By Danny Pugsley

Welcome to a little project.

images__1_.0_standard_783.0

Both here at StatsBomb and over at my more regular home at Bitter and Blue there has been a lot of great work of late looking shots; areas such as placement, distance and conversion. My particular interest though has been to take a step back from this: to look at how shots are generated; what are the successful drivers behind getting a shot at goal and are there are any patterns that a side displays?

The germ of the idea, as, admittedly, a lot of the areas of analytics that spark my interest, was formulated with hockey with some great work having been done over the past couple of seasons (examples here and here) led by Eric Tulsky, who had a paper presented on this subject at last years MIT Sloan Conference.

So, what do we measure? Well, for a start all the data has to be manually recorded so as much as the possibility of watching endless Premier League games thrills me, for now we will have to stick to just Manchester City. (Incidentally, one of the successes of the venture in hockey is the way in which it has spread across the NHL with a number of teams now having this data recorded).

Essentially, every time City either make a final third entry or concede a final third entry this is recorded. We note the entry type (carry, header, turnover or short, medium or long pass), the individual making the entry and the outcome (dispossessed, turnover, corner won, fouled or shot (inside and outside the box) with the shot outcome (saved, wide, blocked or goal)) also noted. Once processed, this then produces the following tables on a weekly basis, with the following tables being City’s early season totals after the Newcastle and Cardiff games:

Team entry summary

Team Carry Turnover Header Short pass Med pass Long pass Total
City 21 2 0 62 28 3 116
Opposition 8 2 0 14 8 11 43

Outcome per entry type

City Enries Dispossed Turnover back out Corner won Fouled Total shots Shot in box Shot o/s box Blocked Saved Wide Goal Shots p/e Goals p/e
Carry 21 8 4 0 1 2 6 5 1 1 3 1 1 28.57 4.76
Turnover 2 0 0 0 0 1 1 1 0 0 0 0 1 50.00 50.00
Header 0 0 0 0 0 0 0 0 0 0 0 0 0 n/a n/a
Short Pass 62 12 21 10 5 1 13 11 2 4 2 6 1 20.97 1.61
Med pass 28 7 8 2 2 0 9 8 1 0 3 3 3 32.14 10.71
Long pass 3 1 2 0 0 0 0 0 0 0 0 0 0 0.00 0.00
Total 116 28 35 12 8 4 29 25 4 5 8 10 6 25.00 5.17
Opposition Entries Dispossed Turnover back out Corner won Fouled Total shots Shot in box Shot o/s box Blocked Saved Wide Goal Shots p/e Goals p/e
Carry 8 1 2 1 0 1 3 1 2 1 1 0 2 37.50 25.00
Turnover 2 1 0 0 0 1 0 0 0 0 0 0 0 0.00 0.00
Header 0 0 0 0 0 0 0 0 0 0 0 0 0 n/a n/a
Short Pass 14 7 3 0 1 0 3 1 2 1 1 1 0 21.43 0.00
Med pass 8 0 8 0 0 0 0 0 0 0 0 0 0 0.00 0.00
Long pass 11 2 6 1 0 0 2 2 0 0 1 0 1 18.18 9.09
Total 43 11 19 2 1 2 8 4 4 2 3 1 3 18.60 6.98

Entry per individual

City Entries Carry Turnover Header Short pass Med pass Long pass
Hart 0 0 0 0 0 0 0
Zabaleta 11 2 0 0 7 2 0
Clichy 20 2 0 0 12 5 1
Kompany 2 0 0 0 0 2 0
Lescott 1 0 1 0 0 0 0
Toure 14 0 0 0 6 7 1
Fernandinho 9 1 1 0 3 4 0
Silva 23 3 0 0 16 3 1
Navas 14 7 0 0 5 2 0
Aguero 8 4 0 0 4 0 0
Dzeko 5 1 0 0 3 1 0
Garcia 1 0 0 0 0 1 0
Nasri 5 1 0 0 3 1 0
Negredo 1 0 0 0 1 0 0
Milner 2 0 0 0 2 0 0
Total 116 21 2 0 60 28 3

Outcome per individual

City Total Dispossed Turnover Back out Fouled Corner won Total shots Shot in box Shot o/s box Blocked Saved Wide Goal Shots p/e Goals p/e
Hart 0 0 0 0 0 0 0 0 0 0 0 0 0 n/a n/a
Zabaleta 11 2 5 1 0 0 3 3 0 1 0 1 1 27.27 9.09
Clichy 20 4 7 2 0 2 5 5 0 1 2 1 1 25.00 5.00
Kompany 2 1 0 0 0 0 1 1 0 0 0 0 1 50.00 50.00
Lescott 1 0 0 0 1 0 0 0 0 0 0 0 1 0.00 100.00
Toure 14 2 7 2 0 1 2 0 2 0 0 1 1 14.29 7.14
Fernandinho 9 4 1 1 0 1 2 2 0 0 0 2 0 22.22 0.00
Silva 24 5 6 2 1 2 8 7 1 1 2 3 1 33.33 4.17
Navas 14 5 4 1 1 1 2 2 0 1 1 0 0 14.29 0.00
Aguero 8 3 4 0 0 0 1 1 0 0 0 1 0 12.50 0.00
Dzeko 5 1 0 1 1 0 2 2 0 0 1 1 0 40.00 0.00
Garcia 1 0 0 0 0 0 1 1 0 0 1 0 0 100.00 0.00
Nasri 4 0 1 2 0 0 1 0 1 0 1 1 0 25.00 0.00
Negredo 1 0 0 0 0 0 1 1 0 1 0 0 0 100.00 0.00
Milner 2 1 0 0 0 1 0 0 0 0 0 0 0 0.00 0.00
Total 116 27 35 12 4 7 29 25 4 5 8 11 6 25.00 5.17

