How do odds-scraping bots price out smaller sportsbooks?

What odds scrapers actually collect

• Pre-match and in-play prices across every market you offer, polled often enough to catch every move your traders make.
• Line movement timing: when you move, how far, and in which direction, which reveals your book's position and liability.
• Market suspension patterns, which tell competitors when your traders are unsure and where the soft spots are.
• Promotional odds and boosts, which get arbitraged or copied within minutes of going live.

How the scraping operation runs

• Lightweight clients hit your odds API or poll the same endpoints your frontend uses, rotating IPs to stay under rate limits.
• The feed is normalized and redistributed: to a competitor's trading desk, to an odds-comparison site, or to a syndicate's models.
• In-play scraping runs hottest, because a few seconds of latency on a live line is the difference between a sharp price and a stale one.
• Some operations do not even hide: they scrape through affiliate or data-reseller accounts, staying inside the letter of your terms while violating the spirit.

The cost to a smaller book

A large book can absorb scraped pricing; it has the volume to shade lines and the trading desk to react. A smaller book cannot. When competitors mirror your lines in real time, your traders' edge becomes a public good, and sharp money flows to whoever prices a tick better. You pay the API and infrastructure cost of serving the scraper, then pay again in margin when the copied lines get picked off.

There is a subtler cost too. If your odds appear on comparison sites within seconds, price-sensitive bettors will always find the better number elsewhere when you move first and they move second. Being the market maker without the market maker's scale is a losing position.

Defenses that keep bettors fast

• Score API consumers by behavior: request velocity, market coverage, and poll regularity separate a bettor checking a match from a bot mirroring your book.
• Add jitter and lightweight obfuscation to the public odds feed. Humans never notice; scrapers' parsers break and their latency edge shrinks.
• Throttle the expensive patterns, full-market snapshots at high frequency, while keeping the single-market views bettors actually use instant.
• Feed suspected scrapers slightly delayed or rounded prices. A scraper working from stale data is worse than no scraper at all to its operator.