The anchor choice is a modeling decision with real consequences: an anchor with stale prices or wide markets produces phantom edges everywhere. Good practice is anchoring on the sharpest available book for that sport, requiring a valid devig pair (both sides present and fresh), and distrusting any “edge” that appears only when the anchor hasn't moved yet.
ParlayAPI's +EV scanner is built on the anchor pattern, and the feed exposes what anchoring needs: every book's two-way prices in one response, plus per-book freshness at /v1/meta/source-quality with thresholds at /v1/meta/per-book-sla, so you can verify the anchor is current before trusting a fair line derived from it.