Track Your AIHL Team's 2026 Season
Final stats and standings for the 2026 season, including Elo ratings, form and player stats for all ten Australian Ice Hockey League clubs.
Standings & Form
→ Scroll to see more columns
| Team | GP | Record | PTS ↓ | Elo | Form (L5) | Goal Diff (season) | Win% | SF% | PP% | PK% | PIM/G |
|---|---|---|---|---|---|---|---|---|---|---|---|
Newcastle Northstars
|
30 | 21-8-1-0 | 65 | 1637 | WLWWW | +51 | 72.2% | 56.2% | 22.2% | 83.8% | 16.1 |
Melbourne Mustangs
|
30 | 19-9-3-0 | 61 | 1614 | WWWWO | +56 | 67.8% | 56.2% | 20.6% | 80.1% | 18.3 |
Canberra Brave
|
30 | 17-6-2-5 | 61 | 1626 | WWWSO | +43 | 67.8% | 55.4% | 29.9% | 80.3% | 10.8 |
Perth Thunder
|
30 | 13-10-5-3 | 48 | 1466 | LLLWL | +11 | 52.2% | 51.8% | 17.0% | 79.7% | 10.5 |
Sydney Ice Dogs
|
30 | 16-12-1-1 | 46 | 1527 | WLWWW | +15 | 56.7% | 47.0% | 29.7% | 79.7% | 15.1 |
Sydney Bears
|
30 | 13-11-3-3 | 46 | 1500 | –WWWL | -1 | 50.0% | 50.6% | 21.9% | 80.0% | 15.6 |
Melbourne Ice
|
30 | 13-12-3-4 | 44 | 1523 | LLLWS | +8 | 51.1% | 53.4% | 22.0% | 71.3% | 18.8 |
Central Coast Rhinos
|
30 | 9-19-1-2 | 30 | 1389 | LLLLL | -49 | 33.3% | 40.5% | 22.0% | 73.6% | 15.9 |
Brisbane Lightning
|
30 | 8-20-1-1 | 27 | 1388 | OWLLL | -52 | 30.0% | 46.8% | 21.9% | 68.4% | 15.2 |
Adelaide Adrenaline
|
30 | 3-23-2-4 | 14 | 1329 | LLLLW | -82 | 15.6% | 43.2% | 15.2% | 78.3% | 21.3 |
Elo Rating Trend
Were the Predictions Any Good this Year? 101/154 called right
Before every game this season, that game's page on this site gave each team a chance of winning — something like “Newcastle 68%, Perth 32%”. The team with the bigger number is the favourite. Here is how all 154 of them turned out.
Was a 70% favourite really a 70% favourite?
For every game this season we calculated the probability of a win for each team,
based on their realtime Elo scores. The predictions ranged from being too close to
call, to choosing a slight favourite, up to a clear and heavy favourite.
We can measure how accurate our predictions were at each band. If our model had
it right, the ‘won’ percentage on each row below should be equal to or
greater than the ‘expected’ percentage. Each bar shows that difference:
right of the line is better than expected, left of it is worse.
percentage points above or below what the model expected
won more often than expected won less often
The model beat its own expectation in every band except clear favourites, where it fell just short.
- It got better as the season went on. Every team starts the year on the same rating of 1500, so in April the model has nothing to go on and barely picks a favourite at all: over the first half of the season it expected its favourites to win 58% of the time, and 63% over the second half. Its record followed: it got 60% of the first half of the season right, and 71% of the second.
- Overtime is where it comes unstuck. Of the 137 games decided inside normal time it picked the winner in 94 of them (69%). Of the 17 that went to overtime or a shootout it picked 7 (41%) — worse than tossing a coin. If two teams are still level after sixty minutes, whatever gap the ratings saw between them has already failed to show up on the ice.
- It was a little too modest. Across the season it expected its favourites to win 60.2% of the time, and they won 65.6%. Home ice is the one place it went the other way: between two evenly matched teams it gives the home side a 55% chance, and home teams won 53% of their games this season.
Changes to the Model for Next Season
The ratings here come from a plain Elo model, the same method used for chess. A full season of results is the best guide to which of its gaps to close first.
- Start from where the season ended. Every team is reset to 1500 each April, so the model spends months working out what it already knew in August. Replaying 2026 from the closing ratings instead turns 101 correct calls into 110, on a hindsight test that flatters the idea.
- Measure home ice instead of assuming it. Every club gets the same bump at home, worth 55% in an even matchup, when home teams actually won 53%. A single number also treats a flight to Perth and a trip across Melbourne as the same journey.
- Treat overtime as its own result. The league gives a team beaten in overtime a point, so a single who-wins percentage answers a narrower question than the ladder asks. Forecasting a win in normal time, a win past it and a loss past it would match how a night is actually scored.
- Use what the site already knows about squads. Signings, departures and suspensions are tracked here already, but none of them move a rating. A team that loses its starting goalie in July carries the same number into its next game.
- Say how sure it is. An Elo rating is one number with no sense of how settled it is, so an April fixture looks as confident as an August one. Glicko, the system built on Elo to fix that, tracks how well established each rating is and pulls its forecasts toward 50% while a team is still an unknown.