How our stats work | Floorball Stats

Power ranking and Elo rating

The table ranks teams by points. The power ranking ranks them by strength: an Elo rating, the system chess uses, that looks at who a team beat, by how much and where.

Rating

An average team is at 1500. The gap between two ratings says how likely one team beats the other: a team rated 100 points higher wins about 64% of games on neutral ground. The home team gets 45 points on top.

After every game, points move from the loser to the winner. How many depends on three things:

Surprise. A favourite that wins gains little, an underdog that wins gains a lot.
Margin. A 9:2 counts more than a 3:2, but with diminishing steps: the margin factor grows with the logarithm of the goal difference.
How the game was won. A win in regulation counts as a full win. A win after overtime or penalty shots counts as 0.60 of a win, closer to a draw.

Change = 56 × margin factor × (result − expected result)

Example

A home team rated 1600 meets a visitor rated 1500. With home advantage its expected result is 70%.

It wins 5:3 in regulation: +17 for the home team, −17 for the visitor.
It wins only after overtime: -4. A 0.60 win is less than the 70% it was expected to reach, so even a win can cost a favourite a few points.

Between seasons

Ratings carry over, but in summer every rating moves 25% of the way back to 1500, because squads change. A team new to the league starts at 1450.

Power rank, 7 days

The power rank is the order of the ratings. "7 days" is the rating change over the last 7 days of games, and the small number next to the place shows how many places a team gained or lost in that time.

How good is it?

We chose the settings on all 25/26 and 26/27 games of every league (October 2026). Teams the ratings gave a 70 to 80% chance won 74% of those games. Measured as log loss, the ratings score 0.545 where a coin flip scores 0.693; lower is better. Ratings know nothing about injuries, transfers or a goalkeeper's day.

Win probability

Each game page shows how likely each team is to win, overtime and penalty shots included, from both ratings plus home advantage. For finished games it shows the chance before the game, which is how we find upsets.

Upset: The winner had less than a 35% chance before the game. The league page lists the three biggest upsets of the season.
When it is shown: Once both teams have played at least 3 rated games. Before that a rating is mostly a guess.

Game story

The line on a game page shows the home team's chance to win at every moment of the game: at the top the home team is sure to win, at the bottom the visitors, in the middle it is even. It starts at the pre-game chance and moves with every goal and with the clock, because a lead is worth more the less time is left.

How the chance is worked out

Three things go in: the pre-game chance from the ratings, the goal difference and the share of regulation time left. We fitted their weights on every minute of 1,814 decided games of the 25/26 and 26/27 seasons (October 2026). On those games the live chance scores a log loss of 0.410, against 0.558 for the pre-game chance alone, and a team given 85% won 85% of the time.

When regulation ends, a lead has won. A tie goes to overtime, where only a little of the pre-game edge remains.

Key moments

The three goals that moved the chance most, in the order they fell, with the scoring team's chance before and after.

Lead changes, ties, largest lead, comeback

A lead change is a team going ahead after the other team had led, with or without a tie in between. Ties count every equalizer. The comeback is how many goals the winner was behind at worst.

More dramatic than

We add up how far the line moved during the game, up and down. A blowout barely moves it, a game with late swings moves it a lot. The percentage compares that with all other games of the same season phase.

Season projections

We play the rest of the regular season 4,000 times. In each simulated season every team gets a "true strength" drawn around its rating (about ±50 points, because a rating is never exact), every remaining game is played out with the win probabilities above, and the points are awarded the way the league awards them, read from its own results (for example 3 for a win, 2 for a win after overtime, 1 for a loss after overtime). How often games go to overtime also comes from the league's results.

Proj. pts: Average final points over all simulated seasons.
1st, Top 4, Playoffs: Share of simulated seasons in which the team finished first, in the top four or on a playoff place. Playoff places follow the league's format, play-in places included (in the Czech Extraliga the top 10). Ties on points are broken by today's goal difference.
Title: Each simulated season continues into the playoffs with the league's series lengths: the best seed meets the worst remaining one in every round, home ice alternates within a series, and a one-game final is played on neutral ground. Formats where teams pick their opponent are only approximated. Title odds are shown until the playoffs start and only for leagues whose playoff format we have.
What 0% and 100% mean: ">99%" and "<1%" mean it happened in almost every or almost no simulated season. 0% and 100% mean never or always in 4,000 seasons, not that it is mathematically decided.

What's at stake

The season simulations also tell us what a single game decides. For every remaining game we split the simulated seasons into those where a team wins it (overtime included) and those where it loses, and compare its chances in both. One sentence sums it up; the first case that applies is used:

Securing a place: a win takes a team that is not yet safe to a playoff place in practically every simulated season.
Hopes all but over: after a loss a team reaches the playoffs in almost no simulated season.
Duel at the playoff line: both teams have between 20% and 80% today.
Big swing: a win and a loss lie 15 percentage points or more apart for the playoffs.
Title race: 8 points or more for the title.
First place: 10 points or more for first place.

