What a run of bets costs on average and how wide the spread around it is. Eight games with measured volatility, 95% result band, no signup.
A house edge of 2.7% does not mean losing 2.7% of a session. It means that if the session were repeated enough times, the average would settle there. Any single run lands somewhere in a band around that average, and for realistic session lengths the band is far wider than the average itself.
Take a thousand even-money bets of ten dollars on a single-zero wheel. The expected result is a loss of two hundred and seventy. The standard deviation of the total is three hundred and sixteen, so nineteen sessions in twenty finish somewhere between losing about nine hundred and winning about three hundred and fifty. Winning is entirely normal over a thousand bets; it is not evidence of anything.
The expected loss grows in proportion to the number of bets. The standard deviation grows with the square root of it. Quadruple the session length and the expected loss quadruples while the spread only doubles — which is why the edge eventually dominates, and why "eventually" is much further away than most players assume.
This is also why volatility matters as much as edge when you are choosing a game. A high-variance slot and a low-variance one can carry the same house edge and produce completely different experiences of it: the same expected cost, delivered in wildly different chunks.
It assumes bets of equal size, independent of each other, at a fixed edge and a fixed per-bet volatility. That covers table games cleanly. Slots are approximated — their payout distribution is heavily skewed by a rare top prize, so the symmetric band is a simplification, and real outcomes cluster below the average more often than a normal curve suggests.
It does not model bet sizing that changes with the bankroll, bonuses, or running out of money mid-session. A ruin calculation is a different question from a range calculation, and mixing them produces numbers that are wrong in both directions.
The spread of possible results around the expected one. Two games with the same house edge can have very different variance, which is what makes one feel steady and the other feel wild.
Because over any realistic session the spread is much larger than the expected loss. Winning is a normal outcome, not evidence that a game is loose or a system works.
It depends on volatility. On even-money bets the expected loss overtakes one standard deviation after a few thousand bets; on a high-variance slot it can take hundreds of thousands.
Approximately. Slot payouts are skewed by rare large prizes, so a symmetric band understates how often you finish below average and overstates the smooth middle.
No. It reports the range of results for a fixed number of bets. Risk of ruin asks a different question — the chance of hitting zero at any point along the way.