Winning Streaks, Losing Streaks, and What Variance Really Tells You

Variance, in Plain Terms

Variance measures how far individual results wander from the average. Put simply, it's the reason your actual night at the casino almost never matches the tidy percentage printed on the game info screen.

A slot might advertise 96% RTP, but that doesn't mean you'll leave with £96 for every £100 staked. You might walk out with £120, or £40. That gap is variance doing its thing.

Low variance: results huddle close to the average — steady, if a bit dull.

High variance: results swing hard in either direction — thrilling, occasionally brutal.

Why Streaks Show Up Even Though Nothing's "Happening"

Here's the bit that catches most people out: truly random sequences look far lumpier than intuition expects.

Flip a coin 100 times and you'll land close to 50/50 heads and tails over the long run. But along the way, you should also expect:

• At least one run of 6 or more identical flips in a row
• Multiple stretches of 4-in-a-row
• Patches where one side clearly dominates for a while

If your 100 flips came out as a perfect H-T-H-T-H-T pattern, you'd instantly suspect it wasn't random at all. Genuine randomness is messy and clumpy — and our brains, wired to hunt for patterns, insist on reading meaning into those clumps anyway.

Tracking 100 Roulette Spins

Let's watch variance play out. Imagine 100 bets of £10 each on red, with a true win chance of 48.65%:

• After 10 spins: 3 wins, 7 losses — down £40. Feels grim, but entirely ordinary statistically.
• After 25 spins: 14 wins, 11 losses — up £30. "I've cracked it." No, you've just landed on the other side of normal fluctuation for a while.
• After 50 spins: 22 wins, 28 losses — down £60. "Table's gone cold." It hasn't — it's just doing what randomness does.
• After 100 spins: 46 wins, 54 losses — down £80.

Final tally: a 46% win rate, respectably close to the expected 48.65%, but the ride to get there was anything but smooth. That's variance. The gambler's fallacy would have you believe the early losing stretch owed you a compensating winning streak. It didn't — probability doesn't keep a ledger.

Where the Law of Large Numbers Comes In

The Law of Large Numbers says that as trials pile up, the average result drifts toward the expected value. Crucially, it has nothing useful to say about short-run results.

Take a 96% RTP slot:

• After 100 spins: actual RTP could plausibly land anywhere from 70% to 130%
• After 1,000 spins: narrows to roughly 85%–107%
• After 10,000 spins: tightens to about 93%–99%
• After 1,000,000 spins: sits very near 96%

That convergence isn't the game "fixing" itself — it's simply that new results dilute the weight of earlier deviations. For more on the mechanics, see our Law of Large Numbers piece.

Standard Deviation: Putting a Figure on the Wobble

Standard deviation gives you a sense of how much your results might scatter from expectation:

• 68% of sessions land within 1 standard deviation
• 95% land within 2 standard deviations
• 99.7% land within 3 standard deviations

High-variance slots carry a big standard deviation, meaning most individual sessions will look nothing like the headline RTP. Low-variance games tend to finish closer to what's expected. This links directly with our piece on slot volatility.

What It Means for Your Bankroll

How Long You Can Realistically Play

Higher variance doesn't just create bigger emotional swings — it demands a chunkier bankroll to ride out the rough patches. A £50 budget on a high-volatility slot might vanish in a quarter of an hour purely through unlucky maths, not poor decisions.

Small Samples Lie to You

A single session can't really tell you much about a game's true nature. Even 100 sessions might not be enough to pin down its real RTP with confidence. That's why testing labs simulate millions of rounds on the RNG powering these games before certifying them.

The Trap of Reading Too Much into a Handful of Sessions

We're built to draw big conclusions from small amounts of evidence. Win three sessions out of five and it's tempting to think you've found the angle. Statistically, that outcome is entirely normal even for a game with negative expected value. Your "system" might just be a friendly bounce of variance.

Hot and Cold Streaks — Why the Labels Don't Mean Anything

There's no actual mechanism behind a "hot table" or a "cold machine." Both are simply clusters that appear naturally within randomness. The labels are a psychological reflex, not a statistical finding.

What makes streaks feel important:

• Apophenia: our built-in habit of spotting patterns even in pure noise
• Selective memory: the unusual sticks; the ordinary fades
• Causal thinking: we want reasons for everything. "The table was hot" sounds like an explanation. It's really just a description dressed up as one.

These are the same cognitive biases that shape decisions well beyond the casino floor.

Variance doesn't take sides. It's a mathematical property, not a mood. Understanding it means you can read your own results honestly — not as omens or proof of a system, but simply as the natural scatter you'd expect around an average.