Been building match-winner models for CS2 and Valorant for a while now, and there's one mistake I see costing bettors more money than bad luck ever could: blindly backing the favorite because of raw team ranking.

Here's the trap. Rankings are lagging indicators — they're built on results from weeks or months ago. They don't account for:

• Roster changes. A team that just swapped an IGL or stand-in is not the same team the ranking reflects. Give it 3-5 series before you trust old form again.

• Map pool drift. Team A being "better" doesn't matter if the series is on their worst map. Always check map-specific win rates, not overall record.

• Fatigue and travel. Back-to-back LAN days or long travel windows quietly tank performance, especially in best-of-ones.

• Bo1 vs Bo3 variance. Underdogs close gaps hugely in single-map formats. A "70% favorite" narrative from a Bo3 doesn't hold in a Bo1 — the real number is much closer to a coinflip in a lot of matchups.

The fix isn't "always fade favorites" either — that's just the same lazy heuristic flipped. The fix is: treat every match as a fresh data problem. Recent map-specific form, current roster, and format all matter more than a ranking snapshot.

This is exactly the kind of signal stack we run through before publishing any pick — happy to break down our full process if people want it. What's the biggest "trap" favorite you've seen burn people this year?