RED HALF LABLiverpool football data
Open REDS
Try a questionGuess the next matchRulesProgressOpen source and dataDevlog and columns
Open REDS

AN INDEPENDENT LIVERPOOL DATA PROJECT

We use match data to answer what Liverpool fans argue about every week.

Red Half Lab is our independent Liverpool project, with no connection to the club. Two things to try: Give your take on a terrace debate, then guess how the next match will unfold.

Try a questionOpen REDSGitHub · coming soon

TWO WAYS IN · START HERE

01Your terrace verdictREDS · Try a question

Start with a fan’s criticism. Give your verdict, then see what REDS says.

Try a question ↗

Zoom into one match

02Every 15 minutesProject B · In progress

We split a match into six periods to see when things happen. Guess the form, cards and style, then reveal the local experiment’s inference.

Guess the next match ↗

TRY A QUESTION

A terrace debate. What’s your take?

Give your verdict, then explore REDS’ answer and evidence.

  1. 1Read
  2. 2Your call
  3. 3Compare
From the stands · Q1Player verdict

Evidence through 11/10

Is Wirtz really invisible in defence, too slow on the ball, short of line-breaking passes, unable to shoot and easy to push off the ball?

Where this came from

We collected this from comment sections: 4 comments on 3 videos from 3 channels, between 16 Sep and 27 Sep 2026. It only represents those comments, not every fan.

6 claims, about 30 seconds

There's more than one complaint in that line

Make your call on all 6, then we'll show you what REDS says.

  • 01Invisible in defence
  • 02Too slow on the ball
  • 03Short of forward passes
  • 04Can't shoot · Off target
  • 05Can't shoot · Shot quality
  • 06Easy to push off the ball
  • Short of line-breaking passesWe can't answer this one yet

PLAY IT · BEFORE KICK-OFF

Your call on form, cards and style

Three-match form. One match’s cards and style. Choose first, then reveal the data’s inference.

MATCHDAY / 02

Liverpool/Next match

Local experimentPre-match inference · awaiting kick-off
MATCH PREVIEW

Reading the inference for the next match…

Have your guess ready. The match preview is loading.

HOW WE WORK

Our rules

The same rules guide our player assessments and pre-match forecasts. We follow them for every question and conclusion.

  1. 01

    Learned from matches

    Every call comes from past matches.

  2. 02

    Only what's known before kick-off

    When we predict, we only use what was known before kick-off. What happens in the match is used for checking.

  3. 03

    Tested on later matches

    We learn from earlier matches and test on later ones. Each prediction has to beat the simplest guess. When it doesn't, we say so.

  4. 04

    New signings too

    If a new signing has only played a few games, we still predict him the same way.

  5. 05

    Set before kick-off, checked after

    Predictions don't change once the match starts. We check every item afterwards and keep the misses on the page.

  6. 06

    Under 70%, too early to call

    If we're less than 70% sure, we say it's too early to call and explain what's missing.

PROGRESS

Where we are

Every step is logged here by date. We only list what's done.

See the full devlog
  1. Tracking demoOpen event and tracking data connectedBoth open datasets are connected, and we've started tidying the code for release.
  2. REDSThe Manchester City match is checked, and we got it wrongBefore kick-off we gave Liverpool a 59% chance of winning. They lost 0–3. The check went out as usual, and none of the misses were removed.
  3. Project BProject B becomes its own projectWe started learning the order of the six spells in a match and the changes from match to match together.
  4. REDSGoalkeepers get predictions, and REDS gets its namePer-90 goalkeeper predictions for the next appearance are live. Saves don't beat guessing the average yet, so they're marked as a rough guide.
  5. Project BProject B starts testingWe're starting with cards. Nothing goes live until we've reviewed it ourselves.
  6. REDSAutomatic post-match checksAs soon as a match ends, we check the predictions automatically and learn from the match for next time.
  7. REDSFirst prediction locked before kick-off: Manchester City at homeAn hour and a half before kick-off we locked every prediction for the match, and none of them changed.
  8. Whole projectEvery call has to be learned from matchesFrom this day the rule was set: every call is learned from matches and has to beat guessing the average. Where it doesn't, we say so on the page.
  9. Whole projectEvery number on the site is now realWe removed every placeholder. Every number on the site is either our prediction or a result from a match already played.

OPEN SOURCE & DATA

Open source and data

Code we can share goes on GitHub. For every open dataset we use, we credit the source and licence the way the publisher asks.

MORE TO COME

More open source

Code or write-ups for our other projects will be added here once we decide to release them.

CITE THIS PROJECT

If you cite our work

Red Half Lab (2026). Red Half Lab: Liverpool football data. We'll add the repository link here once it's up.

DEVLOG & COLUMNS

Devlog and columns

Column · Reading the numbersWhat a 41% chance of winning actually meansA worked example of three possible results, rounding and the difference between the likeliest outcome and a likely win.Column · From the standsTurning a terrace argument into a question we can checkFive familiar criticisms, five separate questions. Define the action, comparison and time period before attaching a verdict to a player.Column · Behind the scenesWhy a prediction needs a record from before kick-offWhat a pre-match lock proves, what a post-match check measures, and why learning from a new match must leave the old record intact.