lots of answers
Depends on the game, the rules, the penetration, your skill and then lady luck.
I'm in the middle of a recently-started experiment. Since sims only go so far in telling you what to expect, since sims expect perfect play and betting and counting, which is unlikely over the long-haul for a human to do, I have done this test twice so far, and am keeping records as I repeat it several times over the coming year I hope.
I started with a bankroll of exactly $1000, after a discussion with my son. I decided to play DD only, spreading $5-$40, heads up, or $5 to 2 x $25 if there was another player at the table.
First experiment was a little skewed because I wanted to answer the question "can you double your bankroll playing as above?"
I started at $1000, reached a peak of $2265, but played another two sessions after that and ended up with some wild variance, dropping all the way back to just under $1000 before ending up at 1650 or so after about 1400-1500 hands played.
I again started at $1000, and hit $2200 (where I stopped) after about 500-600 hands.
So two passes so far. I am probably going to toss the first experiment since I should have stopped when I doubled, and started over. Had I done so, I might well have gone bankrupt on my second session since that tail-end variance on the first test was larger than my $1000 trip bankroll for the experiment.
What does this mean? That there are fluctuations. In this second trial, the variance was pretty smooth. Never lost more than $200 before climbing back and going higher...
I'm going to repeat this a few times as I'm now really interested in what _I_ can do as opposed to what a sim does. I can play BS perfectly. I can count using the CV blackjack drills with almost perfect accuracy, and over 2 deck drills I almost never make a counting error. But I do on occasion mis-calculate the true count by missing the remaining deck estimate, which can lead to a betting error or a BS departure index error. I don't make many of these, and have played hour+ sessions on CV BJ with no betting or playing errors, but I also play the same sessions and make an occasional error here and there. I don't have the option of introducing errors (that I can find) in CVCX running sims, so I'm running a "real sim" by doing it myself.
I could do this by just playing a 1-hour session on CVBJ every night, but even though the BJ program is _very_ good, it is not a casino, those are not real dollars, etc...
The moral of this is that you can win, but there are definitely losing stretches thrown in. If they come after a winning session, they are not as painful as when they occur at the start and drain your BR to zero if it isn't big enough. No, I don't normally gamble spreading $5-40, with a BR of $1000, the ROR is too high. But I have seen several talk about playing like this so it seemed like something interesting to try and appealed to the "scientist" in me.
More on this as the experiment progresses. If I had run the first test correctly, laying all the rounds played so far end-to-end would suggest that by starting at $1000, three times I have doubled the BR, once I went bust. My real BR has grown by over $2000 so far, and I am fixing to start the next round by dropping back to $1000 and starting again...