Betting Education

The betting glossary

Every term behind a sharp bet, in plain English — the same concepts Covrr computes inside its models, backtests, and interactive calculators.

Closing Line Value Confidence Interval Edge Over Implied Probability Expected Value Fair Odds Implied Probability Kelly Criterion Line Movement Public Betting %Reverse Line Movement Sample Size Sharp Money Sharpe Ratio Statistical Significance Units Variance

Closing line value (CLV)

The difference between the odds you bet and the final odds right before the game starts. Consistently beating the closing line is one of the strongest signs a bettor is genuinely sharp, because the closing line is the market’s most accurate estimate.

Example

You bet a side at +150 in the morning; by kickoff it closes at +120. You beat the closing line. Do that across hundreds of bets and it’s strong evidence your numbers are ahead of the market — even before the profits show up.

Confidence interval

The range a true win rate likely falls within, given your sample. A model showing 60% within a 45–75% range is far less certain than one showing 60% within 57–63%. The tighter the range, the more you can trust the number.

Example

A model that starts 12–3 looks like an 80% winner, but its 95% confidence interval spans roughly 54–94%. After 300 bets winning at 58%, the interval tightens to about 52–64% — now the win rate is a number you can act on.

Edge over implied probability

The gap between how often your model actually wins and how often the odds say it should win. If the odds imply a bet should hit 60% of the time but your model hits 64%, your edge is +4%. Positive edge is where long-term profit comes from.

Example

Odds of −110 imply a 52.4% win rate. If your backtest wins 56% on that market, your edge is +3.6 points — real value. If it wins 51%, you have no edge at −110 even though the bet wins more often than it loses.

Expected value (EV)

The average result of a bet if you could place it infinitely many times. Positive EV means it’s profitable long-term even if it loses tonight; negative EV means it loses over time even if it wins tonight. Individual outcomes are noise; EV is the signal.

Example

A +120 bet that wins 50% of the time averages +$10 per $100 wagered (0.5 × $120 − 0.5 × $100). It can lose three nights in a row and still be a good bet — and a −EV bet that cashed tonight was still a bad one.

Fair odds

The break-even price for a given win rate. If a bet wins 55% of the time, its fair odds are the price at which you’d neither profit nor lose long-term. If the market offers better than fair, there’s value; if worse, there isn’t — even on a winning bet.

Example

A bet that wins 55% of the time has fair odds of about −122. Getting −110 on it is value; paying −135 is not — the same winning bet, profitable at one price and a long-term loser at the other.

Implied probability

The win rate the odds are quietly assuming. Any price converts to a percentage: −150 implies about 60%, +150 implies about 40%. It’s the market’s estimate of how likely something is, and the number your own edge gets measured against.

Example

−150 implies 60% (150 ÷ 250); +200 implies 33.3% (100 ÷ 300). If your model says the +200 side actually wins 40% of the time, the market is underrating it — that gap is your edge.

Kelly Criterion

A formula that calculates how much of your bankroll to wager based on your edge and the odds. The bigger your edge, the more it suggests risking; no edge, and it suggests nothing. Designed to grow a bankroll as fast as possible without risking ruin. Many bettors use a fraction of full Kelly to smooth the swings.

Example

With a 55% win rate at −110 odds, full Kelly stakes about 5.5% of bankroll per bet. Most bettors run half-Kelly — about 2.75% here — accepting slightly slower growth for a much smoother ride.

Line movement

How a bet’s odds or spread shift between opening and game time, driven by incoming money and new information. The direction and timing of movement can reveal where informed money is going.

Example

A spread opens −3 and closes −4.5: money and news favored the favorite all day. A total drifting from 8.5 to 8 with no injury news is the market quietly saying “under.”

Public betting percentage

The share of all bets placed on one side. Heavy public money on a favorite is often a fade signal for sharp bettors, since the public tends to lose long-term. It’s a measure of crowd sentiment, not correctness.

Example

If 78% of bets are on the Cowboys and the line hasn’t moved their way, the crowd is loud but the market isn’t listening — books are comfortable holding the other side.

Reverse line movement

When the betting line moves opposite to where the public money is going. If most bets are on one team but the line moves toward the other, it suggests sharp money is quietly backing the unpopular side.

Example

70% of bets land on the favorite at −7, yet the line drops to −6. The money actually moving the line is on the underdog — that’s sharp action showing itself against the crowd.

Sample size

How many past instances a result is based on. A trend built on 5 games can look spectacular and mean nothing; the same trend over 300 games is worth attention. Small samples are the most common way bettors fool themselves.

Example

A pitcher who’s 4–0 in day games tells you almost nothing — four games. A system that’s 210–150 across three seasons is a signal worth pricing. Same win rate, completely different meaning.

Sharp money

Wagers from professional, winning bettors, as opposed to the general public. When the money on a side outweighs the number of bets on it, it often signals sharp action — professionals betting bigger on the side the public is ignoring.

Example

One side draws 30% of the bets but 65% of the dollars: fewer, bigger, smarter wagers. When books move the line toward that side despite the bet count, they’re respecting the sharp money.

Sharpe ratio

A measure of return relative to how bumpy the ride was. Two strategies can both profit, but the one with steadier, more consistent results scores higher. It rewards reliability, not just raw return.

Example

Two models both finish +12 units. One grinded about +1 unit a week; the other swung ±15 units month to month. The steady one has the far higher Sharpe ratio — and is the one you’ll actually stick with.

Statistical significance (p-value)

A measure of whether a result is likely real or just luck. A low p-value means the pattern probably isn’t chance; a high one means there isn’t yet enough evidence to trust it. A 70% win rate over 10 bets is exciting but not significant; the same over 500 bets is meaningful.

Example

A 7–3 heater has a p-value around 0.17 — easily luck. Going 290–210 over 500 bets has a p-value under 0.001 — almost certainly a real edge, not a hot streak.

Units

A way of measuring bets by risk rather than dollars, so a $10 bettor and a $1,000 bettor can compare performance. One unit is your standard bet size. Tracking in units shows how a strategy performs regardless of bankroll.

Example

If your unit is $50, a bettor who’s +24 units is up $1,200 at your sizing. Their actual dollars don’t matter — units let a $10 bettor and a $1,000 bettor compare records honestly.

Variance

The natural swing of results around their expected value. Even a genuinely profitable strategy will have losing weeks; variance is why. Understanding it keeps a good bettor from abandoning a winning approach during a normal downturn.

Example

A genuine 55% strategy still loses 6 of 10 about one week in every four. That’s not the model breaking — that’s variance doing what it always does, and the reason bankroll rules exist.

See the math in action

Fair odds, expected value, Kelly staking and more — run the numbers yourself with the free calculators, no account needed.

Open the calculators →