Soccer Analytics Course

Lesson 1 • Setting Up an Analytics Workflow

Outlines a repeatable process from question formulation to insight delivery. Connects analytical rigor to practical club decision-making.

Lesson 2 • Data Sources and Collection Methods

Surveys primary data providers, tracking systems, and event logs. Students understand data provenance and quality trade-offs before analysis begins.

Lesson 3 • What Soccer Analytics Means

Defines analytics in a soccer context and distinguishes it from traditional scouting. Establishes shared vocabulary used throughout the course.

Lesson 1 • Working with Soccer Data Formats

Covers JSON, CSV, and XML structures common in event and tracking feeds. Prepares students to parse and load real provider data programmatically.

Lesson 2 • Data Cleaning and Validation

Addresses missing values, duplicate records, and inconsistent labels in soccer datasets. Clean data is the prerequisite for all reliable downstream metrics.

Lesson 3 • Feature Engineering for Soccer

Transforms raw events into analytically useful features such as zones, sequences, and time deltas. Engineered features power the models built in later chapters.

Lesson 4 • Merging and Linking Datasets

Teaches joining event data with tracking, lineup, and contextual tables. Accurate merges prevent analytical errors caused by mismatched identifiers.

Lesson 1 • Communicating Findings Visually

Applies design principles—color, hierarchy, annotation—to analytical outputs. Effective visual communication ensures insights reach non-technical audiences.

Lesson 2 • Pitch Visualization Fundamentals

Builds static pitch maps for shots, passes, and defensive actions using mplsoccer or similar libraries. Spatial context transforms raw coordinates into tactical insight.

Lesson 3 • Descriptive Statistics for Soccer Data

Applies mean, median, variance, and distribution analysis to player and team metrics. Descriptive summaries form the baseline before inferential work begins.

Lesson 4 • Radar and Comparison Charts

Designs radar charts and bar comparisons for multi-metric player profiling. These formats are standard in scouting reports and stakeholder presentations.

Lesson 1 • Feature Selection for xG Models

Identifies shot distance, angle, body part, and assist type as core predictors. Feature choice directly determines model accuracy and interpretability.

Lesson 2 • Post-Shot and On-Target xG

Extends basic xG with shot placement and goalkeeper positioning data. Post-shot models provide finer resolution on save difficulty and keeper performance.

Lesson 3 • Training and Validating xG Models

Applies logistic regression and gradient boosting to shot data with proper cross-validation. Validation prevents overfitting and ensures generalizability across seasons.

Lesson 4 • Interpreting and Applying xG Outputs

Translates model outputs into finishing skill estimates and team attack quality. Connects xG to recruitment and tactical decisions made by coaching staff.

Lesson 1 • Tactical Style Clustering

Groups teams by possession and pressing metrics using unsupervised clustering. Style clusters enable like-for-like comparisons and opponent preparation.

Lesson 2 • Building Passing Networks

Constructs directed weighted graphs from pass event data using NetworkX or igraph. Network structure reveals team hierarchy and preferred circulation routes.

Lesson 3 • Possession Sequences and Chains

Segments match data into possession chains and analyzes their outcomes. Chain-level analysis links ball circulation patterns to shot creation.

Lesson 4 • Passing Metrics Beyond Completion Rate

Introduces progressive passes, passes into the final third, and expected threat added. Richer passing metrics capture intent and danger beyond simple accuracy.

Lesson 1 • Age Curves and Development Trajectories

Models how player performance evolves with age using longitudinal data. Age curves inform buy-vs.-loan decisions and long-term squad planning.

Lesson 2 • Building a Recruitment Shortlist

Integrates metric filters, similarity scores, and contextual flags into a ranked shortlist. The shortlist format bridges analytics output and sporting director workflow.

Lesson 3 • Similarity Search and Player Comparisons

Uses distance metrics and nearest-neighbor search to find statistically similar players. Similarity models power replacement and alternative target identification.

Lesson 4 • Composite Player Rating Models

Combines multiple metrics into weighted composite scores for overall player value. Composite models reduce dimensionality and support rapid shortlisting.

Lesson 1 • Off-Ball Movement Analysis

Measures runs, space creation, and positioning quality independent of ball contact. Off-ball metrics reveal player contributions invisible in event data alone.

Lesson 2 • Integrating Tracking with Event Data

Aligns tracking frames with event timestamps to enrich shots, passes, and duels. Combined datasets enable context-aware models unavailable from either source alone.

Lesson 3 • Team Shape and Spatial Metrics

Quantifies team width, depth, and centroid movement from synchronized tracking frames. Spatial team metrics operationalize tactical concepts like compactness and stretch.

Lesson 4 • Introduction to Tracking Data

Explains optical and GPS tracking systems, coordinate frames, and frame rates. Understanding data provenance is essential before any physical metric calculation.

Lesson 1 • Expected Threat and Possession Value

Builds xT and VAEP frameworks to assign value to every on-ball action. Possession value models enable holistic player and team performance attribution.

Lesson 2 • Translating Analytics into Club Strategy

Structures analytical findings into strategic recommendations for coaching and management. Bridges technical output and organizational decision-making at the club level.

Lesson 3 • Match Outcome Prediction Models

Builds pre-match and in-match win probability models using team-level features. Outcome models support betting-market benchmarking and in-game tactical adjustments.

Lesson 4 • Set-Piece Analytics

Quantifies corner, free-kick, and throw-in routines using spatial and outcome data. Set pieces account for a significant share of goals and deserve dedicated modeling.