xg-model

Train an XGBoost xG model to predict hockey shot goal probabilities.

Updated Aug 23, 2026

Search Tags:#sports-analytics #model-training #machine-learning #feature-engineering #xg #hockey #expected-goals

Ready to run

npx skills add https://github.com/ronniepinnell/benchsight --skill xg-model

💬 Prompt for ChatGPT / Claude Chat▼

Please help me install this Agent Skill.
Skill: xg-model
Source: https://github.com/ronniepinnell/benchsight/tree/main/.claude/skills/xg-model
Command: npx skills add https://github.com/ronniepinnell/benchsight --skill xg-model

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Automate hockey shot analysis by predicting goal probabilities using an xG model trained on historical shot data, enabling faster insights and better decision-making for coaches and analysts.

Core Features & Use Cases

Feature-rich inputs including distance, angle, shot_type, coordinates, and contextual features for robust xG scoring.
Use Case: generate per-shot xG predictions in ETL pipelines and compare against actual goals in dashboards for performance monitoring.
Use Case: iterate on feature engineering and model versions to drive improvements across seasons.

Quick Start

Train and evaluate the xG model using your hockey shot dataset to generate production-ready predictions.

Frequently Asked Questions about xg-model

High-intent search queries and answers about installing and using this skill.

FAQPage Schema

How do I train an expected goals model for hockey shots using XGBoost?▼

Expected goals models for hockey use shot features like distance, angle, and shot_type to predict goal probability. XGBoost handles the non-linear relationships between these shot characteristics and actual scoring outcomes effectively.

What features do I need to calculate xG for hockey shot analysis?▼

Hockey xG calculation requires features like distance, angle, shot_type, strength_state, period, score_diff, goalie_movement, and coordinates. These contextual inputs capture the spatial and situational factors affecting shot success rates.

Can I generate per-shot xG predictions within an ETL pipeline?▼

Yes, you can integrate per-shot xG predictions into ETL pipelines. The model outputs goal probabilities for each shot record, which can be compared against actual goals in downstream dashboards for performance monitoring and analysis.

How do I evaluate xG model performance across multiple hockey seasons?▼

Evaluate xG model performance by comparing predicted goal probabilities against actual goals in dashboards. Iterating on feature engineering and model versions helps drive predictive improvements and maintain accuracy across multiple hockey seasons.

Does this xG model include contextual features like strength state and goalie movement?▼

Yes, the xG model incorporates contextual features like strength_state and goalie_movement alongside distance, angle, shot_type, period, and score_diff. These contextual inputs capture the situational factors affecting shot success rates.

What is the best way to deploy a machine learning xG model for hockey analytics?▼

Deploy an xG model for hockey analytics by training XGBoost on historical shot data with features like distance, angle, and strength_state. The model provides a production deployment path for generating goal probability predictions in analytics workflows.