To make $25 per hour playing on-line MTTs is pretty hard these days imo.
To make $25 per hour at a 10% roi you would have to wager on average $250 per hour. In reality you do not get a perfect choice of how to wager this amount as expensive tournaments are few to start and far between.
Warning: the following assumes a very simple model of just a 10% roi. Don't read on if you are of a nervous disposition.
For a very simple model lets say you always played 180 seat ones and these take about 5 hours to actually bink/win one but you usually last on average perhaps 2 hours. Sometimes an individual one will only last minutes, sometimes 5 hours but on average 2 hours regardless of the amount of concurrent tables you are playing or the buy-in amount.
So one way to earn $25 per hour would be to play one table of $512, this way you as you last about 2 hrs you wager $256 on average per hour. (I picked 512 as it can be halved many times) and earn 10% roi = ~$25 per hour.
Lets also say that you could play 2 concurrent tables of $256 ones and again still have 10% in each (easier field, but twice as many decisions balances out the roi. In this model each still lasts on average 2 hours and in each you have 10% roi). So now you are playing on average 2 in every two hours, you again wager $256 per hour and get a return of 10%.
Code:
Tables Cost GamesPlayedPerHr RoiPerGame
1 512 0.5 10
2 256 1 10
4 128 2 10
8 64 4 10
16 32 8 10
If you played these for 8 hours a day on average (this is a simple model, but for eg, balanced over several days you typically played 4 x $512 per day or 8 x $256 per day etc)
If you played 200 days in a year so that is like having 2 days off each week and a few bank holidays, 3 weeks of hols and perhaps 30 days combined for study and "the variance is killing me, I need to stop!" days. You would play 200 x 4 = 800 games of $512 or 1600 games of $256 etc.
The variance per game measured in BI's is the same for all games, assuming the same finish distribution in each, hence giving the same 10% roi.
The $variance per game is proportional to the BI squared, so the variance in $ is 4 times as big for one $512 game as for one $256
The actual variance per game in $ will be approx
6851149 for the $512
1712787 for the $256
but we play 2 x the amount of games for $256 as for $512 and this doubles the variance seen per hour or year.
At the end of a 200 working days year you would have a variance of:
$512 ==> 6851149 * 8 * 0.5 * 200 = 5480919200 so sd = sqrt(5480919200), sd = $74033.23
$256 ==> 1712787 * 8 * 1 * 200 = 2740459200 so sd = sqrt(2740459200), sd =$52349.39
$128 ==> 428196 * 8 * 2 * 200 = 1370227200 so sd = sqrt(1370227200), sd = $37016.58
$64 ==> 107049 * 8 * 4 * 200 = 685113600 so sd = sqrt(685113600), sd = $26174.67
$32 ==> 26762 * 8 * 8 * 200 = 342553600 so sd = sqrt(342553600), sd = $18508.20
The amount you earned on average is exactly the same for each of the buy-ins ($25 per hour) = 8 * 25 * 200 = $40000 per year.
It is pretty bad/great for the $512 buy-in player though, playing a total of 800 games, as 70% of the time in any year they earn somewhere within the range $40000 +/- $74000. About 15% of years they would lose more than $40000 - $74000 = -$34000 and about 15% earn more than $114000. Phew it's a tough way to earn a living playing a few $512 tournies per year at 10% roi.
Even the $32 player, 16 tabling, earns quite a wide (70% confidence interval) $40000 +/- $18508 from playing a total of 200 * 8 * 8 = 12800 games.
I suppose this is a very simple model. Any earned rake-back softens the bumps a little and generally I would hope that a full time professional would be able to muster more than 10% for MTTs although I do think multi-tabling 16 at 10% would be pretty hard.