What goes into the prediction
Each prediction is calculated using only relevant, comparable data scoped to a defined race context. This prevents cross-category distortion and keeps comparisons fair.
Race context
Predictions are always scoped to:
- Division (e.g. Open, Pro)
- Gender
- Age group (where applicable)
This ensures athletes are compared against the right peer group.
Athlete history
For each athlete in the start list, Race Pulse evaluates performance history in comparable categories, including consistency and trends over time.
- Past results in comparable divisions
- Consistency across multiple events
- Performance trends over time
Recency and decay
Not all races count equally. Recent performances carry more influence, while older results are progressively down-weighted to reflect current form.
- Recent races have greater influence
- Older races fade over time
- Long gaps between races reduce confidence
Race Pulse prefers honest uncertainty over false precision.
What the model produces
Predicted finish order
Athletes are ranked by expected overall race time, not reputation or best-ever performance. This is designed for race-day decision making.
Predicted time
Each athlete receives a predicted finish time based on their historical performance distribution and event-level variability. Times are estimates, not guarantees.
Confidence indicators
When the available data is limited, inconsistent, or dated, confidence is reduced and surfaced in the output. This is intentional.
Why start lists matter
Traditional rankings answer: “How good is this athlete overall?”
Race Pulse answers: “How is this athlete likely to perform against these people, in this race context?”
- Field strength varies by event
- A mid-pack athlete in one race can podium in another
- Rankings do not account for who actually shows up
What this does not do
Race Pulse does not:
- Predict injuries, illness, or race-day execution
- Account for weather, heat, or course anomalies (yet)
- Replace coaching judgement or race strategy
It is a decision-support tool, not an oracle.
How to use it well
- Use realistic times where known
- Include only athletes likely to start
- Treat small predicted gaps as interchangeable outcomes
The closer the predicted times, the more volatile the order.