About sport.dev
sport.dev is a free sports fixtures and prediction site. Machine-learning models rate every team and player, publish a win probability for each upcoming match, and every prediction is scored publicly against the final result.
How it works
- We collect the data Results, fixtures, team lists, player stats, injuries and weather are gathered continuously from official sources across every league we cover.
- Models rate teams and players Rating systems track the strength of every team and player over time, and machine-learning models combine them with form, team lists, rest and venue to estimate each side's chance of winning.
- You get a probability, then a scorecard Every upcoming match shows a win probability. When it finishes, the pick is marked right or wrong, and the full record is kept on the Stats page.
What you can use it for
Following your sport
Subscribe to leagues and teams for one clean schedule: kickoff times in your timezone, broadcasters for your country, live scores and results.
Fantasy competitions
Team lists as they drop, late changes, predicted try scorers, expected fantasy points and the casualty ward, all in one place.
Team and player analysis
Ladders, head-to-head history, form and match stats make it a quick tool for studying teams, players and matchups.
An independent signal
Model probabilities are built without bookmaker odds, so they're a second opinion to weigh against the market. They are information, not betting advice.
What's coming
Our data now runs through its own API. Next: an MCP server and agent skills so AI assistants can use it, documented API access and datasets for research, and tools for coaches and performance staff. Read about research and data access →
FAQ
Do I need an account?
Not to browse. Fixtures, results and predictions are free for everyone. Signing in with Google, also free, unlocks each match's team lists, stats and facts, and syncs your subscriptions across devices.
How accurate are the predictions?
It depends on the sport and on how confident the model is: more confident picks win more often. The Stats page has the full record, lifetime and for the last 30 days, by league, sport and confidence.
Which sports have predictions?
Rugby League, Rugby Union, Football (soccer), American Football, Basketball, Baseball and Ice Hockey have model predictions now. Other sports show fixtures and results while their models are trained.
Do the models use bookmaker odds?
No sportsbook data or market price data is used in any models. Market price data can get predictions reasonably close, however the aim of this platform is to compose enough signals purely from data to train improving models for each sport.
Is this betting advice?
No. Predictions are statistical estimates for information and analysis. No outcome is guaranteed.
Credits
- League and Team logos are only used for reference purposes, no products or services are being sold using those protected images
- Live score data from ESPN
- Weather data from Open-Meteo.com
Prediction accuracy
Every prediction is scored against the final result, and nothing is removed or restated. A pick counts as correct only when the predicted outcome happens.
Research and data
Models built to pick match winners learn far more than a win probability. Here is what our data and models can do beyond the app, what's available now, and what we're building for researchers, teams and developers.
The modelling foundation
Skill ratings (Elo and OpenSkill-style, updated after every result) estimate the strength of every team and player. Gradient-boosted models combine those ratings with form, team lists, rest, travel and venue into a calibrated win probability: when the model says 70%, it should win about 70% of the time, and the Stats page holds it to that.
Available now
The site runs on a read-only JSON API: fixtures and results, predictions, team lists, ladders, injuries and accuracy stats for every league we cover. It isn't documented or versioned for outside use yet, so if you'd like to build on it, get in touch first.
AI agents: MCP server and skills
An MCP server and ready-made agent skills, so assistants like Claude can query fixtures, predictions, team lists, injuries and stats directly, with sources and timestamps.
- Ask an agent for this weekend's fixtures with win probabilities and late team changes
- Pull a league's ladder, form and accuracy record into an analysis notebook
- Set up alerts for team-list changes and injury news on the teams you follow
API access and datasets
Documented, versioned API access with keys and rate limits, plus bulk historical exports for universities, schools and analysts.
- Match-level history with predictions, results and team lists
- Model outputs and calibration data for forecasting research
- Teaching datasets for statistics, data science and sport-science courses
Tools for coaches and performance staff
Team dashboards built on opposition-adjusted ratings and per-match stats, separating process from results the way expected-goals models do.