Otherwise not much changes and there is no sentence. The wording comes from fixed templates, never from a language model, so every number in it is a simulation result.

Game of the week: Among the games of the next 7 days, the one where the chances of both teams move most.
Schedule: The average rating of the opponents a team still has to play, and its rank: #1 has the hardest remaining schedule. It explains why a projection can differ from today's table.

Clutch stats

Not every goal weighs the same. These stats count the goals that changed a game, from each game's event list.

Game-changing goals

Goals that changed who leads, at any time of the game. There are two kinds:

Equalizer: ties the game, for example 1:2 to 2:2.
Go-ahead goal: breaks a tie, for example 2:2 to 3:2.

The player page also shows which share of a player's goals were game-changing.

Clutch goal, clutch points

A game-changing goal late in the game: in the last ten minutes of regulation (from 50:00) or in overtime. Every overtime goal is a clutch goal, because overtime starts tied and its goal decides the game. Clutch points are clutch goals plus assists on them.

Game winner (GWG)

The goal that put the winner one goal above the loser's final total. In a 5:3 that is the winner's fourth goal, even if it came early. Games decided by penalty shots have no game winner. On game pages these goals are labelled "Game winner".

Overtime goals

Goals in overtime. Penalty-shootout goals are not goals in these stats.

What does not count

Penalty-shootout goals never count. Own goals change the score but are credited to no player. Penalty shots during the game count like any other goal.

Scoring duos and partners

A duo is two teammates who scored from each other's passes. We count every goal where one assisted the other, in both directions, so "Berg → Lind 5 · Lind → Berg 10" means 15 goals together. Arrows read passer → scorer. Most sources list one assist per goal, so only the last pass counts. On player pages the same data shows the teammates a player scored from (scored from) and set up (set up) most often.

Table history

The chart shows each team's place in the table after every round. The leagues publish no round numbers, so a round is a block of game days: a new one starts when a game day is more than 2 days after the first day of the current round. Friday to Sunday is one round, Wednesday and Saturday are two. A team with a postponed game has one game less in between, as on any table.

The table after each round is rebuilt from the results with points, then goal difference, then goals scored. Some leagues rank ties by head-to-head results, so a place in between can differ from the official table on a tie. The latest round always shows the official table.

Head to head

The league page crosses every team with every other team. A row reads from that team's side: 5:3 means the row team scored 5. Colours follow the form guide: green for a win, red for a loss, amber for a game decided in overtime or by penalty shots. Games still to come show the row team's win chance. The last two columns give points per game against the current top four and against the rest, which shows who beats the strong teams and who collects points against the weaker ones.

Comebacks and leads

Comeback wins: Games a team won after being behind at some point, out of all games it trailed. "Pts trailing" are the points it still took from those games.
Blown leads: Games a team lost, overtime included, after being ahead at some point, out of all games it led.
Ahead or behind after P2: Wins and losses when the team led, or trailed, at 40:00.

When goals fall

Team pages split all goals for (up) and against (down) into 5-minute blocks of regulation time. The best and weakest period compare goals for and against per period; the strongest ten minutes are the two neighbouring blocks with the best goal difference. Overtime goals are left out of the bars.

Streaks and records

Streaks: Wins in a row count wins after overtime or penalty shots too. "Unbeaten" means no loss. A player's point streak counts the team's games in a row in which the player scored or assisted. A game the player missed ends it, because the sources do not tell us who played.
Season records: The standout moments of the season phase: fastest goal from the start, the two goals of one team closest together, the longest run of goals without reply, most goals by one player in a game, the biggest comeback (from two goals down or more), the biggest win, the most goals in a game and the biggest crowd.

How many of a team's goals a player scored or assisted, for example 18 of 38. It shows who a team depends on and compares players across leagues, whatever the scoring level. Lists need at least 15 team goals and 5 points from the player.

Crowds

Average attendance of a team's home games that have a published number, as the leagues report it. The leaders page compares all leagues, with at least 2 home games per team.

Penalties

Every penalty in the game reports, counted per team and per penalised player and split by kind:

2 min: minor penalty, the team plays short-handed for two minutes.
2+2: double minor, four minutes short-handed.
5 min: major penalty, five minutes short-handed.
10 min: personal penalty, the player sits out ten minutes but the team stays at full strength.
Match: match penalty, the player is sent off and the team plays five minutes short-handed.

Penalty minutes (PIM) on player pages come from the leagues' own player statistics; the counts here come from the game events, so a game without an event list adds no penalties.

Data

Results and game events come from the official league and federation sites and are refreshed daily. These stats are worked out on every refresh from the games we have tracked since the 25/26 season. When a source misses an event, for example the last goal of a game, we do not fill it in: that game then has no game winner. Found something that looks wrong? Use the feedback button, we read every message.