CS2 Major Simulator: Singapore 2026 Pick'Em Simulator

This major simulator plays every Swiss stage of the PGL Major Singapore 2026 20,000 times and shows, for each team, the share of runs in which it went 3-0, advanced 3-1 or 3-2, or went 0-3. Build your Pick'Em picks on it, see how often the set would reach 5 of 10, or let it suggest a set. It is an unofficial fan tool, not affiliated with Valve, PGL or HLTV.

PGL Major Singapore 2026 at a glance

The Singapore Major is the second Counter-Strike 2 Major of 2026. It runs from 25 November to 13 December 2026, with 32 teams and a prize pool of $1,250,000, of which $500,000 goes to the winner. The playoffs, 10 to 13 December, move to the Singapore Indoor Stadium.

• Three Swiss stages of 16 teams each. Three wins advance, three losses eliminate, so every stage sends 8 teams on and knocks 8 out.
• Bo1 or Bo3. Matches that decide advancement or elimination are Bo3; every other Swiss match is Bo1.
• Playoffs for the top 8 of Stage 3: a single-elimination bracket of Bo3 matches and a Bo5 grand final.
• Map pool: Ancient, Anubis, Cache, Dust II, Inferno, Mirage and Nuke, as listed for the event.

All 32 teams are invited from the Valve Regional Standings (VRS) published on 2 November 2026, region by region. How many Stage 2 and Stage 3 places a region gets follows how its teams did at the previous Major:

Stage 1 is seeded 1 to 16 by global VRS standing. In Stages 2 and 3 the invited teams take seeds 1 to 8, again by VRS, and the 8 teams that came through the previous stage take seeds 9 to 16 in the order they finished it.

How CS2 Pick'Em works at a Major

Pick'Em is Valve's prediction challenge inside Counter-Strike 2. The rules below are how recent Majors ran it; Valve confirms the details for each event in the game's Major tab.

• 10 picks per Swiss stage: 2 teams to go 3-0, 6 teams to advance with 3-1 or 3-2, and 2 teams to go 0-3. Each team can be picked once.
• A pick counts only on its exact result. A team you picked to advance that goes 3-0 does not count, and neither does a 0-3 pick that finishes 1-3.
• 5 of 10 correct completes the stage's challenge.
• Picks lock when the stage's first match starts, so Stage 2 and Stage 3 picks are made once the previous stage has ended.
• The playoffs are a bracket of 7 picks: the 4 quarter-final winners, the 2 semi-final winners and the champion.

In recent Majors, correct picks also upgraded the Major coin that came with Valve's Viewer Pass, from bronze up to diamond. Whether that returns in Singapore, and what it takes, is Valve's call; nothing on this page depends on it. The Pick'Em page has opened about 12 days before recent Majors, which would put it around 13 November, but only Valve's announcement sets the date.

How to use the Pick'Em simulator

1. Choose the stage. Stage 1 is the first Pick'Em of the event.
2. Read the three percentages on each row: the simulated share of runs in which that team went 3-0, advanced with 3-1 or 3-2, or went 0-3.
3. Tap a percentage to pick that team for that result. The panel below the table counts your picks and, once all 10 are in, shows the share of runs that reach 5 of 10.
4. Disagree with a rating? Use the minus and plus buttons to move a team 25 points at a time, up to 200 either way, and watch the stage rerun.
5. Press "Suggest picks" for a set the search rates highly, then change any pick to compare. The link in your address bar keeps the stage, picks and ratings, so you can share it.

How the simulator works

Team strength: Valve's own ratings

Every team is rated by its VRS points, the ranking Valve uses to invite and seed Major teams. Valve publishes the model: it is Glicko with a fixed rating deviation of 75, so the chance that team A beats team B is 1 / (1 + 10^(−0.973 × (A − B) / 400)). A gap of 200 points gives the stronger team 75.4%, and a gap of 400 gives it 90.4%.

Shrinking the favourites

Valve checks its model against real results and writes that it tends to "overestimate at the high end": favourites win less often than the points say. On Valve's chart the observed win rate rises only about 0.8 points for every point of expected win rate. The simulator therefore pulls every matchup toward 50% by 20% of the gap by default, so the 75.4% above becomes 70.3%. The "Shrink favourites" slider changes that share from 0% (the raw rating) to 60%.

Bo1 against Bo3

Most rated matches are Bo3, so the adjusted figure is read as a Bo3 win chance. A Bo1 is one map, which gives the underdog a better shot: if a team wins each map with chance p, it wins a Bo3 with chance p² × (3 − 2p), and the simulator solves that for p. The 70.3% Bo3 favourite above wins a Bo1 63.9% of the time.

The Swiss pairings, rule for rule

• Round 1 is fixed: seed 1 plays seed 9, seed 2 plays seed 10, down to 8 against 16.
• From round 2 teams only meet teams with the same record. Inside a record group the seed order is the Difficulty Score, also called Buchholz: the current wins minus losses of every opponent a team has faced. Ties go to the higher initial seed.
• In rounds 2 and 3 the highest seed plays the lowest seed available without a rematch. Applied one pair at a time, that rule can leave the last two teams as a rematch, so the simulator picks each opponent so that the rest of the group can still be paired.
• In rounds 4 and 5 the six teams in a group are paired by the rulebook's table of 15 orders, taking the first that has no rematch.
• Matches in which either team is on 2 wins or 2 losses are Bo3. That makes round 3 Bo3 for the 2-0 and 0-2 groups and Bo1 for the 1-1 group, and rounds 4 and 5 all Bo3.

20,000 runs and a fixed seed

Each figure is the share of 20,000 simulated stages. For a 50% share that leaves a standard error of about 0.35 percentage points, so treat a difference of a point or less as a tie. The random generator is seeded, so the same settings always give the same table; "New seed" shows how much the numbers move between runs.

How "Suggest picks" chooses

The search starts from the set with the highest expected number of correct picks: it tries every pair of 3-0 picks and every pair of 0-3 picks and fills the 6 advance slots with the best of the rest. It then scores changes on what matters for the challenge, the share of runs with at least 5 correct, trying every swap of a picked team for an unpicked one and every exchange of slots between two picked teams, and keeps going while a change helps. Because it scores all 10 picks inside the same runs, it rewards picks that tend to come true together. It is a local search, so it finds a very good set rather than a proven best one.

Limitations

• The field is not final. Until the 2 November 2026 snapshot the teams are projected from the latest VRS standings, and until a stage ends its qualifiers for the next stage are stand-ins, marked as such in the table.
• No maps. Real matches are decided by the veto and the map pool. The model sees one number per team.
• Rosters and form. A stand-in, a late roster change or a team peaking at the right time is not in the points until results show it. Nudge the rating if you know something the points do not.
• Independent matches. Each match is drawn on its own, with no momentum between rounds.
• Swiss stages only. The playoff bracket is explained above but not simulated yet.

When this page is updated

• 5 October 2026: the projected field and ratings now come from the October VRS update.
• 2 November 2026: the snapshot that decides the invites. The field switches from projected to invited once teams accept.
• Between stages: the 8 teams that advance replace the stand-ins, seeded 9 to 16 in the order they finished, before the next stage's picks lock.

Frequently asked questions