How our numbers work

A sportsbook's odds turn into chances that add up to more than 100%. The extra is the book's margin (the “vig”). To get the market's fair chance, we convert both sides to percentages and scale them so they add up to 100%.

Over -150

60.0% implied

Under +125

44.4% implied

Total

104.4%: the extra 4.4% is the margin

Fair chance

Over 57.4%, Under 42.6%

Where Pinnacle (a low-margin book that takes sharp action) lists the same bet, we prefer its fair price. Otherwise we use the book's own two sides. A one-sided market, such as anytime touchdown, has no other side to remove the margin from, so we say when a number still includes it.

L10 means how often the player went over this line in his last 10 games, and L20 the last 20. They describe recent form. They are not the chance it happens tonight, for two reasons:

• Small samples swing. A real 55% shot goes 8 for 10 or better about 1 time in 10 by luck alone. Ten games can't tell a hot streak from a real change.
• The line already knows. Books see the same game logs and set the line with them. When we graded last NBA season, no L10 band beat the margin, and our old hit-rate-based NBA picks went 56.1% against a 58.3% break-even.

That is why we show hit rates for context, and never multiply them together or treat them as a chance to win.

Where we show an expected value (EV) built from a hit rate, we first blend the hit rate with the market's fair chance, as if we had also watched 100 more games at the market's price. Ten games then move the estimate only a little, which matches how much they are actually worth. We also cap any EV at +40%: anything higher is a small-sample artifact, not an edge, and we label it “40%+”.

Prop

Over at -150, hit 8 of last 10, fair chance 57.4%

Raw L10 as a chance

80.0% → EV +33.3% (not believable)

Honest chance

(8 + 57.4% × 100) ÷ (10 + 100) = 59.5%

Honest EV

59.5% × 1.667 − 1 = −0.8%

EV = chance × decimal payout − 1. It is a long-run average per $1 staked, before tax, and a single bet usually doesn't look like the average.

Each leg carries the book's margin, and a parlay multiplies the legs together, so it multiplies the margins too. Break-even is the chance the ticket needs to hit for the payout to be fair: 1 ÷ decimal odds. Here is the same coin-flip leg (−110 both sides, fair 50%) stacked up:

A single −110 bet gives up about 4.5% to the book. Six legs give up about a quarter of every dollar.

The hit-rate trap

Four legs at +150, each 8 for its last 10. Multiplying the hit rates says 0.8 × 0.8 × 0.8 × 0.8 = 41%. The ticket pays +3806, so it needs 2.6%. With the margin removed, the market's chance is 2.1%. Our honest estimate, with each leg shrunk as in section 3, is 3.1%. It is nowhere near 41%. That is why Bobby and our parlay pages show the book's number first.

Try it: parlay break-even

Enter each leg's odds. Add the other side's odds to strip the margin and see the fair chance. It uses the same calculation as Bobby and the parlay pages.

A profit boost raises the payout, which lowers the chance the bet needs to break even. We compare two numbers: the fair chance the bet hits (margin removed, as in section 1) and the chance the boosted payout needs. If the fair chance is higher, the boost pays more than the bet is worth.

Bet

+200 (other side -250), fair chance 31.8%

50% profit boost

pays +300, needs 25.0%

Without the boost

EV −4.5%

With the boost

31.8% × 4.00 − 1 = EV +27.3%

Our verdicts: Worth a look at +5% EV or better, Fair between −3% and +5%, and Skip below that. A boost is capped at a maximum stake and the bet still usually loses, especially at long odds. Today's boosts, priced

• EV colors and “40%+”: green, yellow and red rank how far a price sits from our estimate. “40%+” means the number hit the cap and is mostly noise from a small sample.
• Parlay tiers (Safe Play / Conservative, Moderate, Aggressive, Lotto): these describe how long the payout is: more legs and longer odds mean a lower tier. They are not a forecast. A “safe” parlay still needs every leg to land, so read the book's chance shown next to it.
• Edge Score (Strong / Good / Moderate / Weak): a 0–100 ranking that mixes the shrunk value, recent form against break-even and the opponent. It sorts the board. It is not a probability.
• Boost verdicts: see section 5.

• Sports betting is entertainment with a built-in cost: the margin. Most bettors lose over time, and so do most picks, including ours.
• Decide your budget before you start, keep bets small and level, and never chase losses.
• Winnings are taxable, and losses don't always offset them.
• Only bet where it's legal for you, and only if you are 21 or older.

21+ · Gambling problem? Call 1-800-GAMBLER.