skills.rest
najicham
najichamCOMMUNITY
Explore all skills by this author →
GitHub

hit-rate-analysis

Aggregate NBA prop prediction hit rates across confidence levels, models, and time windows.

1|Updated May 24, 2025
Search Tags:#NBA#confidence#analyze#edge#model-comparison#hit-rate#prop-predictions
One-click install
npx skills add https://github.com/najicham/nba-stats-scraper --skill hit-rate-analysis
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: hit-rate-analysis
Source: https://github.com/najicham/nba-stats-scraper/tree/main/.claude/skills/hit-rate-analysis
Command: npx skills add https://github.com/najicham/nba-stats-scraper --skill hit-rate-analysis

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This Skill makes it easy to quantify how often prop predictions hit the actual outcomes across different confidence bands, edge levels, player tiers, and time frames, reducing guesswork in model evaluation.

Core Features & Use Cases

  • Hit rate by edge: measure how edge thickness correlates with accuracy.
  • Confidence-aware evaluation: compare performance across confidence buckets to diagnose calibration.
  • Model comparison & drift detection: track multiple models (e.g., catboost variants, ensembles) over time to spot drift and inform model selection.
  • Use Case: A data analyst wants to decide which model to put into production for daily prop bets based on recent hit-rate stability and edge-driven profitability.

Quick Start

Use the hit-rate-analysis skill to compare the latest catboost_v9 and ensemble_v1_1 hit rates over the last 30 days.

Frequently Asked Questions about hit-rate-analysis

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

FAQPage Schema
How do I analyze NBA prop prediction hit rates by confidence and edge levels?▼

You can compare NBA prediction model performance by tracking hit rates across multiple models like catboost variants and ensembles over time. This detects performance drift and identifies the most stable model for production deployment.

How do I detect prediction model drift for NBA prop bets over time?▼

No, you do not need external dependencies, but you must have access to the nba_predictions dot prediction_accuracy data and related feature tables. These internal datasets provide the required ground truth outcomes for hit rate analysis.

What is the best way to evaluate edge thickness against actual prop bet accuracy?▼

The best way to evaluate edge thickness is by grouping predictions into edge buckets and calculating the hit rate for each band. This measures how strongly edge thickness correlates with actual accuracy to optimize your filter configurations.

Can I use this approach to generate recommended filter configurations for trading decisions?▼

Yes, analyzing hit rates across confidence, edge, and player tier dimensions generates recommended filter configurations. These configurations optimize model selection strategies by isolating the parameter combinations with the highest historical profitability.

Related Skills

Explore More Data & Analytics Skills→
LudiInformatioLudiInformatio

sports-data-model-architect

Audit and design sports analytics data models with migration-safe SQL and Python.

Community
Advanced
kayzaakayzaa
5

Prop Firm Manager

Track prop firm challenges and enforce trading rules with K.I.T. integration.

Community
Advanced
najichamnajicham

model-experiment

Train and evaluate NBA player prop regression and breakout classifier models.

Community
Intermediate
HermeticOrmusHermeticOrmus

Model Evaluation Patterns

Generate classification metrics, calibration analysis, and fairness evaluations for machine learning models.

Community
Advanced