I made some sample simulation once
as you can see it’s normal to have lagging items if probability is the same for every item.
Look how much time it took (my C++ program ) to get last 10% items -
3 times more time than it took to get the first 90%.
In other words - the last items are a nightmare to find and it’s normal.
I was comparing it to a hypothetical scenario of having unique drops if playing the same session all the time (which is not true in the end from what I remember and from other tests (?) ). We made a test with blueprints in 1 session and it became harder and harder to get new ones from chests spawned with command line in modded game. But it’s not 100% random as I remember hearing about some small pity system GD has that rerolls 10 times of something like that if you get the same blueprint.
People don’t have good statistical intuitions. If you keep playing, with every new legendary you get, the next new one is going to be harder to find. The number of items you’ve found is far too low for everything to be spread out equally and for big number laws to start working. For that we should combine results of 100 players for example.