clock menu more-arrow no yes mobile

Filed under:

Analytical Musings: Cranking Creativity up a Notch

Part III

Watford v Chelsea - Premier League Photo by Catherine Ivill/Getty Images

So far, we’ve used dribbling and key passes and shots to help look at how well players this season have done, and in a slightly different way, how well what I consider the top players of the decade have done over the years. It’s time to tweak this one more time, and see if we can utilize it to help identify the best players.

I’m using the Premier League for this analysis, and using data from the 2009-10 season to now, because mainly that’s what Whoscored.com has available at the player level. I’m using only league data, as well. What I’ve done is gather up the following data points for each player (the good ones at least) for each season in that range:

  • Season
  • Player
  • Club
  • Minutes Played
  • Goals
  • Assists
  • Successful Dribbles
  • Key Passes
  • Shots on Target
  • Total Shots

I built myself a model, after pulling together data for about a thousand player-seasons. I scraped the top 100 or so for each of the summary, KP, and DRB options for each season from whoscored’s data, and then pulled it all together via lookup formulas (this dataset is too small to worry about a DB or anything). If I was missing any data from what floated near the top, I went and manually added those entries in. For example, a few players showed up in one or two of the three tables, but not all of them.

At this point, I had a nice big dataset. I had it with Per-90 numbers, and Minutes-Per numbers, and what I then wanted to do was filter it down to give myself the Top 30 player-seasons in my analysis time frame, utilizing these criteria. It’s tough, because when developing criteria weights you want to be objective as possible, while making sure you are best using the metrics you want to track. I’m sure there are better ways, I’m not a stats-degree level of metric-nerd, I just do this sort of thing for a living.

What I ended up with for my criteria weights were the following, based off of Per 90 numbers (the easiest way when adding things up):

  • Successful Dribbles — 1.0 weight value (it’s a baseline)
  • Key Passes — 1.0 weight value (also a baseline)
  • Goals — 3.5 weight value (I wanted them to matter)
  • Assists — 4.0 weight value (again, I wanted them to matter)
  • Shots on Target — 1.75 weight value

Additionally, I applied the following criteria to weed out anomalies:

  • Minutes played >= 1500
  • Goals + Assists >= 10

There were one or two odd results floating up as I was working through my weights and I wanted to maintain a good list. The below is what I came up with (excuse the omitted “Persie” on the 2011-12 season, I missed that when putting the table together):

I included my rank value as a column in the middle. This is simply a sum of the-Per 90 values with the restriction criteria applied and the Per-90 value weights applied. Literally it looks like so:

=IF(OR(C2<1500,(D2+E2)<10),0,SUM((N2*$Z$2),(O2*$Z$3),(P2*$Z$4),(Q2*$Z$5),(R2*$Z$6)))

Super tidy and simple.

A part of me really worried about some sort of bias creeping in with having Eden’s current season be #1, and believe me, I went back and forth a lot on this. The reality is, though, that the #2 listing, Suarez in 13-14, had almost twice the minutes. When you break it down on minutes played, things start looking quite remarkable for Eden’s season this year, ESPECIALLY when you consider he started hurt and wasn’t scoring (and still doesn’t have many assists). But as I tweaked the values around, he was always in the top 3, and I ended up wanting the weights I selected.

How the data breaks down

When looking at the table, the first things you should probably notice are the repetition. The names, the clubs, the years.

For Clubs:

  • Chelsea - 5 entries
  • Arsenal - 6 entries
  • Liverpool - 4 entries
  • Manchester City - 9 entries
  • Manchester United - 1 entry
  • West Ham - 1 entry (Somehow this got labeled as Manchester Ham. Humorous.)
  • Tottenham - 3 entries
  • Leicester - 1 entry

For Players:

  • Alexis Sanchez - 3 entries
  • Carlos Tevez - 1 entry
  • Cesc Fabregas - 1 entry
  • Didier Drogba - 1 entry
  • Dmitri Payet - 1 entry
  • Eden Hazard - 4 entries
  • Gareth Bale - 1 entry
  • Harry Kane - 2 entries
  • Kevin De Bruyne - 1 entry
  • Leroy Sane - 1 entry
  • Luis Suarez - 2 entries
  • Mo Salah - 1 entry
  • Nani - 1 entry
  • Philippe Coutinho - 1 entry
  • Riyad Mahrez - 1 entry
  • Robin Van Persie - 2 entries
  • Sergio Aguero - 6 (!) entries

