Sportra Intelligence

Data pipeline

Sportra Intelligence runs in four steps for every statistics tab:

1. Gather — Sportra loads fixture, team, and player datasets when you open the page (match detail, club season, or player season). Statistics views load their current dataset when requested; live matchday surfaces use a separate background-refresh pipeline.
2. Process — values are cleaned (percentages, ranges, missing fields) so numbers can be compared side by side.
3. Aggregate — season rows, home/away splits, minute buckets, and team-vs-team pairs are rolled up into the blocks shown in the UI.
4. Present — base totals appear in tables and bars; Sportra Intelligence adds derived metrics, scores, and short narratives on top.

Team and player pages use the league and season you select. Match pages use that fixture only.

Overview

The same intelligence layer powers the Statistics, Players, and Predictions tabs on matches, plus Statistics on team and player profiles. Where the dataset includes a field, we show it; where it does not, we omit the derived metric rather than guess.

Labels such as archetypes, impact leaders, and comparison winners are Sportra definitions for reading the game on Sportra — they are not official competition statistics.

Team statistics — Northbridge Athletic (fictional sample)

Where: Team page → Statistics tab.

Season totals are grouped into blocks (fixture record, goals, minutes, formations, and more). Sportra Intelligence turns those blocks into charts, 0–100 scores, traits, and an archetype summary. Full season tables remain available under Full statistics tables when present.

Season data blocks

Season intelligence (0–100)

• Home fortress — home attack share vs an even split, plus home goals vs home conceded
• Away attack — away goal share vs an even split
• Defensive reliability — clean sheet rate blended with goals conceded per game
• Match chaos — share of high-scoring games from threshold data
• Form momentum — recent form string weighted toward wins
• Tactical flexibility — distinct formations used relative to matches played

Traits and archetype

Traits are rule-based tags (for example, a late-finisher tag when a high share of goals come after 76′). The archetype headline (such as home fortress) summarizes the season profile for quick reading.

Player statistics — Leo Castell (fictional sample)

Where: Player page → Statistics tab.

Season output is merged into one profile: hero metrics, card-style attributes, derived rates, then a full stat breakdown. Non-penalty goals are shown as total goals minus penalties scored when both exist.

Stat groups

Appearances, shooting, goals, passing, tackling, duels, dribbling, fouls, cards, and penalties — combined for the selected league and season. If multiple rows exist for the same competition, counting stats are summed and the highest rating is kept.

Card attributes (scouting profile)

Each rating is scaled 0–99 from this season's stats. Higher is stronger in that area relative to elite benchmarks for the metric mix — not an absolute world ranking.

Attack

Goal threat — goals per 90 plus on-target shooting share.

Vision

Chance creation — assists per 90 and key passes per match.

Control

Ball carrying — dribble success rate and attempts per 90.

Precision

Execution quality — shot accuracy, or passing accuracy when shots are scarce.

Pressure

Drawing contact — fouls won per 90 (players who attract challenges).

Defense

Defensive actions — tackles per 90 and duel win rate.

Engine

Usage — average minutes per appearance (stamina / role).

Discipline

Card record — fewer yellow/red cards mean a higher score.

Flair

Take-on volume — dribble attempts per 90 (direct, skillful play).

Derived categories

Some tiles include an intensity bar: value compared to a fixed Sportra benchmark, capped at 100 — a visual cue, not a league percentile.

Match statistics — Northbridge Athletic vs Riverside Rovers (fictional sample)

Where: Match page → Statistics tab.

Home and away team totals are paired stat-by-stat, grouped into six themes, then extended with Sportra Intelligence derived rows (efficiency, pressure, gameplay ratings).

Comparison groups

Derived match metrics

• Shot accuracy — shots on target ÷ total shots
• xG per shot — expected goals ÷ total shots
• Inside box ratio — inside-box shots ÷ total shots
• Attack pressure — (in-box shots × 2) + corners + shots on target + blocks faced
• Gameplay ratings — aggression, control, siege pressure, clinical index, defensive compactness, chaos factor

Comparison bars highlight the leading side; fouls and cards favour the lower value.

Match player statistics

Where: Match page → Players tab.

Each player line from the match dataset is processed into a box score, derived rates, an impact score, optional badges, and team-level duel bars that sum both squads.

Per player

Minutes, rating, position, goals, assists, saves, cards — plus derived passing, duel, and discipline metrics for that appearance.

Impact score

• Outfield: rating, goals, assists, key passes, shots on target, tackles, interceptions, saves, and card penalties combined into one ranking score
• Goalkeeper: rating, saves, goals conceded, and passing volume

Badges mark standouts (best rating, playmaker, wall, and similar). Captain is shown when flagged in the data. Team totals aggregate all players on each side for head-to-head bars.

Predictions

Where: Match page → Predictions tab.

Win, draw and away shares are an estimate. Each team's expected goals average its season goals scored with the opponent's season goals conceded. Early-season rates are pulled towards a typical 1.35 goals, as if each team had already played five typical matches, and the home side gets a 10% edge. A Poisson model then turns the two rates into outcome shares. The estimate needs at least three league matches for both teams; with fewer we show no shares. It uses pre-match data only, ignores injuries and line-ups, and is a statistical guide, not betting advice.

What the tab includes

• Estimated outcome shares (home, draw, away) and expected goals
• Side-by-side comparison (form, attack, defence, goals, head-to-head, overall)
• Season and streak metrics per team
• Recent head-to-head fixtures when available
• Charts — outcome donut, profile radar, scoreline bars, season anchors, signal grids

Sample data only (Sample League · 2019–20 (illustrative)). Live pages use the same views with Sportra data for your selected fixture, team, player, league, and season.

Calculation rules

Cleaning values

• Percentages are stored as plain numbers (58% → 58)
• Ranges such as 3–5 use the midpoint
• Missing values stay out of derived maths

Rates

• Per 90: (value ÷ minutes) × 90 when minutes > 0
• Per match: value ÷ appearances when appearances > 0
• Ratios: shot accuracy, conversion, duel win %, pass accuracy

Match comparisons

Higher is usually better; fouls and cards favour the lower total. Ties show as neutral.

Team of the Week

Team of the Week uses the stored match rating on players who appeared in that round. It is not a season reputation score and it is not a weighted formula.

• A player needs at least 60 minutes in a finished match in the round.
• The position must be a known goalkeeper, defender, midfielder, or forward.
• If a player appears twice, the higher match rating is kept.
• The XI is 4-3-3. At most three players from one club are selected while the side is filled.
• Ties break by rating, then goals, then assists, then player id.
• Player of the Week is the highest-rated selected player.

Limits

• If a stat is missing from the dataset for that page, related derived metrics are not shown.
• Penalty shootout scores on match headers follow the PEN status and phase breakdown used elsewhere on Sportra.
• Match, team, and player detail pages show the dataset from when the page was loaded. To see updated statistics, reload the page — we do not poll upstream on a fixed interval for those tabs.
• Predictions reflect pre-match data only.

Questions: Contact · Data licensing: Data attribution