Variance Explained
Two players sit at the same game with the same odds. One walks away up big; the other busts. Nothing was rigged — that's just variance, the natural scatter of random outcomes around their average. Variance is why a losing game can pay you tonight, and also why it will take that money back if you keep playing. Understanding it is the difference between reading the short run correctly and fooling yourself.
Expected value vs. variance
Expected value (EV) is the average result of a bet if you could repeat it forever — the center of gravity. Variance (and its square root, standard deviation) measures how far individual results tend to land from that center. EV tells you where things settle in the long run; variance tells you how wild the trip is along the way.
They're independent. Two bets can share the exact same EV while one barely twitches and the other swings violently. A bet that pays a little, often, has low variance. A bet that pays a lot, rarely, has high variance — even if, on average, they're worth the same.
A coin-flip example
Flip a fair coin 10 times. The expected number of heads is 5. But you won't always get 5 — that's variance. Here's roughly how 10 flips scatter:
| Heads in 10 flips | Roughly how likely |
|---|---|
| Exactly 5 (the average) | About 25% |
| 4 to 6 | About 66% |
| 3 or 7 | About 23% |
| 0–2 or 8–10 (extremes) | About 11% |
The average is 5, but landing exactly on it is the exception, not the rule. Most sessions miss the mean in one direction or the other. That spread is variance, and it's why no short run ever looks like the textbook average.
Key idea: expected value is the destination; variance is how far you wander off the road getting there. In the short run, variance is louder than EV — which is exactly why short runs lie.
The law of large numbers
Here's the twist that decides everything. Variance dominates a small number of bets, but the law of large numbers says the more you play, the closer your actual results crawl toward the expected value. Ten flips might give you 7 heads. Ten thousand flips will sit almost dead on 50%. The swings don't disappear — but they shrink relative to the growing pile of bets, until the average shines through.
For a casino game, this is the house's quiet superpower. The operator isn't betting once; it's banking millions of rounds, deep in long-run territory where variance washes out and the edge is all that's left. You, betting a few dozen times, are stuck in the noisy short run where anything can happen.
High variance vs. low variance
Games and bets sit on a spectrum. Low-variance bets — even-money roulette colors, conservative blackjack, cashing out crash at 1.2× — produce small, frequent results that hug the average. High-variance bets — slots with rare jackpots, a single number in roulette, holding crash out to 20× — produce long droughts punctuated by occasional spikes.
| Style | Example | What it feels like |
|---|---|---|
| Low variance | Even-money bet, crash at 1.2× | Steady, gentle drift |
| High variance | Slots jackpot, crash at 20× | Dry spells, sudden spikes |
Why variance lets you win — then takes it back
This is the whole story in one paragraph. A casino game has a negative edge, so its expected value is below your stake. But variance scatters real outcomes around that average — and some of that scatter lands above the line. That's a winning session: not a beaten system, just variance breaking your way for a while. The trouble is that the same variance is symmetric, and the law of large numbers is patient. Keep playing and your results march back toward the (losing) expected value. High variance can make the good nights bigger and the wait longer, but it can't change the direction the average points. It buys you hope, not an edge.
On Riskr you can watch variance in the open: stack a pile of even-money bets and see your balance drift, or chase a 50× crash and feel the long droughts and sudden spikes — all with fake money, where the only thing variance can win or lose for you is a spot on the leaderboard.