There is a standard way most players decide how much money to bring to a game. They think about what they can afford to lose, subtract a small buffer for comfort, and call that number their bankroll. It feels like responsible planning. It is, mathematically speaking, almost entirely backwards.
A bankroll requirement is not a personal finance question. It is a statistical one. The number you need is derived from the specific game you are playing—its volatility, its edge, and your target number of decisions—not from your account balance or your tolerance for discomfort. When those two methods produce the same number, it is a coincidence. Usually they don't, and the gap between them has real consequences.
What Bankroll Requirements Actually Measure
A properly calculated bankroll requirement answers one question: how much capital do you need to make it statistically unlikely that variance destroys your session before your edge—or even your intended number of hands—has a chance to express itself?
That question has three inputs: the standard deviation of the game per unit wagered, the number of decisions you plan to make, and the risk of ruin you are willing to accept. None of those inputs are your personal feelings about money.
Standard deviation per hand varies significantly by game. Baccarat banker bets sit around 0.93 units per hand. Basic strategy blackjack runs roughly 1.15 units. A single-number roulette bet carries a standard deviation near 5.8 units. These are not the same game from a bankroll perspective, even if the table minimums are identical.
Once you have the standard deviation and your planned number of decisions, you can calculate the total session standard deviation: multiply the per-hand standard deviation by the square root of the number of hands. For 200 hands of blackjack at 1.15 units per hand, that is approximately 16.3 units of session-level volatility. If you are playing $25 units, that is $407 of standard deviation in a 200-hand session.
That number is one standard deviation. Roughly 16% of sessions will go worse than one standard deviation below expected value. About 2.3% will go worse than two standard deviations below. If you cannot survive two standard deviations of downside—which for 200 hands of $25 blackjack is roughly $814 below expectation—then your bankroll is not calibrated to the game.
The Risk-of-Ruin Connection
This connects directly to risk of ruin, which is where the calculation becomes concrete. Risk of ruin is the probability that you exhaust your bankroll before completing your intended session or reaching your target. There are simplified formulas for estimating it, and while they involve approximations, they produce useful numbers that comfort-based bankrolling cannot.
A rough estimate of risk of ruin for a negative-expectation player in a session context can be expressed as a function of how many standard deviations of downside your bankroll covers. If your bankroll covers one session standard deviation, your risk of ruin for that session is meaningful—over 15%. If it covers two, that drops substantially. Three covers you against nearly everything except the statistical tail.
Most players, when they calculate backward from what they're willing to lose, end up somewhere between one and two standard deviations of coverage—enough to feel safe, not enough to actually be safe by any rigorous definition. They then attribute their busted sessions to bad luck rather than to undercapitalization.
The Game-Specific Gap
The practical problem is that different games require radically different bankrolls for the same bet size and the same number of decisions. Consider a player who brings $300 to a session planning to make 150 decisions at $10 per hand.
In baccarat on the banker bet, the session standard deviation is about $0.93 × $10 × √150 ≈ $114. That $300 covers roughly 2.6 standard deviations of downside. Reasonably well capitalized.
The same player at a Caribbean Stud table, where per-hand standard deviation can exceed 2.2 units once the ante-raise structure is included, faces a session standard deviation closer than $270. That $300 now covers barely one standard deviation. The table minimum is the same. The risk profile is completely different.
This is why copying a bankroll number from one game to another is a category error. The number does not transfer because the volatility does not transfer.
Building the Calculation the Right Direction
The correct process runs like this:
Step one: Identify the per-hand standard deviation for the specific game and bet type you intend to play. This information is available in game analysis literature for every major casino game.
Step two: Decide how many decisions you want to be able to play through. This is a legitimate personal choice—session length is yours to set.
Step three: Calculate session standard deviation. Multiply the per-hand standard deviation (in dollar terms at your intended bet size) by the square root of your planned decisions.
Step four: Choose your acceptable risk of ruin for the session. A common target is 5% or lower. Covering approximately 1.65 standard deviations of downside gives you 95% session survival probability in a symmetric approximation. Two standard deviations covers roughly 97.7%.
Step five: Multiply your session standard deviation by your chosen coverage multiple. That is your bankroll floor.
If that number is more than you brought, you have two options: bring more, or reduce your bet size until the required bankroll matches what you have. There is no third option that doesn't involve accepting higher ruin risk than you calculated.
What This Changes in Practice
For most players, running this calculation does one of two things: it confirms that their bankroll is roughly adequate for low-volatility games at small bet sizes, or it reveals a significant undercapitalization problem in higher-volatility games.
The more important behavioral change is what happens to bet sizing. When you calculate the bankroll requirement properly, the bet size becomes an output of the system rather than an input. You don't decide you want to play $25 hands and then figure out bankroll—you decide how much money you have, what coverage ratio you require, and what volatility the game carries, and the bet size falls out of that.
Players who size bets first and check bankrolls second are running the math in the wrong order. The casino doesn't care which order you run it. The variance does its work either way.