Routster LogoRoutster Docs
Open Routster

WinSplits & Field Comparison

Advanced statistical field comparison, gaffling detection, and ranking.

While GPS data tracks where you went, WinSplits data tracks how you performed relative to the entire field. By linking a WinSplits URL to an event, Routster downloads the entire HTML results table, parses the split times, and unlocks advanced field analytics.

Field Ranks and Graphs

With WinSplits data, the Per-Leg view is augmented with a "Field Chip". For every leg, Routster calculates your exact position in the field, time behind the fastest split, and cumulative time behind the leader.

The Performance Graph is also unlocked, showing a classic "Boomerang" or "Spaghetti" chart. This chart graphs the cumulative time behind the leader for all runners over the course of the race, making it easy to spot lead changes, massive field-wide mistakes, and individual blow-ups.

Statistical Gaffling Detector

Routster features an incredibly sophisticated algorithm to automatically detect forked (gaffled) legs in relay events or special races.

Gaffling Detection (Technical Details)

Real gaffling produces leg-times that fall into two or more distinct clusters (each fork variant has its own typical time). However, a regular individual race with a few massive mistakes can also create long-tail outliers, which used to false-trigger simpler detectors.

Routster solves this using a strict statistical heuristic:

  1. High Dispersion: The leg must have a Coefficient of Variation (CV) > 0.55 and an Interquartile Range Ratio (Q3/Q1) > 1.9. (These thresholds are slightly lowered if the event is explicitly marked as a Relay).
  2. Log-Time Bimodality: The engine takes the natural logarithm of all times to reduce the impact of long-tail mistake outliers. It sorts them and looks for the largest adjacent gap.
  3. Cluster Density: The gap must be ≥ 35% of the total spread, AND at least 25% of the runners must sit on each side of the gap.

If a leg passes this test, Routster marks it as Gaffled. In the Per-Leg view, it handles gaffling via a spanRange, dynamically merging consecutive course legs (e.g. 4-5-6) into a single visual leg so it matches the aggregated WinSplits data chip for that section.

Sverigelistan Points Estimator

For Swedish users, Routster includes a live Sverigelistan Points estimator. It calculates estimated ranking points exactly matching the official SOFT rules:

  • Pm: Average points of the 3 best-ranked runners who start in the class.
  • Tm: Average time of the 3 best times in the results (capped at 1.10x the winning time).
  • Kk: (K + Pm) / Tm, where K = 60 for women, 75 for men.
  • Tb: Tm - (Pm / Kk).
  • Final Points: (Your Time - Tb) * Kk. (Capped at a maximum of 300 points).

If fewer than 3 ranked runners start, or fewer than 3 runners finish, the estimator gracefully aborts for that class.