Every model leaves something out
Models simplify reality. That is what makes them useful, but it also creates blind spots. A team-strength model may not know that a key spine player has been ruled out an hour before kickoff.
The responsible response is not to pretend those gaps do not exist. It is to identify what is included, what is missing and how that should change the way the output is read.
Transparency improves the model
Publishing assumptions makes future improvements measurable. If PTN later adds injury value, strength of schedule or venue-specific effects, the change can be backtested against the earlier baseline.
That is much stronger than quietly changing a model after poor results and presenting the new output as if it had always existed.
The standard PTN should aim for
The goal is not to make the model look perfect. It is to make it understandable, testable and progressively better.
- Date every model snapshot
- Track predictions before results
- Publish core inputs
- Backtest changes
- Separate observed statistics from PTN-derived ratings
Where PTN uses official NRL statistics, the raw inputs are identified as source data. Ratings, percentiles, model probabilities and interpretation are Play The Numbers calculations.