Game Sense

CS2 Major Pick'Em: The Odds, Simulated

20,000 simulated Swiss stages give every team's 3-0, advance and 0-3 odds. Backtested on eight real Major stages first.

Beginner

·

6

MIN READ

·

Beginner

·

6

MIN READ

·

WHAT YOU’LL LEARN

WHAT YOU’LL LEARN

  • 20,000 simulated Swiss stages give every team's 3-0 odds.

  • The model earned the coin in 6 of 8 past Major stages.

  • 4.5 correct picks on average. A random Pick'Em gets 2.8.

  • Only 2 teams go 3-0 and only 2 go 0-3. Six advance.

  • 20,000 simulated Swiss stages give every team's 3-0 odds.

  • The model earned the coin in 6 of 8 past Major stages.

  • 4.5 correct picks on average. A random Pick'Em gets 2.8.

  • Only 2 teams go 3-0 and only 2 go 0-3. Six advance.

CS2 Major Pick'Em guide cover

The Major Pick'Em is ten picks. Two teams to go 3-0, six to advance, two to go 0-3. Get five right and you earn the coin.

Most people fill it in from vibes. The problem with vibes is that the Swiss format is not intuitive: a team you are sure about has to survive three specific matchups against opponents decided by results that have not happened yet.

So we built a model that plays the whole stage 20,000 times and counts what happens. Then we checked it against eight real Major stages before publishing a single number.

The part nobody else publishes: does it actually work

How the Pick'Em model performed across eight past Major Swiss stages

Any site can print percentages. The honest question is whether those percentages were ever right.

We ran the model on eight past Major Swiss stages: Austin 2025, Budapest 2025, and both Stage 1 and Stage 2 of Cologne 2026. For each one it used only the Valve table from before that Major, nothing that happened after.

Filling the safest picks would have earned the coin in 6 of those 8 stages, with 4.5 correct picks out of 10 on average.

For comparison, a random Pick'Em gets 2.8 right and earns the coin about 8 percent of the time. So the model is not magic and it is not a coin flip either. It is roughly one and a half extra correct picks per stage, which over a Pick'Em is the difference between usually failing and usually passing.

Two of eight is also a real number. The model missed the coin twice, and it will miss again. Upsets are not noise to be modelled away, they are the thing that makes a Major a Major.

Every Stage 1 team, with its three numbers

Every Stage 1 team at the CS2 Major with its 3-0, advance and 0-3 odds

Sixteen teams open the Major on 24 November. Here is what 20,000 simulated stages say about each one.

The shape of the table is the useful part. Only two teams can go 3-0 and only two can go 0-3, so those percentages are competing for a very small pot. Six teams advance 3-1 or 3-2, which is why the advance column is full of numbers near 50.

Ninjas in Pyjamas lead the 3-0 odds at 38 percent, well ahead of Liquid at 28 and Eternal Fire at 26, despite Liquid and Eternal Fire sitting below them in points. That is recent form doing the work: both gained the maximum form bonus over the last two months.

At the other end, Chinggis Warriors are at 53 percent to go 0-3 and Lynn Vision at 44. Those are the two picks the model is most confident about, and the 0-3 slots are historically where most people lose their coin.

Worth saying plainly: these numbers move. The Valve table updates monthly and form is a rolling two month window, so the odds on the live tool today are the ones to use, not the ones printed here.

The opening round is already fixed

The eight Stage 1 opening matches with each team's win chance

Round 1 is the one part of the stage you can see in advance. Seed 1 plays seed 9, seed 2 plays seed 10, and so on down the table. Every Round 1 match is best of one.

That matters more than it sounds. A single map is the format where upsets are cheapest, and four of the eight opening matches sit between 61 and 67 percent. Those are not safe games, they are coin flips with a lean.

The two clearest are 100 Thieves at 82 percent against Chinggis Warriors and Eternal Fire at 77 against FlyQuest. Everything else is closer than the seeding suggests.

How the model actually works

Three steps, and none of them are secret.

1. Team strength comes from the Valve Regional Standings, the same table Valve uses to invite and seed teams. With recent form on, a team that gained points over the last two months gets up to 100 added, and a team that lost points gets up to 100 taken away.

2. Each map uses the points gap through Valve's own win probability formula, softened slightly because that formula is too confident at the extremes. Best of three matches are built from the chance of winning two maps first.

3. The whole stage is simulated 20,000 times with the real rules: same record pairings, Buchholz seeding, no rematches, and best of three for every match that decides advancement or elimination.

You can also disagree with it. If you think a team is stronger than its points, for example a new roster or a map pool that suits the Major, you can add or remove strength in steps and the full 20,000 stages run again with your adjustment.

What this does not tell you

Worth being clear about the edges.

It does not know anything a scoreboard does not. No injuries, no visa problems, no stand-ins, no bootcamp reports, no map pool reading beyond what the points already contain. If you have that information, the adjustment buttons exist precisely so you can use it.

Eight stages is also a small backtest. It is the honest number we have, not a large sample, and anybody quoting a model's accuracy from fewer than that is quoting nothing.

And only Stage 1 can be modelled before the Major starts. Stage 2 and Stage 3 fields depend on who survives, so those odds appear once the previous stage is done.

Where to go next

Build yours on the Major Pick'Em helper. It shows your chance of hitting five or more as you pick, and it has a fill safest button if you just want the model's own answer.

For context on who is even at this Major and how they got in, how the Major invites work covers the Valve table that decided all 32 places, including the region where the last invite came down to three points.

For the event itself, the Singapore format guide has the three Swiss stages, the new map pool with Cache in, and the first third place decider in Major history.

And while the Major is on, the Pro Scene matches page carries every live series with the score, updated every minute.

Once the stage starts, the PGL Major Singapore page shows every match by day with a live score and our win prediction, so you can check your picks as they play out.

KEEP IMPROVING

KEEP IMPROVING

GET YOUR DEMOS REVIEWED.

FIND YOUR MISTAKES. FIX THEM.

Want someone to look at your own games? Book CS2 coaching and get your demos reviewed, or join the CS2 Rank Up Society for a structured 20-day plan and weekly accountability.

Join the CS2 Rank Up Society for the 20-day plan, or book CS2 coaching and get your own demos reviewed.

Join the CS2 Rank Up Society for the 20-day plan, or book CS2 coaching and get your own demos reviewed.

$5/MONTH · CANCEL ANYTIME

NEXT STEP

$5/MONTH · CANCEL ANYTIME

COMMUNITY

COMMUNITY

FIND TEAMMATES. ASK ANYTHING.

Queue with players at your level instead of going solo, and ask whatever you are stuck on in the questions channel.

Queue with players at your level instead of going solo, and ask whatever you are stuck on in the questions channel.

JOIN THE DISCORD

→

Brian Sirbu, CS2 coach at GameSenseLab

WRITTEN BY

Brian Sirbu

CS2 COACH // GAMESENSELAB

STOP IMPROVING ALONE

Turn scattered practice into a repeatable system with feedback, accountability, and a clear next round.

Turn scattered practice into a repeatable system with feedback, accountability, and a clear next round.

$5/MONTH · CANCEL ANYTIME · +200 ELO GUARANTEE

GAMESENSELAB

.PRO

CS2 coaching, training systems, and improvement content.

CS2 coaching, training systems, and improvement content.

© 2026 GAMESENSELAB.PRO

© 2026 GAMESENSELAB.PRO

© 2026 GAMESENSELAB.PRO