Please note that Sunderland and Reading have only played 8 games and all other teams have played 9.
If you look at the table at the bottom of the page you will see that Arsenal have attempted the most final-third passes 1599(FTP) so far this season, have completed the most final-third passes (FTP C) and have the highest % completion of final-third passes (FTP C).
Fulham are the most accurate and in their 9 games they have averaged 1 goal per 43.57 final-third completed passes. (Goal/FTP C).
QPR are the least accurate with 1 goal for every 104.85 final-third passes completed.
With the data it is possible to draw a scatter diagram with a regression line drawn in to see if there is a relationship between final third passes completed and goals.
An R2 near 1 indicates that a regression line fits the data well. In this case we see that R2 is 0.452 so not a great result but not a bad result and we should not conclude that there is not correlation between final third passing completion and goals without looking deeper into the analysis.
What we can see is that teams above the line have scored less goals then expected which includes Arsenal, Liverpool, Everton and Man City.
All the teams that are below the line have scored more goals then expected which is even more true for Stoke, West Brom, Newcastle and Fulham.
Man UTD and Chelsea are right on the line and if every team was on the line of regression then we would be very excited and we would be saying that there is correlation between final third completed passes and goals.
Click on Graph to enlarge
I do not have all the answers but there are other variables we must consider.
- Does a team play the long ball like Stoke do? Teams who play long balls as their strategy will have a lower pass completion so the regression line is unable to identify this. Stoke score goals but do not spend as much time in the final third as some other teams.
- Not all teams have played the same quality of teams so the sample size is very small so possibly too early to draw conclusions. Aston Villa are having a poor season and are still to play Man UTD and Man City who you would expect to restrict their final third pass completion.
- Final third passing has many other variables which we have not considered such as the conditions of the pitch. Was the pitch muddy so that passing was difficult or was it windy for example.
- Another interesting variable that we cannot link into the data is how good the final third pass was in terms of the possibility of a goal. Was it for example a 3 on 1 position making a goal likely from the pass or was it a long ball into the box which is easily defended.
Sometimes it is very easy to ignore in terms of analysis and assume there is no validity without checking all the variables, but I cannot say if there is a correlation between final third passing or not at this stage.
What is clear is that data is readily available to have a look at which must be a good thing.
Final third passing is certainly an interesting topic and we can only hope that we will understand it better as more detailed data is made available to us. Fo now here is the data that I have collected from the 4-4-2 StatsZone iPhone App manually.
The final column represents final third passes per goal – you can click on the column headers to sort the data. Enjoy!
| Team | FTP | FTP C | % FTP C | Total Goals | Goal/FTP C |
|---|---|---|---|---|---|
| Arsenal | 1599 | 1259 | 78.73 | 14 | 89.92 |
| Man City | 1561 | 1188 | 76.1 | 18 | 66 |
| Man Utd | 1475 | 1161 | 78.71 | 24 | 48.37 |
| Chelsea | 1441 | 1067 | 74.04 | 21 | 50.8 |
| Everton | 1428 | 1015 | 71.07 | 17 | 59.7 |
| Liverpool | 1389 | 998 | 71.85 | 12 | 83.16 |
| Southamp | 1224 | 824 | 67.32 | 14 | 58.85 |
| West ham | 1190 | 733 | 61.59 | 13 | 56.38 |
| Norwich | 1155 | 697 | 60.34 | 7 | 99.57 |
| Fulham | 1124 | 828 | 73.66 | 19 | 43.57 |
| QPR | 1124 | 734 | 65.3 | 7 | 104.85 |
| Wigan | 1114 | 784 | 70.37 | 10 | 78.4 |
| Swans | 1106 | 790 | 71.42 | 14 | 56.42 |
| Villa | 1075 | 578 | 53.76 | 7 | 82.57 |
| Spurs | 994 | 743 | 74.74 | 17 | 43.7 |
| WBA | 982 | 610 | 62.11 | 13 | 46.92 |
| Reading | 972 | 534 | 54.93 | 11 | 48.54 |
| Sroke | 967 | 543 | 56.15 | 8 | 67.87 |
| Newcastle | 897 | 517 | 57.63 | 11 | 47 |
| Sunderland | 721 | 465 | 64.49 | 6 | 77.5 |
Please do leave any comments below.
- Excellent
- Informative
- Awesome
- Good Read
- ok
About the Author
Categories: Arsenal (NN), Aston Villa, Betting Tips, Chelsea, EPL Index Featured Article, EPL Index Statistical Comparisons, Everton, Fulham, Liverpool, Manchester City, Manchester Utd, Newcastle Utd, Norwich City, QPR, Reading, Southampton, Stoke City, Sunderland, Swansea City, Tottenham Hotspur, West Bromwich Albion, West Ham United, Wigan
Tags: EPL, epl opta stats, EPL Stats, Final Third Passing, Final Third Passing Analysis, Opta Stats, premier league, Premier League Stats
This article has had 6,836 Views
Whilst these footy stats are in their infancy, we need to nail down their definitions. Why is a goal kick that lands on Andy Carroll’s head, for example, a completed final third pass? It could also be a completed first third pass depending on how you define it. I think with a different definition of completed final third pass (that the pass must originate AND finish in the final third) you’d see more teams on your line of best fit.
