Understanding the Winning Probability
Winning Probability is a predictive framework that estimates candidate competitiveness using historical voting, party strength, and current indicators.
Prediction Notice
This is a data-based prediction model. It uses public historical results and measurable factors to provide an estimated probability.
What is Winning Probability?
It assigns each candidate a 0–100% chance of winning for a constituency by combining multiple signals.
Past vote shares, results, and win/loss records (e.g., 2074 and 2079).
Party structure, alliances, continuity, and ground network.
Visibility, activity level, and constituency-specific dynamics.
Data Sources We Analyze
The model uses publicly available sources and structured datasets. These pillars contribute different signals into the final score.
Historical Voting Records
Official results for 2074 and 2079 (PR votes, direct votes, vote share, win/loss patterns).
Party Structure
Mergers/splits, alliances, candidate switching, and lineage mapping.
Political Indicators
Visibility, activity, recent developments, and local dynamics.
How It’s Calculated
We compute a score for each candidate from multiple signals, then normalize it into a percentage share.
Calculation Steps
- 1
Party Foundation
Base score from PR vote strength and past win context.
base = PR_vote × (won ? 1.6 : 1.3)
- 2
Personal History
Bonuses for wins, seat familiarity, and proven performance.
- 3
Competitive Adjustments
Visibility/activity and close-margin effects.
- 4
Normalization
Convert final scores into 0–100% probability share.
Winning Probability (0–100%)
Why New Candidates May Score Lower
Because the model relies on proven historical signals. New entrants have less verified voting history, so the model becomes less confident.
How to Read the Output
Score
Probability Score (0–100%)
Higher means stronger position; it is not a guarantee.
Confidence
Confidence increases when there is more historical data.
- High Confidence Data Rich
- Low Confidence Data Poor