SlickBet ⚽ screens soccer matches via the Livescore API, scores each fixture with a 12-factor statistical model (form, venue, xG, odds, and more), and ranks betting opportunities — including safer double-chance picks.

Same SlickML spirit: prototype fast 🏎, keep axes clear, and measure with backtests 🔎.

fixtures  ×  model  ×  filters  ×  backtest  ×  export

• 📅 Multi-day screening — next N days of fixtures
• 📊 12-factor model — form, position, home, H2H, odds, goals, venue form, defense, momentum, match stats, xG, reliability
• 🎲 Double chance — 1X / X2 / 12 with confidence grades
• 🌍 League coverage — major & minor Europe, Asia, Americas
• 📈 Backtesting — historical accuracy + API response cache
• 🔧 Hyperparameter tuning — search weights on cached data
• 📄 PDF / JSON export — shareable reports and CLI scripting

Install Python >=3.10,<3.14 and uv, then:

git clone https://github.com/slickml/slick-bet.git
cd slick-bet
uv sync --locked --all-extras --all-groups

Task runner is Poe the Poet (same idea as slick-tune). Install once:

uv tool install poethepoet
poe greet

Set Livescore API credentials:

export LIVESCORE_API_KEY="your_api_key"
export LIVESCORE_API_SECRET="your_api_secret"

# Screen all leagues for tomorrow (top 20)
poe run-all --days=1 --top=20

# Major European leagues only
poe run-major

# Backtest all leagues for 1 week
poe backtest-all-global --weeks=1

from slickbet import BettingScreener, ScreenerConfig

config = ScreenerConfig(min_probability=0.60, all_leagues=True)
screener = BettingScreener(config=config)
result = screener.screen_days(3)

for bet in result.get_top_k(10):
    best_dc, prob = bet.double_chance.best_double_chance
    print(f"{bet.match.home_team.name} vs {bet.match.away_team.name}: {best_dc} @ {prob:.1%}")

API cache (fast reruns / offline tune):

# Populate cache
slickbet backtest-all --weeks 12 --cache-dir data/api_cache

# Offline rerun
slickbet backtest-all --weeks 12 --cache-dir data/api_cache --cache-only

# Tune weights
slickbet tune --cache-dir data/api_cache --trials 20

slickbet                          # tomorrow
slickbet --days 5 --top 10
slickbet --major-only
slickbet --asia-only
slickbet --americas-only
slickbet --all-leagues
slickbet --league 2               # competition ID
slickbet --min-prob 0.60
slickbet --json
slickbet --pdf my_report.pdf

slickbet backtest --competition 2 --weeks 4
slickbet backtest-all --include-asia --include-americas --weeks 4
slickbet competitions --country England
slickbet tune --cache-dir data/api_cache

Use slickbet competitions --country <name> to discover more IDs.

flowchart TB
  match[Match + stats / odds / form]

  subgraph factors [12 weighted factors]
    direction LR
    venue["venue_form 20.6%"]
    momentum["momentum 12.0%"]
    defense["defense 10.4%"]
    home["home 9.5%"]
    reliability["reliability 8.7%"]
    goals["goals 8.6%"]
    matchStats["match_stats 7.1%"]
    h2h["h2h 6.1%"]
    odds["odds 5.5%"]
    position["position 4.7%"]
    xg["xg 3.7%"]
    form["form 3.3%"]
  end

  composite[Weighted composite score]
  probs[Home / away win probabilities]
  drawRisk[Draw risk]
  pred[BetPrediction]
  dc["Double chance: 1X · X2 · 12"]
  grade[Confidence grade]

  match --> factors
  factors --> composite
  composite --> probs
  composite --> drawRisk
  probs --> pred
  drawRisk --> pred
  pred --> dc
  pred --> grade

Tuned 12-factor weights (BettingModel.WEIGHTS in src/slickbet/model.py):

• 1X — home or draw · X2 — away or draw · 12 — no draw
• 🔥 HIGH VALUE · ✅ GOOD BET · 👍 DECENT · ⚠️ RISKY

from slickbet import Backtester, format_backtest_report

results = Backtester().run(competition_id="313", weeks=4)
print(format_backtest_report(results))
print(f"Excl. draws: {results.accuracy_excluding_draws:.1%}")
print(f"Best DC: {results.best_double_chance_accuracy:.1%}")

slick-bet/
├── assets/
│   └── logo.png
├── src/slickbet/
│   ├── __init__.py
│   ├── api.py          # Livescore client + domain models
│   ├── model.py        # 12-factor betting model
│   ├── screener.py     # Screening + formatting
│   ├── backtest.py     # Historical evaluation
│   ├── cache.py        # File-based API cache
│   ├── tune.py         # Weight search
│   ├── cli.py          # `slickbet` entrypoint
│   └── pdf_export.py   # ReportLab PDFs
├── tests/
├── .github/workflows/  # ci.yml + cd.yml only
├── .coveragerc
├── ruff.toml
├── mypy.ini
├── pytest.ini
├── pyproject.toml
└── README.md

For educational and entertainment purposes only.

• Past performance does not guarantee future results
• Sports betting involves risk of financial loss
• Predict responsibly and within your means
• Check local laws in your jurisdiction

Join our Slack to talk with the SlickML team.

MIT — see LICENSE.