Given the cost/kill, your actual results, maximum hof possible on them these days and frequency of various loot sizes, I would say right now is an extreme scenario, somehow similar to 2007 (i am not playing atm, just eyeballing the situation). I mean, regarding to max possible loots. Since I don't see the same "omg I got 50% return" nor do I see various EU intelligentia musing on how 90% is optimal, then it's safe to presume MA changed their approach. Meaning they spread the pie with keeping for candy (high loots) a very small slice and rest putting in smaller more often blings, such as results to be somehow bearable. However they very existence of those high loots will imply with necessity more stress in the 90% area.
I would say that, nomatter your efficiency and looter skill, because they will have a minimal impact in this regard as long as you're in some sane point, you can look at a very long interval, in the order of 100k if not more.
Point 1. As such, I would frankly ignore the tt results, as weird as it might sound, and focus on MU. In a hypotethical reasonable 95% (let's not get too optymistic with 98%), would looted MU be +-5%? Is that simple.
Point 2., Disregarding high loots, I would say your sample size is pretty solid for what can be expected in terms of median loot. So then comes the next, way way more important question. Do you have enough bankroll to carry 4-5-6 intervals similar to this one and still be sane and still have SOME reserve worst case scenario?
Worst case would be that neither looted MU checks that 5% (nomatter if you sell them or use them), neither bankroll can carry some projected 100-150k of similar return.
In summary, I would look again at the data over 100k, but only and only if it has any sense MUwise, otherwise just search for luck anywhere, you won't "let peds lost in the mob", that's nonsense, don't let that hinder your felxibility.