How LINE works
From the odds to your edge, in plain numbers.
How LINE reads a betting market, how odds become probabilities, how its model differs from the price, and how it tells you how much to trust each number.
1 · How LINE analyzes a market
Six steps, every game, every market.
A betting market is a price. LINE’s job is to turn that price into a probability, set its own estimate beside it, and tell you honestly how far apart they are.
- 01
Collect the prices
Odds for every game from a licensed odds feed, from several sportsbooks, each with the time it was quoted.
- 02
Translate them
Each price becomes the probability it implies. The sportsbook’s margin is measured, then removed to get the market’s fair view.
- 03
Estimate independently
CMOS calculates its own probability from team performance, matchups and situation. It never looks at the price.
- 04
Compare
Where LINE and the market disagree: the edge in points, and the expected value at the best price.
- 05
Label
Every number says how far its model has been tested, from Experimental to Verified.
- 06
Rank
The biggest gaps in your sports rise to the top as Today’s Top Bets.
Pick your sports
Follow one sport or all of them. Your home screen, alerts and favorites follow your picks.
See the numbers
Every game: LINE’s win probability, the projected score and the odds, side by side, with the edge called out.
Make your call
Start with today’s Top Bets, dig into any matchup, then bet your way, with better information.
2 · Sportsbook odds comparison
Same bet. Different prices.
Sportsbooks set their own prices, and they rarely agree to the cent. A few points of price on the same bet change how often it has to win to pay off.
- LINE lines up every book’s price for every side, with the time each was quoted
- The best available price is marked, and the edge is calculated at that price
- Old quotes are flagged; a stale price is never treated as current
LINE shows prices; it doesn’t take bets or send you to a sportsbook.
| Sportsbook | Price | Implied | Profit on $100 |
|---|---|---|---|
| Sportsbook A | +150 | 40.0% | $150.00 |
| Sportsbook B Best price | +155 | 39.2% | $155.00 |
| Sportsbook C | +150 | 40.0% | $150.00 |
| Sportsbook D | +155 | 39.2% | $155.00 |
The better price needs fewer wins to break even. Sportsbooks here are unnamed examples.
3 · Implied probability
Every price is a probability.
Odds tell you the payout. Turned around, they tell you how often a bet has to win to break even: its implied probability.
Negative odds: odds ÷ (odds + 100). Positive odds: 100 ÷ (odds + 100). Add both sides of a market and you get more than 100%. The extra is the sportsbook’s margin (the “vig” or “juice”).
| Odds | Implied probability | How |
|---|---|---|
| −110 | 52.4% | 110 ÷ (110 + 100) |
| +150 | 40.0% | 100 ÷ (150 + 100) |
| −200 | 66.7% | 200 ÷ (200 + 100) |
Try it
American odds, like −110, +150 or −200.
Your side, then the other side. Past the line at 100%: the sportsbook’s margin.
4 · Model projections vs market prices
Two probabilities, one gap.
The market’s probability comes from the price. LINE’s comes from CMOS, which never looks at the price. Edge is how far apart they are; expected value is what that gap is worth at the price on offer.
Edge · percentage points
How far LINE’s probability sits above the one the odds imply.
Expected value · percent
What each dollar returns on average if LINE’s probability is right, at the actual payout: −110 pays $0.909 profit per $1.
Example: HAR +3.5 at −110 with a LINE probability of 55.0% (fictional game). A positive expected value still loses often in the short run; it pays off only over many bets, and only if the probability is right.
The market is a strong forecaster
Prices pool a lot of information, and they’re usually close to right. In LINE’s own 2020–2025 NFL backtest, the market was more accurate than LINE overall (see the Report Card). A gap between LINE and the market is a reason to look closer, not proof the market is wrong.
What a projection is, and isn’t
A probability says how often something should happen over many games like this one. A 55% side loses 45% of the time. Positive expected value only pays off over many bets, and only if the probability is right. LINE has not established a validated betting edge yet.
5 · Ask Ace
A research assistant, not a tipster.
Ask Ace answers questions in plain English from LINE’s own data: a matchup, the best price for a side, how a line moved, what your teams have coming up.
