We don't just guess.
We verify.

Our accuracy is calculated by dividing the number of correct Safe and Risky predictions by the total number of Safe and Risky predictions made. A prediction is "correct" when a Safe prediction results in a confirmed ticket, or when a Risky prediction results in a non-confirmed (waitlisted/RAC) outcome.

For example, if we made 100 Safe/Risky predictions and 87 matched the actual outcome after chart preparation, our accuracy would be 87%. Medium predictions are excluded from this calculation entirely.

Medium (Uncertain) predictions fall in the 50–60% confirmation probability range, which is essentially a coin flip. These tickets depend heavily on last-minute cancellations, tatkal quota releases, and RAC movement that are inherently unpredictable even minutes before chart preparation.

Including these volatile predictions would artificially skew accuracy numbers in either direction and wouldn't give you a meaningful measure of our model's reliability. By focusing on Safe and Risky predictions, we show you how well our system performs when it is confident enough to give a clear recommendation.

Accuracy is updated after every chart preparation. When Indian Railways prepares the reservation chart (usually 4 hours before departure), our automated system checks the final PNR status within minutes. The accuracy metrics on this page reflect the latest verified data in near real-time.

During peak travel seasons (festivals, holidays), dozens of charts are prepared daily across different trains, so you may see accuracy numbers change multiple times per day as new verifications come in.

RailTC uses multiple data sources to generate predictions: official IRCTC PNR status data, historical confirmation trends for specific routes and trains, seasonal patterns, quota utilization data, and waitlist movement velocity. Our model analyses these factors together to estimate confirmation probability.

For verification, we rely exclusively on official Indian Railways PNR status data obtained after chart preparation. This ensures our accuracy metrics are based on authoritative, tamper-proof outcomes rather than user-reported results.

Yes! The "Recent Verifications" section above shows the latest individual prediction outcomes. Each entry displays the train name, journey date, our prediction (Safe, Risky, or Medium), and the actual outcome (Confirmed, WL, or RAC). A green checkmark means the prediction was correct; a red cross means it was incorrect.

Medium predictions are shown with an amber indicator and are labeled "Not Counted" since they do not factor into accuracy calculations. You can scroll through recent results to see how our model performs across different trains and routes.