Three thoughts: 1) I would expect that those trading at the greatest discount to NAV would have the greatest underperformance vs their index. The payoffs to arbitrage from the creation/liquidation process are greater, which quietly comes out of the pockets of continuing holders of the ETF. 2) Of course, this all depends on whether the calculated NAV is accurate. Pricing grids do not keep up with markets under times of stress. You can only validate the pricing grid off the activity of those doing unit creation and liquidation. 3) I would expect these effects to be larger on a dollar-weighted basis vs time-weighted.