The thing with sample sizes is that they need to be pretty large to be worth anything. If you flip a fair coin with a 50/50 heads or tails outcome 100,000 times, you might expect 50k heads, 50k tails, but the numbers would likely differ. The actual probability of getting exactly 50k heads and 50k tails (1 in 396 chance of this happening, if you’re curious) is weighed up against the probability of all other outcomes . Probabilities will just get closer to the averages as the sample size increases. The problem with this in the case of Dirty Bomb cases is that some of them have pretty small percent chances of dropping, meaning that whilst on average you might expect to see a Cobalt drop once in 1000 cases, but to test this you need a huge sample size.
Essentially, if this were an exercise in determining an unknown probability (so, if the case percentages were hidden from us) of each rarity dropping from regular cases, your numbers would be pretty accurate (within reason). You’d have:
Lead: ~86% / Iron: ~11.5% / Bronze: ~2% / Silver: ~0.5% / Gold <1% / Cobalt <1%
If we had every player who had ever opened a case submitting this, we’d be even closer. But it would still (likely) never be the exact figures specified. Makes me partially glad they announced them openly.
Part of me was tempted to do some ridiculous math based around the probability of getting a cobalt drop for a Merc you actually want to use, with a set of usable augments. Instead I’m going to go and find some lunch. 