VCE probabilistic forecasting
Model uncertainty instead of hiding it.
Enter results, cohort position and practice-exam evidence. The workbench simulates plausible study scores, scaling outcomes and ATAR aggregates entirely in your browser.
Setup
Forecast profile
Model status
Evidence and limitations
Latest result
Forecast summary
No forecast yet
Add subjects and run the model. Results remain on this device.
No subjects added
Add a subject to begin. A valid ATAR forecast normally requires at least four permissible studies, including an English-group study.
No ATAR result
Run a forecast after entering a permissible set of studies.
Distribution
Simulated ATAR outcomes
Thresholds
Probability of reaching targets
Subject contribution
Primary-four and increment probabilities
| Study | Scaled central range | Primary four | Increment | Median contribution |
|---|
No target analysis
Run the main forecast, select a subject and calculate pathways.
Interpretation
Conditional, not guaranteed
Comparison
Saved scenario outcomes
No scenarios run
Select a scenario above.
| Scenario | ATAR median | Central range | Likely range | Confidence |
|---|
No snapshots
Run forecasts with local saving enabled.
| Date | ATAR median | Central range | Confidence | Model |
|---|
Runtime sequence
- Approximate inputs are converted into bounded probability distributions.
- Remaining coursework, final rank and examination performance are simulated coherently.
- A moderated-coursework proxy is generated from rank and cohort evidence.
- Graded assessments are combined into a raw study-score distribution.
- Historical monotone scaling curves generate scaled-study-score distributions.
- ATAR rules are applied afresh in every simulation, including English and increment selection.
This release includes a public VCAA school-performance proxy derived from 2022–2025 school aggregates and 2025 published high-achiever records. It remains an unofficial context prior, not school moderation data. Individual-level fallback coefficients still require institutional validation before consequential use.
External checks
What has actually been validated
School holdouts evaluate the public context proxy. The statewide distribution check evaluates the score transformation. Neither test establishes individual forecast error or advertised interval coverage.
Stored on this device
- Subject inputs and scenarios
- Optional school label
- Forecast snapshots
- User preferences and seed
Storage can be disabled. Exported profiles are ordinary JSON files.
Not included in this release
- Student-level institutional records
- Names, student numbers or credentials
- Private API keys
- Unsuppressed small-cohort statistics
The model bundle is inspectable by anyone who receives the files.
Controls