korb 0.2.7

A CLI toolkit for DBB basketball league analysis

🏀 korb
A CLI toolkit for DBB basketball league analysis

---

What is this?

A zero-dependency Python CLI that parses HTML from the DBB (Deutscher Basketball Bund) legacy JSP platform and gives you:

• 📊 Standings — full league table with points, differentials, averages
• 🏀 Team drill-down — game-by-game results, sparklines, quality metrics
• 📋 Ergebnisse — all completed game results with score differentials, optional team filter
• 📅 Schedule — with back-to-back detection, filtering, pending games
• 🔮 Predictions — efficiency-model-based final standings forecast
• 🥇 Top N — quick leaderboard with ASCII bar chart
• 📥 Download — fetch fresh HTML data directly from basketball-bund.net
• 🤖 Skills — built-in AI skill prompts for team analysis & league prediction
• 🔧 JSON output — pipe-friendly --json flag for all commands

---

Finding your Liga ID

The --ligaid value comes from the liga_id parameter in the DBB league URL:

https://www.basketball-bund.net/index.jsp?Action=103&liga_id=51187
                                                           ^^^^^
                                                           this is your liga ID

Copy the number after liga_id= and pass it to --ligaid.

---

Quick Start

# Install with uv
uv sync

# Download league data
uv run korb --ligaid 51187 download

# View standings
uv run korb --ligaid 51187 standings

# Download fresh data + predict in one go
uv run korb --ligaid 51187 --download predict

---

Installation

# Clone and install
git clone https://github.com/malvavisc0/korb && cd korb
uv sync

# Install with dev tools (black, isort)
uv sync --group dev

After uv sync, the korb CLI is available inside the virtual environment. Use uv run korb to invoke it, or activate the venv with source .venv/bin/activate and run korb directly.

---

Commands

download — Fetch HTML data

# Download results + schedule for a league
uv run korb --ligaid 12345 download

# Refresh all previously downloaded leagues at once
uv run korb download --all

Saves ergebnisse.html and spielplan.html into files/<ligaid>/.

The --all / -a flag scans the files/ directory for all previously downloaded league IDs and re-downloads each one. Useful for bulk-refreshing data across multiple leagues.

standings — League table

uv run korb --ligaid 12345 standings

 #  Team                GP   W   L   D     PF    PA   Diff  Pts   Avg PF  Avg PA
----------------------------------------------------------------------------------
 1  Thunder Academy     12   9   3   0   1248   847   +401   18    104.0    70.6
 2  Riverside Hawks     11   9   2   0    952   668   +284   18     86.5    60.7
 3  Metro Wolves        11   8   3   0    813   694   +119   16     73.9    63.1
...

ergebnisse — All game results

# All completed results
uv run korb --ligaid 12345 ergebnisse

# Filter by team (partial, case-insensitive)
uv run korb --ligaid 12345 ergebnisse --team "Hawks"

# JSON output
uv run korb --json --ligaid 12345 ergebnisse

======================================================================
  Ergebnisse — MFR U12 mix Bezirksliga Nord
======================================================================

   Datum            Heim              Gast              Ergebnis    Diff
-----------------------------------------------------------------------
1  15.01.26 10:00   Team Alpha        Team Beta           79:75      +4
2  15.01.26 12:00   Team Gamma        Team Delta          60:60       0
3  22.01.26 15:00   Team Beta         Team Alpha          82:70     +12

team — Deep dive on a single team

# Basic results
uv run korb --ligaid 12345 team "Thunder"

# With bar chart + quality metrics for last 5 games
uv run korb --ligaid 12345 team "Thunder" --bars --last-k 5 --metrics

schedule — Game calendar

# All upcoming games
uv run korb --ligaid 12345 schedule --pending

# Filter by team + mark back-to-back ⚡ fixtures
uv run korb --ligaid 12345 schedule --team "Hawks" --pending --b2b

# Include cancelled games
uv run korb --ligaid 12345 schedule --all

predict — Forecast final standings

uv run korb --ligaid 12345 predict

Uses a multiplicative efficiency model with recency weighting, recent form blending, home advantage, and back-to-back fatigue modelling.

top — Quick leaderboard

uv run korb --ligaid 12345 top -n 5

skill — Print AI skill prompts

# List available skills
uv run korb skill --list

# Print a specific skill prompt
uv run korb skill analysis
uv run korb skill prediction

Ships two built-in skills: analysis (team deep-dive) and prediction (league top-N forecast).

