Variance vs. House Edge: How You Win on Bad Odds and Lose on Good Ones

What exactly is house edge?

House edge is the average percentage of every wager the house expects to retain, calculated across an enormous number of bets. It is built into the payout structure and rules of a game, not into any single outcome, so it only describes what happens in aggregate, never what happens on one spin or one hand.

Every bet in the salon has a built-in mathematical cost, and house edge is simply that cost expressed as a percentage of what is wagered. A 0.5% edge means that for every 100 units bet over a large enough sample, the house expects to keep about half a unit, or about 0.5 ENT per 100 staked. A 3% to 8% edge, typical of slot reels, means the house expects to keep roughly 3 to 8 ENT per 100 staked across the same volume.

The critical word is expects. The edge is a statement about the average of a great many trials, not a prediction for the next one. A single spin does not know the house edge and does not owe it anything. The edge only becomes visible once enough bets have accumulated for the average to stabilize.

House edge describes the average of many bets, never the outcome of one.

What is variance, and how is it measured?

Variance, also called volatility, measures how widely individual results scatter around the expected average. Its mathematical measure is standard deviation. High-variance games swing hard in both directions over short stretches; low-variance games cluster tightly around the expected result, session after session.

Two games can share a similar edge and still feel completely different to play, because variance governs the shape of the ride even when it does not govern the destination. A high standard deviation means big, rare payouts and long stretches of nothing in between. A low standard deviation means smaller, steadier movement around the average, with fewer dramatic swings either way.

• High variance: slots, keno, and long-shot single-number roulette bets
• Low variance: blackjack, baccarat, and even-money wagers

Why can a bad-odds game feel like a winner and a fair game feel like a loser, in the same night?

Because in a short session, variance dominates and the edge has not accumulated enough bets to show itself. A high-variance game can hand a guest a large early win that has nothing to do with its edge, while a low-variance game can grind out a small, unlucky loss that has nothing to do with how fair it actually is.

This is the mechanism behind a very familiar story: a guest wins big and fast at the slots, then sits down at blackjack and loses steadily, and walks away concluding the slots were the better game. The conclusion mistakes a lucky sample of variance for a judgment about edge. Blackjack, played with basic strategy, carries an edge of roughly 0.5%, while slot edges commonly run 3% to 8%, several times worse. The single night said nothing true about either number; it only reflected how each game's variance happened to land that evening.

The anecdote is not wrong about what happened. It is wrong about what it proves. One session is a sample size of one, and neither the law of large numbers nor the house edge has any obligation to show up in a sample that small.

One winning night at slots proves nothing about which game is fairer.

What is the law of large numbers, and why does the house always get there first?

The law of large numbers states that as the number of bets grows, the average result converges toward the expected value, meaning the house edge. The house reaches that convergence quickly because it books an enormous, continuous volume of bets; an individual guest, playing far fewer hands across far fewer sessions, rarely does.

This asymmetry in volume is the whole story. For the house, every table and every reel is one stream in a river of bets running around the clock, so its results settle onto the expected edge almost immediately in relative terms. For a guest, a night at the tables might be a few hundred bets, nowhere near enough for the average to fully stabilize. That gap between the house's volume and the player's volume is precisely the space where variance is free to dominate, win or lose.

What is gambler's ruin, and does variance change the outcome?

Gambler's ruin describes a player with a finite bankroll facing a house with far greater resources and any edge against them: continued play trends toward exhausting that bankroll. Variance can change the pace of that journey, making it faster or slower, but it does not change where the journey ends.

A finite stack of chips facing a negative edge and a functionally unlimited counterparty is, mathematically, on a path toward zero the longer play continues. Variance decides how that path looks: a high-variance game might produce a dramatic early spike followed by a faster fall, while a low-variance game might produce a longer, flatter, slower decline. Both are heading toward the same destination, because the destination is set by the edge, not by the volatility layered on top of it.

This is a description of a mathematical trend over continued play, not a claim about any single session or any single guest's night.

Does managing a bankroll fix the house edge?

No. Bankroll management addresses variance, not edge. It changes how long a given stake is likely to last and how bumpy the experience feels along the way. It does not change the expected final result, which is set entirely by the edge of the game being played.

Sizing bets carefully, setting limits, and choosing lower-variance games can smooth the ride and stretch a session out considerably. What none of it can do is alter the average percentage the house structurally keeps. That number lives in the rules of the game itself. Bankroll discipline is a tool for shaping the experience of variance, not a tool for negotiating with the edge.

So is low variance simply the smarter choice?

Pairing a low edge with low variance, such as blackjack, produces the slowest expected loss and the smoothest experience over time. High variance produces bigger swings in both directions, but the underlying negative edge still collects at its own pace regardless of how dramatic the ride looks.

Smarter is a matter of preference, not mathematics. A guest who values a smooth, predictable evening is well served by low edge and low variance together. A guest who values the possibility of a dramatic swing, understanding that the same edge is quietly at work underneath it, may prefer higher variance. Neither preference changes the arithmetic. The edge is the cost of the entertainment; variance is simply the texture in which that cost is delivered.

Variance is the texture of the experience; the edge is its price.