Two different ideas
Blackjack is a game of chance with a built-in casino advantage. No strategy, system, pattern or timing changes the odds in the player’s favor over time, and most players lose money over time. Variance does not change that. It describes how far individual results are likely to stray from the average, and it explains why a session can feel very different from what the math predicts. Nobody should gamble money they cannot afford to lose.
Two ideas are worth separating. The expected result is the long-run average loss per amount wagered. Variance is the spread of actual results around that average. In blackjack the spread is much larger than the average loss, especially over a few hours of play.
Ten hypothetical hands
All figures in this article are hypothetical. Suppose a player bets $10 on each of ten hands and the outcomes are as follows:
| Hand | Result | Change |
|---|---|---|
| 1 | Win | +$10 |
| 2 | Loss | -$10 |
| 3 | Loss | -$10 |
| 4 | Win | +$10 |
| 5 | Push | $0 |
| 6 | Loss | -$10 |
| 7 | Win | +$10 |
| 8 | Win | +$10 |
| 9 | Loss | -$10 |
| 10 | Win | +$10 |
That is five wins, four losses and one push, for a net of +$10. Another ten hands, with three wins, six losses and one push, would net -$30. Both are ordinary outcomes. Neither says anything about whether the game is fair or the player is skilled, and a session of ten hands is far too short to show the underlying edge.
Real blackjack adds extra movement. Doubling down and splitting put more than the base bet at stake on a hand, and a natural pays 3 to 2 at most tables. Both make swings bigger than in a simple even-money bet.
Measuring the spread
Standard deviation is the usual way to describe the spread. For blackjack, a figure of a little over one bet per hand is commonly used. The numbers below use 1.15 times the bet, an assumption chosen for illustration, along with a hypothetical edge of 0.5 percent and a $10 bet each hand.
For n hands, the expected loss grows in proportion to n, but the standard deviation grows only with the square root of n. That is the key to the whole subject.
| Hands | Total wagered | Expected loss | Standard deviation (about) |
|---|---|---|---|
| 100 | $1,000 | $5 | $115 |
| 1,000 | $10,000 | $50 | $364 |
| 10,000 | $100,000 | $500 | $1,150 |
| 100,000 | $1,000,000 | $5,000 | $3,637 |
The arithmetic: $10 × 1.15 = $11.50 per hand; times the square root of 100 (10) gives $115; times the square root of 1,000 (about 31.6) gives about $364; times 100 gives $1,150; times about 316 gives about $3,637.
Read the first row. After 100 hands the expected loss is $5, yet a typical result is within roughly $115 of that in either direction. For a normal-shaped spread, about two-thirds of sessions would end within one standard deviation of the average, which here is roughly from a $120 loss to a $110 win, and about 95 percent within two standard deviations. A losing session or a winning one is perfectly consistent with the same edge. Nearly as many 100-hand sessions would end ahead as behind, because the spread is so much larger than the expected loss.
What happens as the hands pile up
Look down the table. The spread keeps growing in dollars, but it shrinks relative to the expected loss. At 100 hands the standard deviation is 23 times the expected loss ($115 against $5). At 100,000 hands it is less than the expected loss ($3,637 against $5,000). That is the law of large numbers: with enough play, results cluster around the average, and the casino’s edge becomes visible. The casino plays vast numbers of hands, so for the house the outcome is predictable. For an individual player, a few hundred hands tell almost nothing.
This is also why a hot run does not show a system works. Someone who happens to finish a weekend ahead has seen ordinary variance. Over a long run, the same decisions would drift toward the expected loss.
Why it feels so different from the math
Variance interacts with how people think. Wins are vivid, which feeds selective memory, and a short winning streak can create an illusion of control over the cards. A losing streak can lead to the opposite error, the belief that a win is due, which is the gamblers fallacy. Each hand is dealt from a shuffled deck, so random streaks happen without any hidden pattern or debt owed by the cards.
Practical consequences
- The expected loss is a long-run average, not a guide for any one evening. A $5 expected loss on 100 hands does not mean the session ends $5 down.
- Larger bets raise the dollar spread in proportion. Doubling the bet doubles both the average loss and the standard deviation.
- A bigger bankroll does not change the edge, but it affects how long play can continue before the money is gone. This is the heart of bankroll management, which is about limiting losses rather than changing odds.
- Chasing losses to get back to even adds more hands at the same edge and raises the amount lost on average.
Because the swings are large, deciding in advance how much to spend, and stopping there, is the sensible approach. If gambling stops being fun or starts to cause stress, local support services are available to talk it through.