HLTV has deployed a significant accuracy update to its Rating 3.0 system, leveraging data from 29,030 maps played since its initial release, including 9,029 LAN matches.
The updated system now features enhanced map-specific calculations. Previously, Rating 3.0 struggled with newly added maps like Cache, which lacked CS2-specific data and had to rely on global averages. Cache proves to be CT-sided in buy rounds with Terrorists winning just 47.1% of the time, but the system treated it as an average map. Now, each map receives tailored win probability calculations, with differences reaching up to ten percentage points between maps like Anubis and CT-favored Overpass.
The timing of this update addresses another critical issue: Valve's July 2025 economy changes shifted the game toward CT-side advantage, but Rating 3.0's calculations were based on pre-change data with limited samples under the new economy.
Round swing mechanics have been refined as well. Kills occurring when the bomb has fewer than 10 seconds remaining (or 5 seconds if CTs have a defuse kit) are now treated like exit frags, preventing inflated swing values for essentially decided rounds.
A bug causing negative round swing after kills has been fixed. In rare low-sample scenarios where win probability decreased after a frag, players now receive zero swing instead of negative values.
The impact on existing ratings remains minimal. Across 6,687 CS2 maps played between July and September 2026, single-map ratings shifted by just 0.021 on average. Only 9.4% of player-map ratings moved by 0.05 or more, with the largest change being 0.31.
These improvements arrive ahead of the 2026 HLTV Awards and Top 20 Players rankings.
