Glossary
Betting terms, in plain English.
The words we use on the board, in alerts and in our research, each in a sentence or two. Link anyone straight to a term.
Prices and odds
- American odds #
- The price on a bet. A minus number is what you risk to win $100; a plus number is what you win on a $100 bet. −110 means risk $110 to win $100. +150 means a $100 bet wins $150.
- Implied probability #
- The win chance a price implies: how often the bet has to win to break even at that price. −110 implies 52.4%. +150 implies 40%.
- Juice (vig) #
- The sportsbook's cut, built into the price. A fair coin flip would pay even money; at −110 you pay $110 to win $100, and the extra $10 is the juice. Books usually charge more at halftime, often −115.
- Hold #
- The share of all money bet that the book expects to keep because of the juice. With −110 on both sides it's about 4.5%. Props and parlays usually carry a bigger hold.
- No-vig line (fair line) #
- The price with the juice taken out, so both sides add up to 100%. It's the market's real opinion of each side. −110 / −110 is a 50% / 50% fair line. At the half, the app shows the feed's fair line and stars a book whose line beats it.
- Break-even rate #
- The win rate you need just to not lose money at a price. 52.4% at −110 and 53.5% at −115. Every HalfSharp method has to clear it on seasons it never saw. How we test
- Unit #
- A fixed bet size you choose, so records mean the same thing whatever you bet. HalfSharp sizes calls in units: halftime BETs 2 to 5 by edge, the deep dive's best bet 3, the parlay of the day 0.5, everything else 1. Units and bankroll
- Bankroll #
- The money you set aside only for betting. A unit of about 1% of it keeps even a 5-unit call a modest bet. With a $1,000 bankroll, a unit is $10 and a 5-unit BET is $50.
Lines and markets
- Spread #
- A bet on the margin. The favorite has to win by more than the number; the underdog can lose by less. Chiefs −3: the Chiefs need to win by 4 or more. A 3-point win is a push.
- Total (over/under) #
- A bet on the combined points both teams score, Over or Under the number. Most HalfSharp halftime calls are 2nd-half totals.
- Moneyline #
- A bet on who wins, no spread. The price does the work: favorites pay less than even money, underdogs more.
- Opener (opening line) #
- The first number a book posts for a game. Openers are often a little off, and the market spends the next day correcting them. The opening-line edge
- Closing line #
- The last number before the game starts, after all the betting and news. It's the market's best guess, and the hardest number to beat.
- Line movement #
- How a line changes between the opener and the close, from bets and news. When it moves toward the side you bet, you got a better number than the market ended at.
- Key numbers #
- Scores and margins that come up much more often than others. In NFL spreads they're 3 and 7. The most common NFL 2nd-half totals are 17, 24, 20, 13 and 27, so buying a half-point through one of those is worth more than through 18 or 19. More in our research
Halftime and 2nd half
- 2nd-half line #
- A new total (or spread) the book posts at halftime for the 2nd half only, usually within minutes. It moves for the first few minutes and comes down when play resumes. Books grade 2nd-half totals with overtime included. Second-half betting guide
- Delta #
- How far the book changed its mind after one half: the pregame total minus (1st-half points + the 2nd-half line). Pregame total 160, 60 points at the half, 2nd-half line 75: the book now expects 135, so the delta is 25. Halftime calculator
- Buying points #
- Moving a spread or total in your favor for a worse price. You win more often but each win pays less, so it's only good when the book charges less than the points are worth. Under 44.5 at −115 might become Under 47.5 at −160. On its own it isn't an edge; it rides on a good pick. Buy points calculator
- EARLY call #
- A card on the board when the model likes the very first 2nd-half line, before it settles. Early calls have their own record and don't send alerts; the official call, about 4 minutes into the break, is the one the main record counts.
Measuring an edge
- Closing-line value (CLV) #
- Whether your number was better than the closing line. Beating the close regularly is the earliest sign of a real edge, long before wins and losses settle. You bet Under 47 and it closes at 45: you beat the close by 2 points. HalfSharp tracks it on every call. See the record
- Backtest #
- Running a betting method on past games with the lines that were posted then. HalfSharp's halftime methods were tested on about 51,800 games. Backtests are hypothetical, not results of actual bets.
- Held-out seasons #
- Seasons a model never saw while it was built, used only to test it. A method that only works on the seasons it was built on doesn't really work. The college football model was built on 2015–2022 and tested on 2007–2014 and 2024 on.
- Pre-registration #
- Writing down the test, the data and the bar a method has to clear before looking at any result, so the goalposts can't move afterward. How we test
- Sample size #
- How many bets a record rests on. Even a real 58% edge loses 42% of the time, so a good or bad week means little; judge a record over hundreds of plays.
- Push #
- A bet that lands exactly on the number. Your stake comes back, and it counts as no action in every record.
- Void #
- A bet that's cancelled and graded as no action. A player prop is voided when the player doesn't play, and a pregame pick pulled for injury news is voided and left out of the record.
HalfSharp's labels
- BET #
- A call we stand behind, from a method that beat break-even on seasons it never saw: halftime calls in the NFL and college football, and opening-line calls in the NBA and college basketball. Only BETs count in the bet record. What to do with a BET
- LEAN #
- The strongest basketball halftime spots, short of break-even, so never called a bet. They're graded on their own record, and you can turn their alerts off with BET calls only.
- No bet #
- What the board says when nothing clears the bar, or when the line has moved past the worst number a call is still good at ("Good down to X; pass past that"). Most halves are a no bet, on purpose.
- Opening-line call #
- An NBA or college basketball BET sent the moment DraftKings opens a line our model disagrees with by enough, before the market moves toward it. Why the opener
- Deep dive #
- A daily research report, live at 9:30 AM ET, with one best bet (3 units) and up to two more. It has its own record, kept apart from the halftime and opening-line records. The Deep Dive tab
- Good looks #
- 4 to 8 plays a day in the Picks tab that we like but that don't clear the deep dive's bar: team picks and player props at 1 unit, plus one parlay of the day at 0.5 units. No proven edge, and their own record.
- Research play #
- An idea still being tested. It's graded in public, labeled unproven, and never counted as a bet. Research plays on the record
- Replay #
- A tag on three calls from October 15–17, 2026: the halftime method updated on October 19 was replayed on those games with its settings frozen. Everything from October 19 on was called live.
Missing a term? Tell us. More answers in the FAQ.
21+ only. No call, method or record is a guarantee, and past results, including backtests, don't predict future results. Gambling problem? Call or text 1-800-GAMBLER.