Skip to analysis
MatchLensFOOTBALL INTELLIGENCE
Local workspace

A NEW PERSPECTIVE ON PREMIER LEAGUE FOOTBALL

See the game.
Understand
the why.

Every match has a score. Every moment has a story. MatchLens connects football data and AI to help fans see more — and analysts go deeper.

Explore the experience Discover the vision

One shared game. Two perspectives. More understanding.

FOOTBALL, IN FOCUS.
PUMA STELLAR Match Premier League Reactivity football
FROM EVENTS
TO UNDERSTANDING
THE NEXT QUESTIONWhat made the difference?
A DIFFERENT READ OF THE SAME GAME ↗

A CONCEPT FOR FOOTBALL’S NEXT CHAPTER

Premier LeagueMicrosoft

Premier League ambition. Microsoft technology. AVATARZ imagination.

Independent project presentation · Evidence-led Microsoft AI

THE VISION, IN MOTION

A different read.
A closer look.

A short cinematic introduction to MatchLens. Then explore the working experience, from the moment on the pitch to the evidence behind it.

CONCEPT TEASER · 10 SECONDSOpen video ↗

MORE VALUE FROM EVERY MATCH

The game goes deeper.
So can the experience.

The ambition: help the Premier League extend its spirit of innovation into the way football is understood, experienced and shared.

01 / CONNECT

Bring fans closer.

Turn complex match data into clear stories. Help every fan understand the chances, turning points and choices behind the result.

MORE ENGAGEMENT. MORE UNDERSTANDING.
02 / REVEAL

Give analysis depth.

Explore the event behind the insight. Compare chance quality, passing and player contributions with evidence you can trace.

CLEARER INSIGHTS. STRONGER CONTEXT.
03 / EXPLORE

Make decisions tangible.

Step into the Decision Lab. Use fictional replay scenarios to explore passing options, space and risk from different perspectives.

MORE CURIOSITY. NEW WAYS TO LEARN.

ONE DATASET. TWO WAYS IN.

Built for the way
you see football.

The same facts, with the right depth for each audience.

FOR THE FAN 01

The story
behind the score.

Key moments. Clear explanations. A match that makes more sense.

Enter Fan view FOR THE ANALYST 02

The evidence
behind the insight.

Event exploration. Research models. Match reports and a personal notebook.

Enter Analyst view

AI WITH A PURPOSE

Football first.
Technology in support.

Start with a question about the game. Find the evidence. Explain it clearly. Microsoft AI turns calculated facts into accessible explanations, with clear sources and limits.

WORKING TODAY

A real foundation
you can explore.

131historical matches
216k+recorded events
  • Two perspectives on shared match facts
  • Local xG & xPass research models
  • Evidence-linked statistical answers
  • Interactive synthetic Replay Lab
Explore the data & model limits Historical Wyscout data · Research prototype.
Statistics use a local question engine. Microsoft AI explanations activate when the cloud connector is available.
MicrosoftMICROSOFT ARCHITECTURE

Microsoft Foundry

Explain the evidence in Fan or Analyst style, using only the facts supplied by MatchLens.

Azure Functions

Recalculate each request, protect service access and check the AI response before it reaches the experience.

Azure Storage

Cache validated explanations and maintain a shared usage allowance for the demo.

Cloud features require a configured API. AI Search and additional agent workflows remain future extensions.

DESIGNED & BUILT BY AVATARZ

A shared passion.
A new perspective.

Bringing product, creativity and technology together around the game.

Simon RamesIulian VarzaruEliton AlmeidaMathues Souza

THE NEXT MOVE IS YOURS.

See what you’ve
been missing.

Open MatchLens
MatchLensBY AVATARZ

Project presentation · Historical data & synthetic decision scenarios.
Premier League and Microsoft marks used with permission for this presentation, as confirmed by the project owner.

Data & methodology ↗

WORKSPACE / MATCH CENTRE

The game, with perspective.

One dataset. Two ways to understand it.

HISTORICAL MATCH
Recorded events
Loading the local match library…
BEHIND THE NUMBERS

Real calculations. Honest limits.

These xG and xPass values are demonstration estimates.

Training, temperature calibration and evaluation use a fictional generator only. Neither model is validated or calibrated on real matches. No professional or proprietary model is used.

xPass

Pass completion estimate based on distance, forward progress, static lane blockers, receiver clearance and opponents near the passer.

xG if shot

A hypothetical immediate foot shot from the receiver’s decision-time position: goal distance, opening angle, shot-lane blockers and clearance. No actual shot is asserted.

Pass → shot

xPass × xG, assuming an immediate shot after a completed pass. Not a causal value or a complete possession model.

“xS” is ambiguous. MatchLens uses the explicit term xPass for pass completion.

For each model: 6,000 training examples, 1,500 calibration examples and 2,500 held-out test examples. Labels come from fictional probabilities defined in code. Replay outcomes never train these models.

Inputs, models, formulas, cutoff and missing data
The recorded result is reviewed after the action. Estimates always use the 6,000 ms snapshot, without the later pass resolution. Alternative replays are authored branches; they do not validate probabilities.
Definitions: xG — Hudl · xPass — Hudl Statsbomb. These references imply no affiliation or use of their models or data.


THE HACKATHON PROTOTYPE
From event to understanding.
Microsoft×Premier League
MatchLens combines a historical football event workspace with a separate synthetic Decision Lab. Fan and Analyst views use the same data. Research xG and xPass models run locally; match exploration, source-linked questions, imports, exports and browser-local notebooks are available without an external account.
Traceable match intelligence Running locallyHistorical xG / xPass Local research models2D sprites, articulated players & POV Original procedural reconstructionMicrosoft Azure / Foundry Optional cloud connector
Historical teams, players and results come from the attributed public Wyscout dataset. The separate Replay Lab remains fictional; its club nicknames and colours are cosmetic. No real match footage, likenesses or client assets are used. Microsoft and Premier League names identify the hackathon context only; no endorsement is implied. The logos identify the hackathon context only and do not imply endorsement of MatchLens.