NBA Moneyline

Simulates a simple NBA betting strategy against historical moneyline odds: pick a team, season, prediction direction (bet on win vs. bet on loss), and wager size, and see what the season result would have been.

Status: no longer updated with new seasons. The web app and the dashboard both still work, with the 2016-17 through 2025-26 regular seasons (11,979 games).

Shipped two ways:

	Link	Stack
Tableau dashboard	public.tableau.com/…/NBAMoneyline	Tableau Public — parameter-driven calculated fields, a running-total table calc, a favorite/underdog breakdown
Web app	nba-moneyline.vercel.app	Node/Express + PostgreSQL + vanilla JS, deployed on Vercel

Both are fed by the same Python/Selenium data pipeline (data/). The rest of this README covers the web app; see Tableau dashboard below for that version.

Overview

The web app lets a user choose a team, season, prediction direction (bet on win vs bet on loss), and wager size, then computes what the season result would have been. It combines:

• A Node/Express API for querying outcomes
• PostgreSQL for game and odds data
• Vanilla JavaScript frontend for fast interaction
• Python/Selenium scraping pipeline that collected the data, one season at a time

Highlights

• SQL-driven profitability and ROI calculations
• Two focused API endpoints: result summary and top bets
• Shared Express route factory to keep backend handlers DRY
• Season-by-season scraper workflow with verification and migration steps
• Production deployment pattern that mirrors local app behavior

Tech Stack

• Node.js + Express
• PostgreSQL (pg)
• Vanilla JavaScript, HTML, CSS
• Python (Selenium, BeautifulSoup, psycopg2) for scraping/migration
• Vercel deployment
• Tableau Public (for the dashboard version — see below)

Project Structure

app/
  routes/
    resultSummary.js          # /api/result-summary
    topBets.js                # /api/top-bets
  utils/
    createQueryRoute.js       # Shared API route factory
    parseSQL.js               # Loads SQL files from disk
    dbConfig.js               # Shared PG pool
  queries/
    resultSummary.sql
    topBets.sql

public/
  index.html
  js/
    api/                      # Fetch + response adapters
    events/                   # Submit/interaction orchestration
    view/                     # Result rendering

data/
  main.py                     # The whole pipeline: extract, transform, load
  extract/                    # Step 1: OddsPortal (odds) + basketball-reference (schedule), then checks
  transform/                  # Step 2: one dated record per game
  load/                       # Step 3: load into production, update frontend
  util/                       # Shared data model, constants, and output formatting
  tests/                      # pytest suite, mirroring the folders above
  sample_queries.sql          # SQL for inspecting the database by hand
  README.md                   # Setup, adding a season, the database, troubleshooting

tableau/
  NBA Moneyline.twbx          # Published Tableau Public workbook
  games.csv                   # Data export the workbook reads

API Endpoints

• POST /api/result-summary
  • Input: season_start_year, team, prediction, wager
  • Output: aggregated outcomes grouped by favorite/underdog and result
• POST /api/top-bets
  • Input: same payload
  • Output: highest-earning individual bets for the chosen strategy

Getting Started

1. Install Node dependencies

Requires Node 24 (pinned in .nvmrc; fnm switches to it automatically).

npm install

2. Configure environment

Create .env.development.local in the project root:

POSTGRES_URL=postgres://username:password@host/database

3. Run locally

npm run dev

App runs at http://localhost:3000.

Scripts

npm start      # Production-style local start
npm run dev    # Development with nodemon
npm run sync-env

Data Pipeline

The data/ directory contains the ingestion flow that collected each season:

1. Extract one season: odds and results from OddsPortal, and the official schedule (dates, home/away) from basketball-reference - then check it: game counts against that season's expected pattern, and every team's opponents against the official schedule
2. Transform: match every game to the official schedule, turning two scraped rows per game into one record with its real date and home team
3. Load into PostgreSQL (teams, one row per game in games, and a team_games view the app reads) in one all-or-nothing transaction, only with all checks passing or an explicit override; the tables' constraints reject bad rows too - then update the web app's season options

See data/README.md for setup, how a season is added, and troubleshooting.

Tableau dashboard

tableau/NBA Moneyline.twbx is the published workbook (Tableau Public) — a data-analyst-portfolio rebuild of the same simulator.

以上为 README 的开头部分。其余内容请阅读原文。 查看完整 README ↗