For the Years:

  • 2009-10 - 2 entries
  • 2010-11 - 3 entries
  • 2011-12 - 2 entries
  • 2012-13 - 2 entries
  • 2013-14 - 3 entries
  • 2014-15 - 3 entries
  • 2015-16 - 4 entries
  • 2016-17 - 5 entries
  • 2017-18 - 6 entries

In my opinion the talent level of the Premier League has been steadily increasing, and we are seeing that with more of this type of player.

The Visuals

For those that want to see some visual representations of the data, let’s go down that rabbit hole. Keep in mind the weight-values I used to select the top players aren’t coming into play for these charts, these are using raw Minutes-Per values. I could weight them, I suppose, but I’m not yet to the point of wanting to build some sort of new metric from all of this, so let’s keep it straight forward.

First, let’s take a look at a very simple Dribbles vs Key Passes.

No surprise who is in that lower left corner all by himself. And it’s really interesting how clustered certain players are, too. Sergio Aguero up high, Sanchez all grouped up, Suarez practically on top of each other. It’s intriguing by the way how many entries from this current season are in here, too.

Next, looking at shots on target vs goals, the forwards come into play a bit more in this one:

Notice that there’s a certain cluster that seems to be where the ‘best of the best’ in the league seem to be pushed up against. Definitely worth a closer look. I find it really intriguing that these top players (Aguero, Kane, Drogba, Suarez, etc), and especially those a bit more focused on scoring, all have a certain wall they hit with respect to minutes per goal, as well as also having a limit on their shots on target. Those 9 all tightly bunched up, and the other 5-7 nearby comprises half of the list, and they are all pushing that same limit. Beyond CR7 and Messi in their absolute prime, I’m curious if anyone has pushed beyond these limits in the modern era.

Lastly, let’s take a look at what I’ll call the “Complete Player” combination. I’ve got Dribbles and Key passes on the X vs Shots on Target on the Y:

Notice that the all-elusive lower left corner is very sparsely populated. It’s tough to be a Top Level player. The scorers leaning towards the bottom, as they shoot more, and the passers/creators drifting up a bit, as they obviously do more with the ball to create scoring opportunities.

In Summation...

I wanted to see how the best players in their best seasons behaved in relation to each other, especially with respect to the metrics I was using for review. I potentially created the very beginnings of a ranking model in terms of the offensive end for players, at least the ones that have the ball the most, and I’m happy with that. It’s still incredibly basic, but I think with proper weighting it did quite a good job of identifying top performers. And I think I’ve put some nice visuals up for folks to look at, and review, and maybe ask some more questions from.

In terms of specifics, the most interesting returns were that certain players not only excelled but consistently excelled, with similar top results (such as Sanchez, or Aguero, or Suarez), whereas others (such as Eden) continue to progress and improve as their career continues.

I was also surprised to see some of the dribbling and key pass numbers for a lot of these forwards in their best years. I didn’t think they would be as high as they were. And seeing a few players sneak in for ‘career’ years, but then tend to revert back to more representative numbers, like Payet or Nani or Mahrez.

I’m also a bit blown away at how the phenomenal season that Manchester City are having is inflating so many numbers for all of their players. Three different players for City this season made this list. And interestingly enough one each for the Top clubs (beyond Manchester United, strangely enough).

I think the next thing after this would be to make similar comparisons to the top seasons from other top players in Europe, and I’d probably want to go back farther if possible for Messi and CR7, and maybe other players and see how things measure up comparatively.

What do all of you think? Interesting? Good read? Waste of time? Boring? Too simplistic? Want to see more?

Please share in the comments section. Thanks for reading!

Sign up for the newsletter Sign up for the We Ain't Got No History Daily Roundup newsletter!

A daily roundup of Chelsea news from We Ain't Got No History