The Engine Room exists for one reason: so that when someone in a meeting asks "where does that number come from?" you can open this page and show them, line by line, exactly how it was derived.

Every equation is shown. Every data source is cited. Every assumption is stated explicitly and flagged where it is a proxy rather than a measured value. Nothing is hidden and nothing is invented.

The female WSL burn rate was derived entirely from StatsBomb open data. No male benchmarks were used in the calculation. The derivation followed this sequence:

1. Load player appearance timelines. For each player-season combination, extract all match dates from wsl_scored_players.csv (8,950 records, 14 clubs, 2018–21).
2. Filter to regular starters. Only players with ≥5 appearances AND ≥40% appearance rate (appearances ÷ total team matches that season) are included. This removes fringe players whose absences are tactical, not injury-related.
3. Identify selection gaps. For each regular starter, scan consecutive match dates. Any gap ≥21 days between appearances, where the team continued to play, is flagged as a probable injury absence.
4. Classify gap severity. Gaps are classified by duration: ≥84 days = Severe (ACL/fracture), ≥42 days = Major (muscle/ligament), ≥21 days = Moderate (strain/tear).
5. Calculate total days lost. Sum all gap_days across all 243 identified gaps.
6. Calculate total player-days available. For each regular starter, sum the days between first and last appearance in the season.
7. Derive burn rate. Apply the equation below.

The Injury Double Tap is not a metaphor — it is a financial accounting framework. Every injury triggers two separate cost categories simultaneously: the ghost wage already sunk, and the cascade of new costs that follow. Together they represent 14.4% of the WSL wage pool every season.

Traditional threshold models are brittle. A player with ACWR 1.49 is "safe" and one at 1.51 is "danger" despite being physiologically identical. Fuzzy logic handles this gracefully by assigning partial membership across multiple zones simultaneously. A player at 1.4 ACWR is 70% "optimal" and 30% "high risk" both true at once.

The Mamdani inference system fires multiple rules simultaneously, weights their outputs by membership strength, and defuzzifies to a single crisp readiness score via the centroid method. This mirrors how an experienced physio actually thinks not binary, but weighted and contextual.

The simulator below is the extended version of the FALI engine. It adds real telemetry inputs HRV, resting heart rate, sleep efficiency, deep/REM sleep, RPE alongside the standard load and wellness inputs. It also incorporates hormonal phase modifiers specific to female athletes.

When a club connects real wearable or GPS data, this is the engine that runs. In proxy mode (no wearables), the wellness scores act as validated surrogates for the telemetry inputs, consistent with the SoccerMon methodology.

Complete run order for reproducibility. All scripts located in the GENDERGAP folder.

The Mamdani engine produces a readiness score using centroids expressed on a 0–10 scale (Green = 8.6, Amber = 6.0, Red = 3.2). The PITCH interface and all reporting displays this on a 0–100 scale for clarity. The conversion is a direct multiplication by 10.

The PITCH daily wellness check-in is built on the Oslo Sports Trauma Research Centre Questionnaire on Health Problems (OSTRC-H2), developed by Clarsen, Myklebust and Bahr at the Norwegian School of Sport Sciences (British Journal of Sports Medicine, 2013, updated 2020). It is the validated standard for athlete health surveillance in elite sport globally — used across Olympic programmes, professional football, rugby, and handball worldwide.

The four questions were designed through a specific clinical process. Traditional injury surveillance only captures time-loss injuries — players who cannot train at all. This misses the much larger population of athletes competing and training while impaired. The OSTRC team worked with elite athletes and medical staff to identify the four domains that most reliably capture the full spectrum of health impact: participation, training volume, performance, and symptoms. Each domain has a non-linear weighting to reflect that the jump from "some difficulty" to "unable to participate" is not a linear progression — it represents a qualitatively different health state.

The specific response option weights (Q1/Q4: 0, 8, 17, 25 · Q2/Q3: 0, 6, 13, 19, 25) were derived empirically from the validation study across 313 elite athletes. They are not arbitrary — they reflect the relative severity of each response option as rated by the athletes themselves. The asymmetric weighting (larger jumps at the top of each scale) captures the exponential nature of injury impact.

The four intervention savings in the PITCH briefing are derived from the 81-incident/season baseline at £17,613 per incident. Each prevention rate represents the estimated proportion of incidents that intervention would realistically prevent. These rates are Signal Analytics modelling assumptions — they are not published research findings. They are stated explicitly here.