Everything is possible. Many things are unlikely. Winning the lottery is very unlikely but people do so every day.
The issue you flag has never been properly studied. In Color Stacking Fairness Project you will see @Garanwyn and others have carefully studied the potential inflence of hero selection on the starting board. They conclude that there is no evidence of any systematic skewing, though Everybody Gets Bad Boards.
If you would like to take on the project of testing whether tile replacements are skewed, that would be a tremendous service.
Suggested approach:
-
set a raid attack team that is heavily stacked in a color, e.g. 3 purples, a green and a blue.
-
Video record 100 consecutive raids using this team.
-
Review the videos, tallying each color’s count of the total tiles dropped.
-
Publish your data here. One of the stats gurus here can do the number crunching.
Your dataset will look something like this:
| Raid# | Dark | Holy | Ice | Nature | Fire |
|---|---|---|---|---|---|
| 1 | 6 | 4 | 5 | 3 | 8 |
| 2 | 4 | 11 | 9 | 21 | 7 |
| 3 | 8 | 12 | 15 | 9 | 11 |
Where the numbers in the second through sixth columns are your count of the tile drops of that color.