Pregame total
160pts
What the market expected for the whole game.
Methodology
How we find an edge.
Sportsbooks are good at what they do. Most of the time the line is right, and the honest answer is "no bet." HalfSharp is built to find the few spots where it isn't, prove they held up before we trust them, and show you every result after.
A halftime call, start to finish
Four numbers, one decision.
At halftime the book posts a new second-half line in minutes. Here's how the board reads it.
Points so far
60pts
Already scored. The first half is settled.
2nd-half line
75pts
The book now expects 135 for the game, 25 below where it opened.
The call
Over 75
When the move is bigger than the historical pattern supports, the board says bet. Most halves, it says pass.
Worked example. The exact trigger sizes stay private.
The approach
Built to say no.
Tested on seasons it never saw
Every team carries a rating that updates after every game, from the result and from how the market priced it. Every rule is built on older seasons and tested on later ones it was never tuned on. If it doesn't beat break-even there, it never becomes a pick.
What a model can't see
Every play is checked against the latest injury report, starting quarterbacks, wind at outdoor stadiums and how far the line has moved. Key numbers in football (3 and 7) are priced in. When the market clearly knows something, we pass.
Fewer, better bets
Best bets come only from rules that held up on seasons they were never built on, at most two per league a week, and often none. New ideas are labeled unproven and graded in public. A good live run alone doesn't promote them; they also have to pass a test on seasons they were never built on.
Live from the system
Check our work.
These update on their own. The ratings move after every final; the report changes as injury news and forecasts come in.
- Loading ratings…
Top 25 per league, in points better than an average team on a neutral field. Members see every team.
Loading…
HalfSharp · Best bet- League
- —
- Market
- —
- Locked
- —
- Kickoff
- —
Pick revealed at kickoff
Members get the calls, not just the method. Every halftime call the moment it fires, every best bet before it locks.
Become a memberWhat we don't publish
Why some of it stays private.
We explain the approach so you can judge it. We don't publish the exact rules, thresholds, weights or data pipeline, because anyone could copy them and the edges would close. Members get every pick and signal the system produces.
All backtest results are hypothetical, based on historical games and lines1, and are not results of actual bets. Live results are on the results page. No method guarantees future results.
FAQ
Common questions.
What's a "unit"?
A unit is a fixed bet size you choose, such as 1% of your bankroll. Records are shown in units so they mean the same thing whatever you bet. HalfSharp sizes each call by how selective it is: halftime BETs are 5 units, the deep dive's best bet of the day 3 units, and every other call (halftime leans, early calls, opening-line calls) 1 unit. Smaller bettors who want steady action can play everything at 1 unit; bettors who want to be very selective can play only the 5-unit BETs.
What's the break-even rate?
The win rate you need to not lose money at a given price: 52.4% at the standard −110, and 53.5% at −115, the extra juice books usually charge at halftime. Every call we make has to clear that bar on seasons our models were never built on.
What does the halftime board do?
It lists every game in order of how close it is to halftime. At the half it fills in the second-half line from DraftKings' live odds, compares it with the pregame total, and tells you bet or pass in plain English, with the historical record behind the call.
Why do you sometimes have no picks?
Because most lines are fair. Betting anyway is how bettors lose. When nothing clears the bar, the right play is no play.
Do picks change after they're posted?
No. A call is fixed at the line and price it was made at, with its timestamp. If big news breaks before kickoff on a pregame play (a starting quarterback or a team's best player ruled out), the play is pulled, left out of the record, and you'll see why.
Which sportsbook should I use?
Any legal book in your state. The halftime board fills in DraftKings' live line automatically; if your book's number differs, type it in and the board recalculates.
How is the record graded?
Automatically, from the final score, at the price the play was posted. Halftime calls are graded at −115 (the extra juice books usually charge at the half) on second-half points only. Pushes are no action. Nobody edits a result.
What's the difference between bets, leans and research plays?
Bets are the halftime calls (NFL and college football) and opening-line calls (NBA and college basketball): methods that beat break-even on seasons they were never built on. Leans are the strongest basketball halftime spots, short of break-even, labeled LEAN and kept on their own record. Research plays are ideas still being tested. Everything is graded in public; only bets count in the bet record. All questions →
- About 51,800 games from 2013–2023 (college basketball through 2020–21) with real second-half lines from sportsbookreviewsonline.com archives. The archive doesn't record the juice, so backtests assume the line was available at the number shown. ↩
Founding members
Put the method to work.
Founding members pay $49 $24.50 a month with code SHARP50: 50% off for as long as you stay. Cancel anytime from the app.