What is variance in sports betting?

Variance is the statistical measure of how far individual outcomes deviate from the expected average. In sports betting, a positive expected value (EV) bettor still loses many bets, variance determines how wildly your bankroll swings before converging toward the true win rate. A 55% bettor at -110 odds has positive EV but can easily endure 10+ bet losing streaks in a season. Understanding variance prevents abandoning a winning strategy during an inevitable downturn.

How many bets do I need to confirm I have an edge?

For a bettor winning at 55% against -110 lines, you need roughly 1,000-2,000 bets to be statistically confident (at the 95% level) that your results aren't just luck. At 500 bets, a 55% true-skill bettor could easily appear to be hitting at anywhere from 50% to 60%, making it impossible to distinguish skill from chance. The larger the edge, the fewer bets needed. A 58% bettor might need only 400-500 bets for significance.

What is the law of large numbers in betting?

The law of large numbers states that as the number of trials increases, the observed average converges toward the expected value. In betting, this means your actual win rate will approach your true win rate over time. But the convergence is slow; it's proportional to the square root of sample size, not the sample size itself. You need 4x the bets to cut the confidence interval in half.

What is CLV and why does it matter for measuring edge?

Closing Line Value (CLV) measures how much better your bet price was compared to the closing line, the final odds before the event starts. If you bet a team at +3 and the line closes at +2.5, you captured 0.5 points of CLV. CLV is the most reliable proxy for long-term profitability because it requires a much smaller sample size to become statistically significant than raw win/loss records, approximately 200-400 bets versus 1,000-2,000.

How does standard deviation affect bankroll management?

Standard deviation quantifies the expected size of your bankroll swings. For flat betting at -110 with a 55% true win rate, one standard deviation over 100 bets is approximately 5 units. This means your bankroll will swing by ±5 units from expected about 68% of the time, and by ±10 units about 95% of the time. Setting your unit size so that a 2-standard-deviation downswing won't bust your bankroll is the mathematical foundation of proper bankroll management.

Statistical Variance: Why Sample Size Is Everything

The most dangerous phrase in sports betting is “I'm on a hot streak.” Short-term results in betting are dominated by variance, the random noise that makes a 55% bettor look like a genius over 50 bets or a fraud over the next 50. Understanding variance isn't optional for serious OwnTheLines players; it's the mathematical foundation that separates sustainable profit from gambling on luck.

Standard Deviation: Measuring the Swings

Standard deviation (SD) quantifies how much your results deviate from expectation. For flat bets at standard -110 odds, one SD over N bets is approximately √(N × p × (1 - p)), where p is your true win probability. For a 55% bettor over 100 bets, one SD is about 4.97 units, meaning you'll finish within ±5 units of expected roughly 68% of the time.

Confidence Intervals for a 55% Bettor at -110

SampleExpected Profit1 SD Range2 SD Range

100 bets+5.0 units0.0 to +10.0-5.0 to +15.0

500 bets+25.0 units+13.9 to +36.1+2.8 to +47.2

1,000 bets+50.0 units+34.3 to +65.7+18.6 to +81.4

2,500 bets+125.0 units+100.2 to +149.8+75.3 to +174.7

Note that at 100 bets, the 2 SD range includes negative territory (-5 units). This means a true 55% bettor has roughly a 16% chance of being unprofitable after 100 bets. Only at ~400+ bets does the 2 SD floor consistently stay above zero.

The Law of Large Numbers

The law of large numbers (LLN) guarantees that your observed win rate will converge toward your true win rate as sample size grows. But the convergence rate is painfully slow; it scales with √N, not N. Doubling your sample from 500 to 1,000 bets only narrows the confidence interval by about 29%. To cut the interval in half, you need four times the bets.

This has a direct practical implication: be patient. A new bettor who evaluates their strategy after 50 or 100 bets is making decisions based almost entirely on noise. The minimum viable evaluation window for standard spread/total betting is 500–1,000 bets, roughly one full sports season of consistent volume.

CLV: The Fast-Track Proxy

Closing Line Value (CLV), the difference between the odds you bet and the closing odds, is the single best proxy for long-term profitability. Efficient closing lines incorporate all available information, so consistently beating them implies you have an edge the market eventually recognizes.

The key advantage of CLV as a metric is sample size efficiency. Because CLV is a continuous measurement (not binary win/loss), it converges to statistical significance much faster, approximately 200–400 bets versus 1,000–2,000 for raw win rate. If you're averaging +2% CLV across 300 bets, you can be highly confident you have a genuine edge, even if your win/loss record looks unremarkable.

Practical Application: Bankroll Sizing

Variance directly informs unit sizing. The Kelly Criterion suggests optimal bet size = edge / odds, but full Kelly produces extreme swings that most bettors can't stomach. Quarter Kelly (25% of Kelly) reduces bankroll volatility by ~75% while sacrificing only ~43% of long-term growth. For a 55% bettor at -110, this means bet sizing around 1.25% of bankroll per bet instead of ~5%.

For deeper bankroll management frameworks, see Bankroll Management 101. To apply these statistical principles to building a projection system, explore Building Your Own Forecasting Model.