Statistical outputs

[_Ben_]

Hi all,

I have a dream that FM becomes even more of a spreadsheet in future and allows access to all of the data that it collects. However, rather than turning this into a rant about that, I'd like to ask for some suggestions or feedback.

I've been building some stuff over the past months in Excel that try to mirror what you might see on Twitter, or elsewhere, about statistical analysis for real life players. I have created a few different views with what I can easily export (caveat is that one screen does require input of team possession, goals conceded and the number of rounds played in the league). I'm at a point now where I'm using these in my game and reasonably happy with what I've created but I'd like to get some feedback on what is missing and what else I can do with the numbers I have collated. 

At some point, I do plan to release this along with the custom views for the data collection, but that won't be until I've ironed it all out and am happy with it (even though it's still pretty basic!). Before anyone asks, this is all automated and I only copy and paste the raw data onto an Excel sheet before selecting (or typing) the name of the players involved.

----

Overall View:

This compares their per/90 outputs with whoever else's data I have collated - in this case, all wingers from the Turkish Super Lig. It then also adds the team performance to it. I like to screenshot pictures of their face, badge, nationality and positional abilities to add in here, also.

Comparative View:

Builds on this, adding another player into it to create a comparison.

Advanced Attacking Stats:

Really pushes the depth of what I can create using the numbers I have to create the fullest picture of a player in their attacking sense.

Advanced Defensive Stats:

A similar view for defensive players.

---

Any feedback would be hugely appreciated! I hope that this is the kind of thing somebody might find useful! It's certainly allowed me to express my love of data alongside the way I play FM.

[Shrewnaldo]

I love it. The graphics are just so good. I love the comparison bars and octagons in particular. I think they just make the data so clear

If you're looking for some suggestions for sharing these with others then I think it'd be worth articulating how you get the baseline for the percentages in the top graphics - just so others can understand what is being compared against.

But I think what you're asking for are suggestions on other stats or means to compare players? I think the ones that I have found most useful but I can't see covered above (apologies if I've missed it), are NPxG/shot and then some sort of comparison between the player's metrics and his teams - e.g. percentage of team goals (player goals per 90 as a percentage of team goals per 90), or player's direct contributions per 90 (player goals and assists per 90 as a percentage of team goals per 90). I think these combination of stats are key, and Forza's suggestion to try and root out high-performing players in low-performing sides was really helpful for me.

[fc.cadoni]

+1

I would love to see a filter / pre-made with Defensive, Possession, Offensive Actions. Some fields right now allows you to put between 0.00 & 2.00; but not between 0.10 etc.

 

Edited June 5, 2022 by fc.cadoni

[danielgear]

Massive fan of the comparative one, really makes it easier to compare between two players.

The graphics themselves look smart too. Out at the moment will have a proper look into the data later

 

I bet @FMStag would like these too and have some suggestions. 

[_Ben_]

10 hours ago, Shrewnaldo said:

But I think what you're asking for are suggestions on other stats or means to compare players? I think the ones that I have found most useful but I can't see covered above (apologies if I've missed it), are NPxG/shot and then some sort of comparison between the player's metrics and his teams - e.g. percentage of team goals (player goals per 90 as a percentage of team goals per 90), or player's direct contributions per 90 (player goals and assists per 90 as a percentage of team goals per 90). I think these combination of stats are key, and Forza's suggestion to try and root out high-performing players in low-performing sides was really helpful for me.

Thanks for the feedback, Shrew! I will attest to saying that a lot of the stats, for attackers, have come through from what you had created.

NpxG/shot is there, although there seems to be a bit of an issue with the formula at present so I'm looking at that.

I have that percentage but I've just re-worked the pie charts to make it clearer:

Here, you can see that CSKA Moscow rely heavily on Daryl Dike as he's scored 57% of the goals that have been scored whilst he is on the pitch and assisted 3% of them, giving him a total involvement of 60%. 

[_Ben_]

10 hours ago, danielgear said:

I bet @FMStag would like these too and have some suggestions. 

His blog, and Twitter, is of huge inspiration to me so I'm sure he'll see some links to how he creates his overview stats!

[danielgear]

30 minutes ago, _Ben_ said:

His blog, and Twitter, is of huge inspiration to me so I'm sure he'll see some links to how he creates his overview stats!

Yes Adam has done a great job with stats to the point where he needs to rebrand to @FMStats  

I agree with what Shrew said about the actual visual representation is spot on. I can see a lot of people wanting to use this. Also his point on explaining how to get the baselines makes sense. 
 

is it compatible with Google sheets too?

 

[Shrewnaldo]

2 hours ago, _Ben_ said:

Thanks for the feedback, Shrew! I will attest to saying that a lot of the stats, for attackers, have come through from what you had created.

NpxG/shot is there, although there seems to be a bit of an issue with the formula at present so I'm looking at that.