I agree with you Paul and I have noticed for example that some sites will give a cross that the keeper catches as a shot on goal so it makes it very difficult to analyze data. Football stats are certainly in their infancy and I wonder how easy it will be to access them in the future? Most data is provided by bookmakers to trigger a bet and is not driven by data analysis.
I think Liverpool and Final third passing make a very interesting contribution to data analysis. in simple terms they have the goal to Final pass completion record of a team who will be fighting relegation. Against Newcastle they had 187 final third passes and 132 completed and all they could do was score once. Add the 23 shots and the 36 crosses and there is something going wrong at Liverpool. Less is more? 1638 total final third passes and 1194 completed so and just 13 goals so 1194/13=1 goal per 91.8 final third completed passes. This means that Liverpool should have scored around 25 goals this season so around 12 short of the data prediction. The most interesting point is that if you back Liverpool because they are doing more of everything then any other team then think how well they are doing it. The answer is not very well.
Anyone thinking of backing Liverpool to beat Chelsea?
Perhaps Liverpool could have a look at Danny Kedwell from Gillingham. Goal scoring machine this season.
interesting.^ I enjoyed reading your article and your comments are very
You then quoted another chap which I have copied below
When we take these individual match numbers of shots, accurate shots, and goals – of which there were 32,789, 10,396, and 2,954 across the three seasons we collected data for – and put them in relation to each other, it turns out that the odds of any one shot actually being on target was 32%, while the odds of an accurate shot finding the back of the net was similarly around 30 % (28% to be exact). Plenty of teams shoot enough to score, but very few of them consistently score.
Clearly, “normal” football isn’t always normally distributed. As a general rule, the more common an event on the pitch is, the more the distribution looks like a bell-shaped curve (graph the frequency of passes per match and you’ll see what I mean). This has important implications: using some of the most common statistical techniques to deal with these data may be problematic, standard (canned) versions of techniques like correlations and linear regression assume normally distributed data. The stuff we care about the most – goals – is the least “normal” of all the events above. But as importantly, think about what the picture above tells us: there is enormous slippage from one stage of the goal production process to the next. Understanding why and how this slippage occurs should be important questions for any budding analyst.
It brings everyone back to the point of why do Liverpool have more crosses, more shots on goal, more possession, more Final third passing and not score enough goals.
You touched on Shalke, I watch around 50 games a month and I often see the better side being beaten. Over the season this will average out so in the case of Shalke as you say it will not effect their ability to win for the rest of the season.
In terms of that chap talking about slippage and people who assume normally distributed data, I am aware of the variables involved such as the weather conditions, motivation, time of the goal, expectation of a goal given the current score, the fact that research shows that if the home team score first that expectation of an away goal is reduced by 10% for 10 mins…
It is easy to read the past research and agree or disagree with their conclusion.
What is clear Richard is football predictive models IN RUNNING are based on
1. Shots on target
2. Crosses
3.Weather conditions
4.Motivation
5. Red cards etc
6.expectation of goals given the current score
I feel slightly amused that my first article for EPL advises me that …”believes it may be on to something”
I was very clear about pointing out all the variables….
Kind regards
Jonny
For your information I am a paid football watcher and former professional gambler.
Sometimes it is very easy to ignore in terms of analysis and assume there is no validity without checking all the variables, but I cannot say if there is a correlation between final third passing or not at this stage.
lol
Richard in response to your friend as I cannot contact him directly.
I cannot reply to the chap on soccerbythenumbers- brilliant website so I will have to post here.
But the most fundamental lesson is often the hardest: before you go looking for fancy ways of analyzing your complex dataset, understand the nature of your variables; having a feel for their shape and nature in a large enough sample of observations is critical to extracting useful information.
I think the irony is that this where football models break down in terms of thinking that a sample size needs to be large enough. In football if you are betting the first thing you want to see is if both teams are motivated. This has no relationship to the previous games. They either are or they are not. Also Liverpool as is being pointed out time and time again to not fit the “football predictive Model” but it does not stop people from backing them every weak. Intelligent people who see that in essence Liverpool are not good at winning games against all grades of teams. Sometimes too much data can be a bad thing in terms of the P and L .
weak*fraudian slip
Another variable is the teams that play with two out and out wingers , not many final third passes completed by those teams such as Spurs and Manu .
Spurs not many FTP but Man UTD right up there.171 v Villa
Interesting read. You can see a moderate positive relationship, but the “sample” size could be questionable! It would be interesting to monitor this and maybe find a correlation coefficient for a larger data set to see if the relationship is statistically significant! I think Arsenal may be missing RVP lol
Hello Matt. I thought i recognized your hand writing. If we find correlation at the end of the season it will be too late to take advantage. lol.
Good point. With 20 games however, if it was significant, you can be “statistically” sure that 95% of the time there is a relationship (with a 15% margin for error) this season between final third passes completed and goals. Obviously you can never be sure what the next 18 games will bring though!
Matt. Poisson Distribution.
The controversial pronunciation of Poisson
Yes not the chicken.
Why not plot ftp/games rather than ftp to take account of the fact some teams have played less?