- Shows where each number comes from and how old the price is
- Uses the same labels as the app: Verified, Not validated, Experimental
- Says “I don’t have that” instead of guessing, and never promises a result
Ask Ace is in development and not yet in the app. Try the scripted version in the demo.
Which NFL player props have the strongest supporting data?
None qualify right now. LINE’s NFL rushing-yards model is still in research: it has been backtested, but it hasn’t passed validation, and no licensed player-prop prices are connected.
So there are no prop opportunities to rank. I can explain how the rushing-yards research model works, or show this week’s game-level numbers instead.
Preview · scripted example
NBA isn’t live in LINE yet. It’s in development and waiting for a licensed data source, so I don’t have tonight’s schedule, prices or a LINE probability for that game.
I won’t guess. When NBA launches, I’ll be able to walk through LINE’s estimate, the current price and the matchup factors behind it.
Here’s what’s next for the teams you follow (demo account, fictional teams):
- Harbor City at RidgeviewNFL · Sun 1:00 PM
- Northgate at LakeshoreCollege FB · Sat 12:00 PM
None qualify right now. LINE’s NFL rushing-yards model is still in research: it has been backtested, but it hasn’t passed validation, and no licensed player-prop prices are connected.
So there are no prop opportunities to rank. I can explain how the rushing-yards research model works, or show this week’s game-level numbers instead.
Every side of Harbor City at Ridgeview, with LINE’s example probability next to the one each price implies:
| Side | Price | Implied | LINE | Edge |
|---|---|---|---|---|
| Harbor CityMoneyline | +155 | 39.2% | 43.0% | +3.8 |
| RidgeviewMoneyline | −165 | 62.3% | 57.0% | −5.3 |
| Harbor City +3.5Spread | −105 | 51.2% | 55.0% | +3.8 |
| Ridgeview -3.5Spread | −105 | 51.2% | 45.0% | −6.2 |
| Over 44.5Total | −105 | 51.2% | 47.0% | −4.2 |
| Under 44.5Total | −110 | 52.4% | 53.0% | +0.6 |
Example prices from an unnamed sportsbook.
6 · Validated vs experimental signals
How much should you trust a number?
Every edge in LINE carries a label that says how far its model has been tested. A model moves up the ladder only by passing the test for the next step, never by marketing.
Research
A model being built and backtested. Its numbers aren’t shown as bets.
LINE today: NBA, NHL, Horse Racing
Experimental
A research model running on live, licensed data, saved before every game and graded. Shown, labeled Experimental, while it builds a track record.
LINE today: College Football, College Basketball
Not validated
An official model: picks locked before kickoff and graded in public on the Report Card. Its edges against the market haven’t been proven yet.
LINE today: NFL
Validated
Beat the market out of sample (on games it never trained on), with calibrated probabilities: its 60% calls win about 60% of the time.
No LINE model here yet
Verified Top Bet
A validated model, a price under 5 minutes old from a named sportsbook, and positive expected value at that price. All three, or it isn’t Verified.
No LINE model here yet
“Out of sample” means tested on games the model never saw while it was built. “Calibrated” means its probabilities match how often things actually happen. LINE publishes misses as well as hits.
Putting it together
The biggest gaps rise to the top.
Every game day, CMOS scores every market in the sports you follow. The biggest gaps go to the top of the app as Today’s Top Bets, each with its price, its book, its age and its label.
An edge is a probability, not a promise: Top Bets still lose, often. Bet only what you can afford to lose.
Today in the LINE app
Top Bets, every game day
Every day there are games, CMOS scores every market in your sports and LINE puts the biggest edges at the top of the app. Your Top Bets are waiting when you open it.
- Ranked by edgeWhere LINE’s probability beats the odds by the most.
- Across your sportsNFL, college football, college basketball and every sport as it launches.
- The full pictureThe price, LINE’s probability, the edge and expected value on every pick.
- Labeled honestlyEach one says how far its model has been tested: Verified, Not validated or Experimental.
See the method in the app.
Every screen in the demo uses the math on this page.