Well Magic
is random, you can be the best player in the world with the best deck and still lose to mana screw. That's why top players only have 60% win rate (it's apparently 65% at GP day 1 level). You still go "positive over time" of course, because any win % that is higher than 50% is enough to go "positive over time". You can also argue it's a skill-based game, since in any game where the top players can consistently win more than 50% of the time is arguably skill-based, but it's still 60% win rate.
For illustration purposes here is a screenshot of the leaderboards from another game I play. The game mode is 5v5, but you can queue at most with two players. The first column on the right is the elo, the second is the number of wins and the third is the number of losses.
I don't know how many of these top-ranked players are solo queuing and how many only duo (and queue during peak hours when the opposition is on average softer), but it should be pretty clear that this game is less random than Magic. 75% win rate over a hundred matches is plausible.
The question is how random is poker? The only thing in your post that touches on it is
Quote:
Multi-table tournament players (the softest fields in poker) generally cash in one out of four tournaments (the top players anyway)
I don't know how often top Magic players cash in tournaments, but 1/4 does not seem like a high rate.
If you have a 60% winrate, then for every dollar you risk, you make $1.20.
Now, multiply that by 100k hands. 200k. 300k hands, etc.
Most good players can generate a winrate of about 8-15 bb/hr playing live, at reasonable stakes. In a 5/10 game, that works out to between $80 per hour to $150 per hour. Then you just add volume.
Winrates on a per hand basis is much smaller playing online, because the fields are tougher, but suddenly you can massively increase volume. That's why most online players try to get about 5 bb/100 hands, then just play multiple tables to try to jam 500 hands per hour, or 25 bb/hr.
Any game where variance matters, you just eliminate variance with volume. It doesn't matter if you run way below expectation for 10k hands, because it's a tiny sample. You just get more volume in and you inevitably revert to your standard winrate. At about 500 hours of playing live or maybe 50k hands online, you probably have some idea of how you're playing.