I draft, in person, on most Fridays. I know what a reasonably likely hand looks like. I know how to shuffle to avoid predictable patterns. I'm good at choosing the right number of lands, and the right colors of lands (what I choose to draft and what I manage to draw are different matters).

I just played a Quick Draft. 7 Swamps, 7 Forests, 3 Islands, totally reasonable. The ONLY game where I saw a reasonable distribution of colors was the first - where I saw 7 lands out of the first 10 cards. Otherwise it was all Forests, all Swamps, or somehow all Islands; the last match of 6, I conceded after mulliganing away 3 Forests, into 3 Swamps, mulliganed into 2 Islands. This is not a playable game, not in Best of 1 (and I'd play a 'Quick Draft Best of 3', for the record, but you'll say "the demand isn't there").

I play for real; I know how unlikely it is to see 8+ opening hands that skewed.

Given how this seems even more likely to happen after a win - and also given that I almost never actually spend money on Arena - it seems very likely to me that this is intentional, to 'balance' win rates and especially for players that don't pony up (I spend plenty of money on real cards, and your system knows this, it knows when I draft and play prereleases).

But even to give the benefit of the doubt, that this is unintentional, it still ruins games. If the intention is to make Best of 1 a bit less swingy and dependent on play-draw, well, it's failing miserably at that. I would perhaps suggest, if it is merely looking at two hands and picking the 'best' one, it's criteria for 'best' is wrong - particularly that 'more' lands is not necessarily better than a spread of colors for decks that are primarily using basic lands (then again, I regularly play against Standard decks using seemingly only basics and it's relatively rare that they are screwed on colors, so I still don't think this is unintentional).