MLS Playoff Chances

A live playoff-probability dashboard for MLS's Eastern and Western Conferences. Standings, results, and the remaining schedule are pulled from ESPN on every request (server-cached for 5 minutes); three independent models then project each team's rest-of-season outlook.

Models

• Points Pace — extrapolates each team's current home/away points-per-game across its remaining fixtures into a projected final total, with an approximate playoff probability from a normal approximation around the conference's projected cutoff.
• Monte Carlo — simulates the rest of the season 8,000 times using a current-form power rating (from points-per-game and goal differential) and each team's own home-win rate (shrunk toward the league average for small samples).
• Elo Rating — replays every completed match chronologically (with a home-field adjustment and a margin-of-victory multiplier) into a per-team Elo rating, then runs the same Monte Carlo engine using Elo-implied win probabilities.

Draw probability in both simulations is derived from the league's actual draw rate this season, tapering as the rating gap between two teams widens.

The playoff line shown in the table (top 9 per conference, top 7 direct to Round One, 8–9 to the Wild Card round) reflects today's actual standings, not a projection — only the probability/projected-points columns change with the selected model.

Data source

ESPN's public soccer API (standings + full-season scoreboard). No API key required. This is an unofficial endpoint; if ESPN changes its response shape, src/lib/espn.ts is the only file that needs updating.

Development

npm install
npm run dev

Open http://localhost:3000.

Project layout

• src/lib/espn.ts — fetches standings + full schedule from ESPN
• src/lib/elo.ts, src/lib/strength.ts — the two rating systems
• src/lib/probability.ts — rating differential → win/draw/loss probabilities
• src/lib/simulate.ts — the Monte Carlo engine (shared by the Monte Carlo and Elo models)
• src/lib/pace.ts — the points-pace projection engine
• src/lib/compute.ts — orchestrates the above into one payload
• src/app/api/data/route.ts — serves the payload, cached in-memory for 5 minutes
• src/components/ — the dashboard UI (React + Recharts)

Deploy to Vercel

npx vercel

Follow the CLI prompts to log in and link the project, then npx vercel --prod to ship. No environment variables are required.