The XIV is (was) meant to track the daily inverse of (roughly) the same underlying index as VXX. So a 10% up day for the index should be 10% up for VXX and 10% down for XIV. Now we can plug in rough approximations for Monday 2/5 and Tuesday 2/6.
The underlying index was up something like 90%. So VXX up 90% and XIV down 90%. Next day the index goes down 25% so VXX down 25% XIV up 25%. The two day returns for VXX will be 1.90 * .75 = 42% up and the two day returns for XIV will be 0.10 * 1.25 = 87.5% down. You can see how the daily tracking blows out the tracking over longer terms (just 2 days in this case). Since the VIX moves had been relatively small on a day to day basis, it sort of looked like they tracked each other on inverse terms over longer periods but it was just an illusion.
The underlying index was up something like 90%. So VXX up 90% and XIV down 90%. Next day the index goes down 25% so VXX down 25% XIV up 25%. The two day returns for VXX will be 1.90 * .75 = 42% up and the two day returns for XIV will be 0.10 * 1.25 = 87.5% down. You can see how the daily tracking blows out the tracking over longer terms (just 2 days in this case). Since the VIX moves had been relatively small on a day to day basis, it sort of looked like they tracked each other on inverse terms over longer periods but it was just an illusion.