Guide · How betting software works
How odds scanners get their data
Where arbitrage and +EV scanners get sportsbook prices, how often that data refreshes, and the four places errors creep in between the book and your alert.
Updated Nov 1, 2026Facts checked Nov 1, 2026
See the software we recommend for thisOn this page
Every arbitrage and +EV scanner does the same four jobs: it collects prices from sportsbooks, matches identical bets across those books, does the math, and alerts you. The scanner's quality is mostly decided before the math, in how fresh and how accurately matched its data is. That's also where bad alerts can start.
Step 1: How odds scanners get their data
Scanners get prices in two broad ways.
Odds feeds. Specialist data companies collect prices from many sportsbooks and sell them through an API. A scanner built on a feed inherits its coverage and its refresh rate. The Odds API, one commercial aggregator, publishes its rates: main pre-match markets every 60 seconds, in-play main markets every 40 seconds, player props every 60 seconds, futures every five minutes before an event, and betting exchanges every 20 seconds pre-match and 10 seconds in-play.[1] Other feeds may refresh at different rates, and a scanner that doesn't say where its data comes from leaves you guessing.
Their own collection. Some scanners gather prices directly from sportsbook websites and apps. That can be faster, but sportsbooks' terms can forbid it. DraftKings' terms of use, for example, prohibit using automated means such as bots, spiders and screen scrapers to collect information from its site.[3][4] Collection that breaks a book's terms can be cut off without notice, which shows up for users as a book that silently drops out of coverage.
Either way, the scanner is working from polled snapshots. A price is always some seconds old by the time it reaches you.
Step 2: Matching the same bet across books
To compare prices, the scanner has to know that "Lakers ML" at one book and "Los Angeles Lakers - Moneyline" at another are the same bet. Team names, player names, market labels and line formats all differ between sportsbooks, so every scanner keeps a mapping layer that translates them. This is the least visible part of the product and one of the most important.
Matching fails in two ways:
- Missed matches. The scanner can't tell two listings are the same, so a real opportunity never shows up.
- False matches. The scanner lines up two bets that look alike but settle differently. That's worse, because it produces alerts for an "arbitrage" that isn't one.
Settlement rules are where false matches hide. A Pennsylvania sportsbook's house rules, for example, define regular time in soccer as 90 minutes including injury time but excluding extra time and penalties, and settle bets on that result.[2] A market that includes extra time, or a two-way line that refunds on a draw, is a different bet even if it carries the same teams and a similar name. Player props raise the same problem when books treat a player who doesn't play, or a stat correction, differently.
Step 3: Doing the math
Once prices are matched, the scanner runs one of two calculations:
- Arbitrage: whether the best prices across books, taken together, imply total probability under 100%. Our arbitrage calculator does the same check by hand.
- +EV: whether one book's price beats an estimated fair price, built for example by removing the margin from a sharp book's line. See implied probability and no-vig odds and the EV betting calculator.
The arbitrage math is mechanical. The +EV math depends on which books the scanner treats as sharp and how it removes the margin, which is a large part of why +EV tools disagree.
Step 4: Alerting you
The alert is the last link and the slowest one, because it includes you: between collection and your bet slip, the book may have moved the line, suspended the market or cut your maximum stake. Scanner alerts and stale odds covers what that means for whether an alert can actually be bet.
Where errors come from
| Stage | Typical failure | What you see |
|---|---|---|
| Collection | Book cut off from a scraper | A book quietly disappears from coverage |
| Matching | Different settlement rules lined up | An "arb" that isn't one |
| Math | Weak fair-price estimate | +EV alerts that don't beat the closing line |
What to ask a scanner vendor
- Where does your price data come from, and how often does it refresh for the markets I bet?
- How do you handle markets with different settlement rules?
- Which books do you cover in my state, and has that list changed recently?
- For +EV alerts, which books set your fair price?
A vendor that can answer those with numbers and specifics is telling you how its product actually works. Our scanner testing methodology measures the same things from the outside.
Frequently asked questions
Where do odds scanners get their prices?
From commercial odds feeds sold through an API, or from their own collection from sportsbook websites and apps. Either way, the scanner works from snapshots taken on a cycle, not live quotes.
How often do odds scanners update prices?
It depends on the data source. One commercial feed refreshes main pre-match markets every 60 seconds, in-play markets every 40 seconds, and exchanges every 20 seconds pre-match and 10 seconds in-play.
Why do scanners show arbitrage that isn't real?
Because the scanner matched two bets that look alike but settle differently, such as a market that includes extra time against one settled on regular time, or because the price changed before you reached the bet slip.
What should I ask a scanner vendor about its data?
Where its prices come from and how often they refresh for your markets, how it handles markets with different settlement rules, which books it covers in your state, and which books set its fair prices.
Sources
- 1.Odds API Update Intervals, The Odds APIOfficial vendor documentation · accessed Nov 1, 2026
- 2.House Rules (Exhibit B2, IC and Exhibit No. 1401.6, submitted Sept. 9, 2024), Mount Airy Casino Resort sportsbook, PennsylvaniaOfficial vendor documentation · accessed Oct 31, 2026
- 3.DraftKings Terms of Use, DraftKingsOfficial vendor documentation · accessed Nov 3, 2026
- 4.DraftKings Terms of Use: automated scraping of website information prohibited (archived provision, last verified Oct. 5, 2026), ConductAtlasIndependent research · accessed Nov 3, 2026
About this page
- Editorial checks
- Checked against the sources cited on this page. We correct it when a source changes or a reader reports an error.
- Facts last checked
- Nov 1, 2026
How we research and label evidence: our methodology.
Keep reading
GuideHow betting software works
How to evaluate scanner alerts and stale odds
Why some scanner alerts can't be bet, how to spot stale odds, and a simple log for measuring how many alerts you can actually act on.
Updated Oct 28, 2026
GuideHow betting software works
Odds comparison software: how it works
What sports betting odds comparison software does, where its prices come from, which features matter, and how it differs from an arbitrage or +EV scanner.
Updated Nov 1, 2026
GuideHow betting software works
Why do EV betting tools disagree? Devig methods compared
The same bet can look +4% on one tool and slightly negative on another. How reference prices, devig methods and timing change the number.
Updated Oct 30, 2026
ResearchHow betting software works
The Sports Betting Scanner Industry: Who Owns the Tools and Where the Data Comes From
How the betting software industry fits together, from league data deals to odds feeds and scanners, plus Gambling.com Group's purchase of OddsJam.
Updated Nov 13, 2026