CASE STUDIES
Validated against real elections.
RPTK has been backtested against real campaigns using real articles. No synthetic data. No retrospective fitting. Here are the results.
NC Governor 2024
Robinson vs. Stein — 39 real articles from AP, CNN, NPR, and WRAL. Could the model detect a scandal-driven collapse and predict the winner?
3.2pt
Margin Error
All
Personas
39
Articles
UNC Student Body President 2026
Three candidates, one mid-race scandal. Could the model predict the winner, the runner-up, and the dropout — in 6 minutes?
3/3
Calls Correct
12
Personas
$2.43
Total Cost
HOW WE VALIDATE
Every prediction is auditable.
RPTK backtests process articles in strict chronological order — no future information leaks into past predictions. Every persona reaction, approval score, and vote share calculation is logged and traceable to specific research-calibrated parameters.
86%
Focus group match rate
3.2pt
Best margin error
100%
Winner accuracy