Parcae
Glossary · strategy

Variance

Variance is the natural fluctuation in betting results around the long-term expected outcome. Even bets with positive expected value go through losing streaks, and even -EV bets occasionally hit hot streaks. Variance is the gap between what should happen on average and what actually happens in any specific sample.

How It Works

Imagine flipping a fair coin 100 times. The expected outcome is 50 heads and 50 tails. But you’d be unsurprised to see 47-53, or 56-44, or even 41-59. That deviation is variance.

The same applies to betting. A bettor with a 5% true edge over hundreds of bets:

  • Could be down 10% after their first 50 bets (variance running negative)
  • Could be up 20% after their next 50 bets (variance running positive)
  • Should average around +5% over thousands of bets (variance averaging out)

The underlying edge doesn’t change. The path to realizing it is bumpy.

Why It Matters

Variance is the single most misunderstood concept in betting:

Short-term results don’t reflect skill. A bettor who wins 7 of 10 +EV bets isn’t 70% likely to win the next 10. A bettor who loses 7 of 10 +EV bets isn’t broken — they’re experiencing variance. The math only delivers reliably over hundreds or thousands of bets.

This cuts both ways. Bettors who win streaks early often mistake variance for skill and increase their stakes too aggressively. Bettors who lose streaks early often quit before the math catches up. Both errors are caused by misreading variance.

Variance is why bankroll management matters. Even with a 5% edge, you can go broke if you bet too much per game and run into a bad variance stretch. Conservative bet sizing (like fractional Kelly) is designed specifically to survive variance.

Variance is why CLV matters more than ROI in the short term. ROI fluctuates wildly with variance over small samples. CLV is a cleaner skill signal because it doesn’t depend on whether bets won or lost — only on whether you got prices better than the closing line.

How Variance Works for Different Bet Types

Different bets have different variance profiles:

Coin-flip bets (-110/-110): Lower per-bet variance but higher cumulative variance over many bets due to the volume.

Underdog bets (+200, +500, etc.): High per-bet variance — most bets lose, occasional bets pay big. ROI takes longer to stabilize.

Heavy favorites (-300, -500): Low per-bet variance — most bets win, occasional bets lose big. ROI stabilizes faster but the edge is also smaller.

Parlays: Extreme variance because all legs must hit. A 4-leg parlay has dramatic swings.

Arbitrage: Effectively zero variance per bet, by design. Each bet locks in profit regardless of outcome.

Estimating How Long to Wait

A rough heuristic for evaluating whether your strategy is working:

  • Under 100 bets: Mostly noise. Don’t draw conclusions either way.
  • 100-300 bets: Some signal, especially in CLV. ROI is still highly variable.
  • 300-1,000 bets: CLV converges to a reliable signal. ROI is starting to be informative.
  • 1,000+ bets: ROI converges to underlying edge. CLV is highly reliable.

A bettor evaluating their performance after 50 bets is essentially evaluating noise. Patience and volume are essential.

For more on the relationship between expected value and variance, see +EV Betting Explained.