Your analysis here assumes that the player will match tiles only for the current move and does not plan ahead. The major part that you are not accounting for is what matches the player CHOOSES to play. When you color stack, especially if you run a monochrome team, you are selectively looking to make matches of your strong color to hit the opponent with, or ghost after a hole has opened; even to the point where you are playing the board to set up future matches of your strong color (or diamonds to clear one of the missing colors). By selectively playing the tiles, you skew the damage per tile higher than it should be from random matching. Additionally, if you are only cascading the missing colors, you will give your opponent less mana per tile:
If you match tiles likes the AI, I think your assessment is spot on (and it is nice to see it worked out mathematically). I’ve run color stacked teams on the auto while playing map stages before, and it is disheartening to watch. That why I only let it run on rainbow teams 
.
Color stacking has worked very well for me, and using some strategies I’ve seen @Anchor use (via streams) has yielded positive results and made my raids more enjoyable 