I did a few calcs for downswings in 1000 seat MTTs a while back and produced some graphs for the likelyhood of swings with various ROI's.
http://www.deucescracked.com/blogs/b...ability-Graphs
If you play with a 10% roi it is truely brutal - as you can see you should expect an 800+BI swing in 1 in 10 sets of 10k games you play, so 800BIs is not really all that unusual. (Note that you will typically have earned 10k * 0.1 = 1000 BI's for any block of 10k games.)
High roi is what protects most from downswings.
The variance of hypers/turbos/normal are all the same per game, it is dependent on the payout structure, and the finish distribution, and these are usually about the same.
You can play a lot more games per hour with hypers/turbos and as the variance is additive if you play 5x the number of games you get 5x the variance for the same hourly period. Juice has some very small affect and also the roi has little effect on the variance per game.
Perhaps an interesting thing is if you play:
(a) 5k games with a 20% roi at normal speed
(b) 10k games with a 10% roi at turbo speed. (So play twice as many as (a) in the same time.)
For both (a) and (b) you win on average the same amount.
For a) the swing to be expected 50% of the time is 300+BI
For b) the swing to be expected 50% of the time is 500+BI
So 500BI swing is about the equivalent swing faced by (b) compared to the 300BI swing for (a)
At 20% roi it takes 300 * (1.0/0.2) = 1500 games to earn the equivalent financial amount as the swing size for (a).
At 10% roi it takes 500 * (1.0/0.1) = 5000 games to earn the equivalent financial amount as the swing size for (b).
If it take A hours for (a) to play 1 game so it takes 1500 x A hours to earn this (a) swing amount on average
It takes (b) 5000 x A/2 = 2500A hours to earn this (b) swing amount.
In hours it takes 2500/1500 = 5/3 times as long to earn the amount of BIs as the average swing size you would expect playing if you halve your winrate but double your amount of games per hour.
If you had the choice of twice the games but half the roi you will typically face deeper and longer swings. Twice the games and half the win-rate makes everything a bit more of a gamble in the short term, the 'bell curve' will be centred on the same $ amount but will by about √2, or approx 1.4 times wider (the bell curve will look approx +/- 3 sd's wide for both but for (b) the sd is √2 times bigger than a's sd).
Nothing wrong in playing turbos if it increases your hourly it is great to do.
If it increases the number of games per hour but doesn't improve your win-rate per hour very much it is pulling in more of a short term gamble so just be aware of this effect.
It is nearly impossible to get an accurate value for your roi anyway, you really do need 100k games to get perhaps a 90%CI to within a +/- 5% value and who manages that in a short period.