- Player performance tracking and development curves
- Weekly reviews against a league baseline, with early-warning flags
- Opposition scouting, matchup planning and recruitment shortlists
Injury, load and scheduling models
The shift in win probability when a player is ruled out is a data-driven estimate of their value. The same machinery can model injury risk, return to play, rest and travel.
- Injury impact, and each club's injury burden across a season
- Return-to-play curves by injury type, position and age
- The cost of short turnarounds, congested schedules and long-haul travel
Working with us
We're a small team and the platform is young, but the pipelines and models run every day and the accuracy record is public. If you're a researcher, teacher, club analyst, sports scientist or company that wants to use the data or models, or you have data that would sharpen them, get in touch through this form.
Changelog
What's new on sport.dev: features, fixes and model updates.
Known issues
- Right now every model is designed to work optimally for regular season play roughly after the first 10% of matches have been played.
- Start of season lacks enough recent data for player form and roles/combinations.
- End of season/finals for each sport is also played differently and may be hit or miss. Model weight adjustments for end of season/finals should be ready soon.
- Football Draw prediction weights needs a lot more work.
- Every league in a sport relies on global models for that sport rather than being league specific. Some leagues like French Ligue 1 or Brazilian Serie A do not fit my current euro models very well due to a few things: financial and squad disparities, travel volatility and heavy squad rotations. I'll take a deeper look at this soon when building out league specific steps for football.
- Rugby League (NRL) had model degradation from round 13 to round 23 from an upstream model stall that forward filled stale data into several primary feature inputs. This was difficult to detect because it started during State of Origin (a mid season tournament using the same players) so I had assumed the model drift was from player fatigue or injuries. This has been fixed but it did lower the overall accuracy % for this season.
Terms of use
The short version of what you agree to by using sport.dev.
What sport.dev provides
sport.dev is a free sports fixtures, results and predictions site, provided as-is. Features and coverage can change or be removed at any time.
Predictions are information, not advice
Predictions are statistical estimates. They are not betting, financial or professional advice, and no outcome is guaranteed. If you gamble, do so legally and responsibly, and only with money you can afford to lose.
Accuracy of data
Data comes from public and official sources and is updated automatically. Fixtures, scores and team lists can be late or wrong, so check the official competition source for anything that matters.
Your account
An account is optional and you're responsible for what happens on it. Accounts used to abuse the service may be suspended.
Intellectual property
League and team names and logos belong to their owners and are shown for identification only. Screenshots and quotes are fine with attribution; republishing our data or predictions in bulk needs permission.
Automated access
Scraping or heavy automated use of the site or its API isn't allowed without permission. Documented API access is on the way; see the Research page.
Liability
To the extent the law allows, sport.dev is not liable for any loss arising from use of the site or from relying on its data or predictions.
Changes
These terms may be updated. Material changes will be noted in the changelog.
Last updated: 1 October 2026
Privacy
We collect as little as possible. Here is what we store and why.
Browsing without an account
No account is needed. Your settings (language, theme, timezone, country) and subscriptions are kept in your browser's storage, not on our servers. We count page views with Umami, a cookie-free analytics service that doesn't identify you. We don't run advertising trackers or sell data.
If you sign in
Signing in with Google gives us your email address, name and profile picture. We store those with your subscriptions and settings so they sync across your devices, and a session cookie keeps you signed in.
Timezone and country
Your timezone and country are worked out in your browser to show local kickoff times and broadcasters. You can change both in settings.
Infrastructure
The site is hosted on Cloudflare, which processes requests and keeps short-lived logs for security and performance. Live scores (ESPN) and weather (Open-Meteo) are fetched by our servers, not by your browser.
Deleting your data
Clearing your browser's site data removes your local settings. To delete your account and everything stored with it, ask us through the feedback form.
Last updated: 1 October 2026