Sitemap

Explanation of my Visualisations

This is a living document and further detail will be added over time

5 min readFeb 7, 2017

--

A number of these visualisations has been inspired and influenced by the work of other especially Ben Mayhew. He runs Experimental 3–6–1 and produces visualisations for various European leagues.

A-League Visualisation

Attendances

Press enter or click to view image in full size
Press enter or click to view image in full size

A-League Player Visualisations

Goals Per 90

How many goals players have scored per 90 minutes of playing time. In the example below Santalab is known to come on as a substitute and tends to score in the last 25–30 minutes of matches.

Press enter or click to view image in full size

A-League Team Visualisations

Shot Dominance

Press enter or click to view image in full size

Defensive Effectiveness

Press enter or click to view image in full size

Attacking Effectiveness

Press enter or click to view image in full size

Good vs. Lucky

Press enter or click to view image in full size

Goal Difference vs. Shots on Target Difference

Defensive vs. Attacking Total Shot Ratio

Visual Cann Table

Press enter or click to view image in full size

Permutations

Press enter or click to view image in full size

Scored First, Didn’t Lose

Simple track of how many times clubs scored first and either won or drew the match. So in the example below Western Sydney Wanderers have a streak of 21 games where they’ve scored first and went on to not lose the game.

Press enter or click to view image in full size

Time Spent

Tracking of the percentage of match time club have spent in certain positions: winning, losing, drawing. In the example below Sydney FC have spent 5.8% of their entire match time in a losing position. This equates to approximately 5 minutes per 90 minutes of football.

Press enter or click to view image in full size

ELO Probabilities

Part of the calculation of ELO is a pre-match probability of each club winning. This visualisation is simply the calculation and visualisation of this. See the glossary for further information about ELO.

Press enter or click to view image in full size

Round Predictions

These are the predicted match outcomes that my model spits out. It’s not a particularly complex model and is therefore not particularly accurate. There is intent to improve it however that requires further time investment.

Press enter or click to view image in full size

ELO Ratings

Relatively standard Football ELO Ratings (see Glossary) with the change over the previous week of games.

Press enter or click to view image in full size

Glossary

If I’ve missed something let me know and I’ll add it. Disclaimer: not all data sources are the same since definitions can be somewhat subjective, ie was a shot on target or not needs to fit within a certain criteria however the categorisation is, in most cases, still done by a human.

Shot

All shots taken including on target, blocked, missed etc

Shot On Target

Attempts on goal that would have gone into the net or would have had there not been intervention by the goalkeeper or defender.

Shot Off Target

Any attempt on goal where the ball is wide of the target or hits the woodwork.

ELO

The Elo rating system is a method for calculating the relative skill levels of players in competitor-versus-competitor games such as chess. — Wikipedia

The ELO rating system has been adapted for football taking into consideration size of victory, home field advantage and match importance.

My rating been created specifically for an A-League specifically using it’s own K (weight index) value and HFA (Home Field Advantage) value.

Shots on Target Ratio (SoTR)

The ratio of shots on target for vs total shots on target for and against, ie SoT For / (SoT For + SoT Against)

Total Shots

The total number of shots attempted in a match, ie Shots For + Shots Against.

Total Shot Ratio (TSR)

Ratio of shots for vs the total number of shots in the game, ie Shots For / (Shots For + Shots Against)

PDO

Scoring % + Save %

Per 90

Match statistics are challenging to compare individuals directly given not everyone plays an entire 90 minutes of football each week. Factoring in the amount of playing time allows a closer comparison of how much impact each player has regardless of playing time.

StatsBomb has a great explanation of Per 90.

Frequently Asked Questions

Where do you get your data from?

The first rule of scraping football data is: You don’t talk about scraping football data.

Result based data is generally from Ultimate A-League. They’ve kindly allowed me access to their data and it is credited accordingly. If you’d like to share A-League data with me I’m more than happy to use it and credit you as a source.

What Software do you use?

I’ve a technical background dealing with all kinds of data so my toolset sits at the more technical end.

  • For data storage Postgres
  • Scripting done using Python
  • Visualisations using Tableau, however I’m testing ggplot and some other similar libraries with Python to incorporate a lot more automation.

How do I contact you?

The easiest way is via @rovingrob on Twitter.

Who else is doing cool stuff with football data?

--

--

Rob Scriva
Rob Scriva

Written by Rob Scriva

Data and viz guy. Singlehandedly destroying people’s hopes of a prosperous @aleague | @AdelaideUnited

No responses yet