← Back to Injury Watch

How the IRS works

The Injury Risk Score is a community-powered metric that turns anonymous player reports into a reliable, comparable signal. Here's the full methodology.

What is the IRS?

The Injury Risk Score (IRS) is a number from 0 to 100 assigned to each padel racket, reflecting the injury risk reported by the players who use it. A score of 0 means no reports. A score of 100 means extremely high reported injury risk across many players.

The IRS is not a measure of a racket's quality, playability, or brand reputation — only the injury signal from community reports.

Step 1 — Severity weighting

Each report carries a severity weight. Not all injuries are equal — a severe injury that ended your season is weighted more heavily than mild post-match soreness.

SeverityWeightDefinition
Minor1Soreness, no missed play
Moderate2Modified play or 1–4 weeks off
Severe3Stopped play, 4+ weeks off

The average severity score for a racket is computed as: Σ(severity_weight) / n_reports

Step 2 — Bayesian shrinkage

A racket with 2 severe reports looks worse than one with 200 moderate reports — but it shouldn't, because the sample is tiny. We use Bayesian shrinkage to blend each racket's score toward the global average when the report count is low.

bayesian_severity = (Σ weights + K × global_avg_severity) / (n + K) K = 10 (prior weight — controls how fast trust builds with volume)

At n=0, the score equals the global mean. At n=30+, the racket's own data dominates.

Step 3 — Volume factor

A racket with 5× more reports than average carries more signal. The volume factor scales the score based on how many reports a racket has relative to the global average across all rackets.

volume_factor = min(n_reports / global_avg_reports, 3.0)

Capped at 3× to prevent a single popular-but-injured racket from dominating the leaderboard.

Final IRS formula

IRS = min(bayesian_severity × volume_factor × 25, 100)

The scale factor (25) maps the 0–4 severity range to the 0–100 output range. The score is capped at 100.

Spike detection

A spike is flagged when a racket receives significantly more reports in the last 7 days than its historical baseline suggests. We use a Poisson model:

expected_recent = (reports_23_to_30_days_ago / 23) × 7 spike_threshold = expected_recent + 2 × √(max(expected_recent, 1)) spike = recent_7d > spike_threshold AND recent_7d ≥ 3

The √ term is the Poisson standard deviation. The threshold corresponds roughly to the 95th percentile upper bound — so spikes are meaningful, not noise.

Anti-abuse

To prevent a single person from spamming reports for or against a racket:

  • One report per IP address per racket per 24 hours
  • IP hashes are stored (not raw IPs) for rate limiting — no PII retained
  • Reports are flagged if submitted within seconds of page load

IRS risk bands

ScoreLabelMeaning
0–9MinimalVery few or very mild reports
10–34LowSome reports, typically minor
35–59ModerateMeaningful injury signal — worth noting
60–100High riskStrong and severe injury signal

Limitations

The IRS is community-sourced data, not a clinical study. It reflects what players report — not causation. A high score may indicate a racket's design is problematic for certain play styles, but it could also reflect the profile of players attracted to that racket (e.g. aggressive smash players reporting elbow issues).

The IRS improves in reliability as report volume grows. Treat low-volume scores (under 10 reports) as indicative only.

The IRS is not medical advice. Always consult a physiotherapist for injury assessment and management.

View leaderboard → Submit a report →