The Newsroom
How news arrives on this site:
1. Sourcing
Every hour the news pipeline pulls public RSS feeds from sports outlets covering NBA, NHL, MLB, and NFL. Headlines that match our subject patterns (injuries, transactions, milestones, results, quotes) get queued for editorial review. Headlines that look like betting previews, listicles, or speculation get rejected at the door.
2. Sports-desk pipeline
Each candidate runs through five stages, modeled on a real newsroom:
- Assignment Editor — classifies the story (injury / transaction / milestone / game-result / quote / feature) and dedupes against the last 48 hours of published posts. Same player + same story type = killed before it gets written.
- Researcher — looks up every named player in our canonical roster JSON and prepares a facts block (current team, position, jersey number) so the writer can’t drift to stale-knowledge errors.
- Drafter — composes the article using ONLY inputs from the source headline plus the researcher’s facts. No outside knowledge.
- Fact-Check Editor — compares the draft against canonical facts. If the draft says “the Lakers’ Doncic” but the roster has him on a different team, the post is killed, not corrected.
- Publisher — attaches player tags + team category, runs final dedup at the URL level, and pushes to WordPress with a timestamp.
3. Daily / weekly digests
Once per morning a “Today’s Top Headlines” digest collects the best 5 items per league. Sundays get a Rafty Week-in-Review post: 7-day prediction record, biggest hit, biggest miss per league. Both are auto-generated and idempotent — re-running the same day overwrites that day’s post.
Open the live newsroom feed, the editorial category, or read about the prediction model at methodology.