CS2 BO3 Veto Trainer
Play a full best-of-three veto over the 7-map Active Duty pool. You ban and pick first; a simulated opponent answers using the pick and ban rates professional teams actually recorded, and every choice you make is explained with the figures behind it.
See also: CS2 tools, build your crosshair, pro settings, lineups, guides, maps, console commands, pro crosshair codes.
This is a description, not a strategy. Every number here reports what a sample of tracked professional teams did with each map. Those teams have their own rosters, anti-strats and permabans, and none of that is your team. A map the field bans constantly may well be the right map for you to pick.
The order you are playing
A best-of-three consumes three maps — one pick each, plus the decider — and bans remove the rest. With 7 maps in the pool that is 4 bans, split around the two picks:
- 1You banremoves a map from the board
- 2Opponent banremoves a map from the board
- 3You pickchooses a map the series will play
- 4Opponent pickchooses a map the series will play
- 5You banremoves a map from the board
- 6Opponent banremoves a map from the board
- 7Deciderthe one map nobody removed
Side selection on a map normally goes to the side that did not pick it, which is why teams sometimes leave a strong map in the pool for the decider rather than spending their own pick on it. The trainer does not model side selection.
What the tracked teams did with each map
These are the only figures the simulated opponent uses. Pick rate and ban rate are the unweighted mean across the sampled teams — “the average tracked team spends this share of its picks or bans on this map”. Round win rates are pooled over those teams' rounds; because a tracked team's opponent is not always tracked too, CT and T need not sum to 100.
| Map | Avg ban rate | Avg pick rate | CT rounds won | T rounds won | Team map results | Teams |
|---|---|---|---|---|---|---|
 | 26% | 20.3% | 53.5% / 10,514 rounds | 47.8% / 10,573 rounds | 927 | 19 |
 | 19.7% | 13.1% | 51.9% / 8,521 rounds | 50.9% / 9,048 rounds | 756 | 17 |
 | 14% | 23.9% | 51.7% / 9,638 rounds | 51.3% / 9,820 rounds | 862 | 21 |
 | 24.7% | 7.2% | 44.8% / 4,385 rounds | 54.2% / 4,455 rounds | 399 | 19 |
 | 30.7% | 8% | 53.2% / 7,333 rounds | 48.8% / 7,035 rounds | 660 | 17 |
 | 26.4% | 11.5% | 56.6% / 8,464 rounds | 45.5% / 8,018 rounds | 721 | 19 |
 | 8.6% | 4.5% | 48.5% / 1,195 rounds | 52.8% / 1,217 rounds | 103 | 12 |
Aggregated from 28 professional teams via PandaScore, covering every competitive match tracked for each of them, and refreshed hourly (server cache). Sample: 4,428 team map results over 117,811 rounds. Pool as of 2026-07-08 — "Season 5, Armory, and More" 8 Jul 2026: Cache replaced Overpass (premier-seasons.ts, map-data/cache.ts, overpass.ts).
The line the averages point at
If both sides behaved exactly like the average tracked team, this is the veto the model says is most likely at each step. It is the same model the trainer's opponent uses and the same one behind the veto forecast on our match pages. Read confidence as falling down the list: each step assumes every step above it went the expected way.
- You banAncient17.1%Nuke 15.6%, Mirage 15.5%
- Opponent banNuke19.4%Mirage 19.2%, Anubis 18.6%
- You pickDust 228.7%Mirage 24.9%, Inferno 18.6%
- Opponent pickMirage40.6%Inferno 26%, Anubis 18.1%
- You banAnubis46.6%Inferno 34.9%, Cache 18.4%
- Opponent banInferno76.8%Cache 23.2%
- DeciderCachewhatever remains
No step is shown near certainty on purpose. The probabilities come from ranking the remaining maps on one measured rate and softening that ranking, so a heavy permaban lands somewhere around half rather than at ninety per cent — a veto model that printed one confident sequence would be claiming knowledge these figures do not contain.
How to read the explanations
After each of your choices the trainer states the map's average ban rate and pick rate, its rank on whichever of those two figures your action was about, and the sample behind it. Where your choice differs from the map the field removed or took most often, it names that map and both percentages so you can see the size of the gap yourself. It never blends the numbers into a single score — a “map strength” figure would hide which measurement actually moved, and there is no honest way to weigh a ban rate against a round win rate.
Rankings on this page are always on one named figure at a time. “Most banned” means highest average ban rate across the sampled teams and nothing else. On the current sample that is Ancient at 30.7%, while the most-picked map is Dust 2 at 23.9%.
Two things the trainer knowingly does not model: side selection, and any specific opponent. If you want a veto priced between two named teams, the veto simulatordoes that with each team's own record; for the underlying per-map figures see map meta stats and the map comparison table.
Frequently asked
- How does a best-of-three CS2 map veto work?
- Over the 7-map Active Duty pool the order is: the first side bans, the second side bans, the first side picks map one, the second side picks map two, then each side bans once more and the single map left over becomes the decider. The trainer puts you on the side that bans and picks first.
- What drives the simulated opponent?
- Only two measured figures: the average pick rate and the average ban rate for each map across the 28 professional teams in our sample. The model ranks the maps still on the board on the relevant one of those two rates, converts the ranking into probabilities, and draws from that distribution rather than always taking the single most likely map. It has no knowledge of any specific team, so it prices no matchup at all.
- Does the trainer tell me the correct ban?
- No, and it deliberately never claims to. It reports what tracked professional teams actually did with each map — which is a description of a field of teams with their own rosters, preparation and permabans, not a recommendation for your team. A map the field bans constantly can still be your best map.
- How big is the sample behind these numbers?
- 4,428 team map results across 28 tracked teams, covering 117,811 rounds. A team map result is one map from one team's record, so a game between two tracked teams contributes one result to each of them — it is not a count of distinct matches. Per-map sample sizes are in the table below and differ map by map.
- Which map do professional teams ban most?
- In the current sample it is Ancient: the average tracked team spends 30.7% of its bans on it, over 660 team map results. Ban rates move with the meta and with Valve's map changes, so this figure is generated from live data rather than fixed.
- Why is the pool exactly these maps?
- The veto uses the Active Duty pool — Mirage, Inferno, Dust 2, Anubis, Ancient, Nuke, and Cache — which is the pool Premier and professional tournaments restrict their pick-ban phase to. The current rotation took effect on 2026-07-08 ("Season 5, Armory, and More" 8 Jul 2026: Cache replaced Overpass (premier-seasons.ts, map-data/cache.ts, overpass.ts)).