Model Predictions
10th round predictions for Super Netball.
To refresh your memory, here’s a description of how my model works.
Standard Model
Predictions for the 10th round are given in the table below. The priors for these distributions are based on the final abilities from last season and the results from rounds 1–9. However, given the Super Shot, it’ll take a little while for the model to adapt to the adjusted goals.
Home | Away | Chance of home team winning |
---|---|---|
Melbourne Vixens | West Coast Fever | 70% |
NSW Swifts | Queensland Firebirds | 59% |
GIANTS Netball | Sunshine Coast Lightning | 28% |
Adelaide Thunderbirds | Collingwood Magpies | 63% |
Score Differential Distributions
The distribution of the predicted score differentials is shown in the figures below. For each game, the chance of the home team winning is calculated and shown in the figure title. Each distribution is coloured by the team’s colours, and the more of a single team’s colour, the higher the chance that team has of winning. Overlaid on each figure are two vertical, dashed dark green lines; these show the score differential that has a 50% predicted chance of happening.
Non-HGA Model
I’ve also built a model that doesn’t utilise a home ground advantage parameter. Given the majority of teams don’t have a home crowd cheering them on, it seems unlikely that there’s any advantage to being named the ‘home team’ in a game. The non-hga predictions are below.
Model Comparisons
In this section we look at how the model estimates team ability from the previous round, as well as model performance.
Team Ability
Team ability can be understood as the expected score difference that the team contributes to the final expected match score difference. For example, if the Vixens are playing at home and have an estimated ability of 1 against the Magpies who have an estimated ability of -1, the expected match difference will be 2 (1 - -1).
Latest Team Ability
The figure below shows the latest team ability estimate. The points are the median predicted ability (as estimated by the model). The dark shading shows 50% prediction intervals, the lighter shading shows 50% prediction intervals.
Season Team Abilities
The figure below shows how team ability has evolved over the season to date.
Comparison of Game Results Against Predictions
The following figure shows the results (score difference of the home team) in round 9 (blue vertical line) against the predictions from the model.
The mean average prediction error for the model over the rounds to date is 12.03. The results are shown below, with the MAPE for each game.
Home | Away | Chance of home team winning | Home Goals | Away Goals | MAPE |
---|---|---|---|---|---|
GIANTS Netball | Melbourne Vixens | 34% | 53 | 73 | 16.3 |
Collingwood Magpies | NSW Swifts | 32% | 54 | 69 | 11.8 |
Adelaide Thunderbirds | Sunshine Coast Lightning | 28% | 50 | 56 | 8.2 |
Queensland Firebirds | West Coast Fever | 18% | 60 | 71 | 7.7 |