The active share between the unconstrained and constrained versions of each of the strategies we analyze drives the tracking error figures we observe. As we might expect, higher active share leads to higher tracking error. Although a simple and neat relationship does not exist, our findings show that tracking error is highly correlated with the active share induced by customization. We observe an average ratio of tracking error to active share of 8.5% with 8.56 basis points (bps) of standard error at a substantial R² of 68%. This means we would expect every 100 bps of active share generated by portfolio customization to yield a custom portfolio whose returns will deviate from those of its original portfolio by at least 8.5 bps in roughly one out of every three years and by at least 17 bps in one out of every 20 years. More-material constraints than we consider in our analysis could lead to even higher deviations.