Ben also took a retrospective look at the zone entries for the Champions League final, and whilst the purpose of the project is to ultimately look at the data over the course of a season, it may also shed an interesting and insightful light on individual games.

We can also see over even such a short sample size that City appear to favour certain entry types (short pass or carry) with certain players also driving these entries.

What was interesting to compare between the two games in particular (which you can do with the data dump over at Bitter and Blue) was the increased number of turnovers against Cardiff than against Newcastle, the use of their full backs high up the field and how certain players (notably Edin Dzeko and Jesus Navas) had far less entries.

There is the data dump over at Bitter and Blue which I will update weekly and this is a topic I will return to regularly here as the season progresses but in the meantime I’d be interested in any initial thoughts or comments.

  • Toshack

    Ben,

    Except that it is generally interesting facts…

    If I were the opposition team manager – would I use this to set up a game plan to negate City’s final third entry?

    If I were Pellegrini would I use this to compare two players playing the same position to see who “contributes” the most? Or to check if the system I’m trying to implement (tiki-taka, counter, “normal” possession based) is what is actually happening on the pitch?

    //Peter

    • http://BenPugsley Ben Pugsley

      I didn’t write this one Peter but it can obviously be used by opposition managers.

      If City are heavily using one method to enter the final third. and thereby getting close to the oppositions goal. I want to know what the method is. If it’s Clichy and Navas trying to dribble and carry the ball in, I want to stop it. If it’s Silva I want to know.

      Conversely, we seee City’s opposition consistently get into final third using long passes (far more than City do). This tells us about the oppositions reliance on counters against a possession bully like City.

      • Toshack

        Sorry Ben,
        Saw Pugsley and didn’t note properly that the piece was written by your brother (?).
        Anyway, thanks for the answer!
        //Peter

  • Nick

    What is considered a turnover entry? Where the opposition passes it backwards into their back 1/3 & loses it? How do you deal with throw-ins into final 1/3? are they considered passes?

    I know that we are used to dividing the pitch into 1/3s but I feel the data would be more interesting if we went to 1/4s. 40 yards out is a lot different to 30 yards out. 25% closer to goal!!

    • Danny Pugsley

      Hi Nick,

      Turnovers would be as you describe or when a player is dispossessed to gain entry.

      A throw in would be considered as a pass, generally a short one of course.

      It’s an interesting point about final 1/4 rather than final 1/3. We’ve always going with final 1/3 due to the other data that is available for the final 1/3. From the games I’ve watched so far this season it feels like there wasn’t many occasions where a final 1/3 entry wouldn’t also have ended up being a final 1/4 entry either. One Ill check on when I do the next game though.

      • http://blogs.columbian.com/portland-timbers/ Chris Gluck

        One thing to consider is that pitch size varies within FIFA – up until this year most had never associated possession as something of concern outside of the defending half versus attacking half.
        1/3 of 120 yards is 40 yards but 1/3 of 105 yards is 35 yards and to be honest a cross delivered from 40 yards can be just as brutal as a cross delivered from 35 yards Nick. What Danny is seeing is the same thing I am seeing here in analyzing the MLS – some teams look to penetrate more often against possession based teams by using long balls from the back four or square one or switches (right midfield to left forward) (vice versa) to get over the top.

        In my research to date though, I don’t need to know that detail of penetration anymore (especially since that statistic is not measure by OPTA or others) All I need to know is the level of passing attempts / completion “within the final third” versus the level of passing attempts “outside the final third”. That difference in positioning of the ball for possession already tells me where a team is more productive in penetrating the attacking third based upon possessions.

        Likewise, I also don’t need that granularity of information because I have defensive data to go with it. the more defensive clearances, tackles. cross blocks and interceptions a team makes ‘within’ the defending final third compared to the ‘entire pitch’ the more that team relies on playing deeper as opposed to higher… with that you can easily tell which teams are parking the bus and which teams aren’t.

        this does vary from game to game and opponent to opponent and it only becomes more apparent when you have ‘more data’ to analyze for one team versus another.

        I also have done extensive analysis on just one team (Portland Timbers) and have found that when I have all the data from all the teams it is easier to compare and create an analytical package that has correlation near .9998 when using a polynomial regression.

        that’s probably more than you wanted to know…

  • Pingback: Imagine There’s No Bayern | Bundesliga Fanatic()