How CS Data calculates CS2 predictions and stats
This page shows, with the numbers used in the code, how CS Data arrives at its win prediction, series simulation, match significance, Hype Score and day score. No number is tuned by hand: when a formula changes, this page changes with it.
Methodology reviewed on
What is CS Data?
CS Data is a free Counter-Strike 2 stats dashboard that puts live matches, the day's schedule, the world ranking, head-to-head records, win predictions and series simulations on one screen. It needs no account and shows every time in the visitor's own timezone.
Where does CS Data get its data?
Matches, scores, tournaments, rosters and streams come from the PandaScore API. The world ranking, map ban and pick rates, CT and T side win rates and recent results come from HLTV. The ranking syncs every 6 hours and veto data every week.
- The live dashboard polls matches every 60 seconds.
- CS Data covers S, A and B-tier tournaments; the live ticker shows every tier.
How is the win prediction calculated?
The win prediction adds four factors with fixed weights: head-to-head (35%), recent form (25%), world ranking (25%) and map pool (15%). Each factor scores 0 to 1 per team, and the total is normalized so both teams' probabilities add up to 100%.
- Confidence is high when the gap between the teams reaches 25 percentage points and medium from 10.
- Below a 5-point gap, the match is treated as even.
| Factor | Weight | How it is measured |
|---|---|---|
| Head-to-head | 35% | Direct wins divided by all matches between the two teams; 0.5 with no history |
| Recent form | 25% | Win rate in recent matches, ±3 points per game of the current win or loss streak (up to 15) |
| World ranking | 25% | 1 − (position − 1) ÷ 30; teams outside the top 30 score 0.5 |
| Map pool | 15% | Average win rate across the maps played; 0.5 with no data |
How does the BO3 series simulator work?
The simulator plays the same series 10,000 times using the Monte Carlo method. On each map, the win chance comes from each team's win rate on that map, adjusted by recent form, with up to 15 points of random variance and a 5% to 95% cap. The output is each team's chance and the most likely scorelines.
- The swing map is the map whose win split across simulations is closest to 50/50.
- The "What if" controls shift each team's form by up to 20 points and rerun the simulation.
- The same teams and adjustments always give the same result, because the random draw uses a fixed seed.
What does match significance measure?
Significance runs from 0 to 10 and measures a match's competitive weight. It adds the tournament tier (up to 3 points), the stage (up to 3.5), the series format (up to 1.5) and both teams' ranking (up to 2). From 7 points a match is critical; from 5, important.
- In a group stage, the internal bracket name (for example "upper bracket quarterfinal") does not turn the game into a playoff.
| Axis | Option | Points |
|---|---|---|
| Tournament tier | S-Tier | 3 |
| Tournament tier | A-Tier | 2.2 |
| Tournament tier | B-Tier | 1.2 |
| Stage | Grand final | 3.5 |
| Stage | Final | 3 |
| Stage | Semifinal | 2.4 |
| Stage | Quarterfinal | 1.8 |
| Stage | Playoffs | 1.5 |
| Stage | Group stage | 0.5 |
| Format | BO1 | 0.3 |
| Format | BO3 | 0.9 |
| Format | BO5 | 1.5 |
| Ranking | Both teams at the top of the ranking (drops with the average position) | 2 |
How is the day score calculated?
The day score runs from 0 to 10. It starts from the day's most significant match, multiplied by a tier weight (S 1, A 0.8, B 0.6) and by 0.95. It then adds a bonus for the number of critical and important matches plus 0.1 per relevant match, up to 1. Canceled or postponed matches do not count.
- 7 or more: Unmissable day. 5 to 6.9: Great day for CS2. 3 to 4.9: Decent day. Below 3: Quiet day.
- The count bonus grows logarithmically, so the tenth critical match is worth less than the second.
What is the Hype Score?
The Hype Score runs from 0 to 10 and estimates how much attention a match should draw. It adds the tournament tier (S 3, A 2, B 1), the two teams' average ranking (up to 3 points), how close their ranking positions are (up to 2) and the series format (BO1 0.5, BO3 1.5, BO5 2.5).
- Match of the Day is the day's match with the highest Hype Score.
How does the Veto Oracle predict bans and picks?
The Veto Oracle simulates the veto order of a BO3 with the active map pool. Each team first bans the maps it banned most in recent matches recorded by HLTV and, without history, the map where its win rate is lowest. For picks, each team takes the map with its highest win rate.
- In the scouting report, a permaban is a map the team bans in more than 40% of its series.
Are CS Data predictions reliable for betting?
No. CS Data does not yet publish a historical hit rate, and its predictions are statistical estimates built from public data, not a guaranteed outcome. They help explain a matchup. During live matches, unexpected results show up as UPSET alerts.
Sources
CS Data is an independent project with no affiliation to Valve, HLTV or PandaScore.