sing the relative performance across a universe of managers is made difficu
lt because of small peer group and relative short return histories. Simulat
ed performance universes are used to decompose the source of relative perfo
rmance and to overcome the small sample problems. Surprisingly, little fore
casting ability is required, on average, to provide positive excess returns
. Portfolio risk, measured as the probability of underperforming the benchm
ark, decreases with forecasting skill and is unaffected by tilt size. The a
uthor uses actual return and portfolio weight histories to assess the forec
asting skill of U.S. TAA managers and demonstrates that after accounting fo
r equity tilt, size, sample period, and length, seemingly equivalent: manag
ers possess vastly different levels of forecasting ability.