--download — Fetch fresh data before any command

# Download + show standings in one step
uv run korb --ligaid 12345 --download standings

# Download + predict
uv run korb --ligaid 12345 -d predict

--json — Machine-readable output

Add --json before any subcommand to get JSON instead of tables:

uv run korb --json --ligaid 12345 standings
uv run korb --json --ligaid 12345 team "Hawks"
uv run korb --json --ligaid 12345 ergebnisse
uv run korb --json --ligaid 12345 ergebnisse --team "Hawks"
uv run korb --json --ligaid 12345 schedule --pending
uv run korb --json --ligaid 12345 predict
uv run korb --json --ligaid 12345 top -n 3

---

CLI Reference

$ uv run korb --help

usage: korb [-h] [--version] [--results RESULTS] [--schedule SCHEDULE]
            [--json] [--ligaid LIGAID] [--download]
            {standings,team,ergebnisse,schedule,predict,top,download,skill} ...

Basketball league analysis tools

positional arguments:
  {standings,team,ergebnisse,schedule,predict,top,download,skill}
    standings           Display league standings
    team                Display results for a team
    ergebnisse          Display game results
    schedule            Display game schedule
    predict             Predict final standings
    top                 Show top teams from standings
    download            Download results & schedule HTML
    skill               Print a skill prompt or list available skills

options:
  -h, --help            show this help message and exit
  --version, -V         show program's version number and exit
  --results, -r RESULTS HTML results file path
  --schedule, -s SCHEDULE
                        HTML schedule file path
  --json                Output as JSON instead of formatted tables
  --ligaid, -l LIGAID  Liga ID (e.g. 51491)
  --download, -d        Download latest data before running the command

standings --help

usage: korb standings [-h]

options:
  -h, --help  show this help message and exit

team --help

usage: korb team [-h] [--bars] [--last-k LAST_K] [--metrics] name

positional arguments:
  name             Team name (e.g., 'Hawks')

options:
  -h, --help       show this help message and exit
  --bars, -b       Show point differential bar chart
  --last-k LAST_K  Analyze only the most recent K games
  --metrics        Show win-rate + margin quality metrics

schedule --help

usage: korb schedule [-h] [--all] [--pending] [--team TEAM] [--b2b]

options:
  -h, --help       show this help message and exit
  --all, -a        Show cancelled games
  --pending, -p    Show only pending games
  --team, -t TEAM  Filter by team name (partial match)
  --b2b            Mark back-to-back fixtures (≤36h)

ergebnisse --help

usage: korb ergebnisse [-h] [--team TEAM]

options:
  -h, --help       show this help message and exit
  --team, -t TEAM  Filter by team name (partial match)

predict --help

usage: korb predict [-h]

options:
  -h, --help  show this help message and exit

top --help

usage: korb top [-h] [-n N]

options:
  -h, --help  show this help message and exit
  -n N        How many teams to show (default: 3)

download --help

usage: korb download [-h] [--all]

options:
  -h, --help  show this help message and exit
  --all, -a   Refresh all previously downloaded leagues

---

How Predictions Work

The predict command estimates final standings using a multiplicative efficiency model:

Factor	How it works
Offensive rating	Team's scored points vs. league average (>1.0 = above avg)
Defensive rating	Points allowed vs. league average (>1.0 = worse defense)
Recency weighting	60-day half-life — recent games count more
Recent form	Last 5 games blended at 30% weight into ratings
Home advantage	3% scoring boost applied symmetrically
B2B fatigue	≤36h between games → 5% offense/defense penalty
New teams	<3 games → ratings blended toward league average
No draws	Ties broken by home advantage (basketball has OT)
No double-counting	Already-played games excluded from schedule scan

---

🧠 Skills

The skills/ directory contains reusable AI skill definitions — structured prompts that can be loaded by an AI assistant (e.g. Roo/Cline) to run analysis workflows using the CLI.

Skill	Output	Description
SKILL_TEAM_ANALYSIS.md	Paragraph	Short team summary (position, identity, form, outlook) — ready for a webpage card
SKILL_LEAGUE_TOP_N_ANALYSIS.md	Table + paragraph	Predicted final standings table with a brief explanation

Both skills accept a LANGUAGE parameter (en/de/es) and return output directly (no file saved). To use a skill, point your AI assistant at the markdown file or load it as a skill definition.

---

Project Structure

├── korb/
│   ├── __init__.py      # Package marker
│   ├── __main__.py      # CLI entry point & download command
│   ├── core.py          # Shared models, HTML parsing, utilities
│   ├── ergebnisse.py    # Game results viewer & filter
│   ├── predict.py       # Multiplicative efficiency prediction model
│   ├── schedule.py      # HTML schedule parser & filters
│   ├── standings.py     # Standings calculator
│   └── team.py          # Team results viewer & metrics
├── skills/              # AI skill definitions for automated analysis
├── files/               # Downloaded HTML data (git-ignored)
├── tests/               # Tests
├── pyproject.toml       # Project metadata & uv config
└── README.md

Requirements: Python 3.10+ · uv · No runtime dependencies

Note: Downloaded HTML files and --ligaid paths resolve relative to your current working directory (files/<ligaid>/). Run korb from the project root or pass explicit --results / --schedule paths.

Dev tools

# Format code
uv run black korb/

# Sort imports
uv run isort korb/