I have that percentage but I've just re-worked the pie charts to make it clearer:

Here, you can see that CSKA Moscow rely heavily on Daryl Dike as he's scored 57% of the goals that have been scored whilst he is on the pitch and assisted 3% of them, giving him a total involvement of 60%. 

Ah my bad. Perils of trying to read these things on the phone

I think we had a few conversations about trying to find similar stats for the defensive side but struggling. I've seen the "defensive actions per 90" thing but I'm not 100% keen on that. If only we could easily extract possession stats for the teams to compare defensive actions against average time defending or attacking. Maybe comparing defensive actions to passes per 90 is a helpful, if imperfect, proxy? 

[FMStag]

Hi @_Ben_.

Thanks @danielgear for looping me in on this!

This is absolutely excellent and I love it. As others have also said, I especially love the presentation. It's so clear to read and looks more dashboard than spreadsheet, which makes it appealing and accessible.

I love the parts which drop simple insight on your lap that FM doesn't natively give you. Like:

 and 

These are great, and provide a different (and ultra-useful) perspective when trying to compare individual players from high-performing teams vs ones in low-performing sides, as per @Shrewnaldo's earlier observation.

I also LOVE that you've managed to incorporate the possession adjusted variants in your sheet. That's something I should really figure out how to do. I assume the team's average possession so far this season needs aggregated manually and input somewhere for each individual player analysis?

I've done a similar thing to your 'as a creator,' 'as a finisher,' 'as a defender' sections, just with different names (creative efficiency, striking efficiency, defensive interferences), but the concept is the same. I've tied in the player's salary to try and tie the defensive or offensive actions to a value, almost in line with the "trying to buy points" value-driven stats analysis recruitment philosophy, but not sure if something like that fits in with your intentions here?

I am writing a guide just now for my Excel bit, but my main frustration is wishing I could automate multiple 'find and replace' commands to occur the moment the user hits 'paste.' I don't know VB though I wish I did! Does your automation involve any clever macros? If so, teach me!

I am keen to help however I can on this, even if it's just as a sounding board for feedback.

What's your Twitter @_Ben_, so I can follow?

Edited June 6, 2022 by FMStag

[Jimbokav1971]

Brilliant stuff @_Ben_

Lets be honest here, (without being too critical of FM), but that's sort of where I expected the Data Hub to be long before now. If anything I expect the FM World to lead the real World in terms of data analysis, because that's basically what it is at it's core. Instead it seems to lag badly behind. It's an area that SI could literally explode and improve, but for whatever reason things seem to happen in cycles and by the time they are introduced they are already out of date. 

There has been a fair bit of talk in the Premier League related this year of partnership goals scored with Kane/Son breaking the Lampard/Drogba record, (I think), and the way that modern full-backs now perform in attacking areas, (especially the Liverpool lads). SI could absolutely hammer this area to death and make it mind-blowingly good with, (to my mind at least), limited real effort because the info is already there. It just needs to be extracted and presented. 

For years, (actually it's more than a decade), I've wanted a top 10 ALL TIME appearance, goal-scorer, assister etc etc list for all competitions. The info is already being calculated. All they have to do is extract it and present it again. It seems an area that is such an easy win. 

I would love to follow you on Twitter too if you tweet about FM. 

[Shrewnaldo]

41 minutes ago, Jimbokav1971 said:

 

There has been a fair bit of talk in the Premier League related this year of partnership goals scored with Kane/Son breaking the Lampard/Drogba record, (I think), 

Would absolutely love this assist-scorer combination to be available somewhere

Or, as we've discussed a few times Ben, the ability to toggle set pieces in and out of the stats for key passes / chances created. Or just keep separate stats for those

Appreciate this is tangential to the premise of the thread... 

[FMStag]

7 minutes ago, Shrewnaldo said:

Would absolutely love this assist-scorer combination to be available somewhere

Or, as we've discussed a few times Ben, the ability to toggle set pieces in and out of the stats for key passes / chances created. Or just keep separate stats for those

Appreciate this is tangential to the premise of the thread... 

I love that player combination suggestion! Especially since in Analytical Data after a match you can see the passing combinations counted between each of the players. Simplifying that to just 'assist provider' and 'goalscorer' and making it an available chart for a team or competition seems like a nice lift and drop of similar fields of information.

I've made some suggestions to SI regarding future statistics, so I think it's a case of trusting that our feedback is taken under consideration going forward. Guess we just need to 'watch this space!' There will be a plan for how the Data Hub and available metrics are expanded upon, I'm sure.

My very first and most critical recommendation was a request to have key passes/chances created split into 'from open play' and 'from set pieces,' exactly like you've mentioned above @Shrewnaldo. Great minds, etc.

The others were:

- Every performance metric to have a 'per 90' variant included so they can be more functionally compared between players and teams who may have played a different number of games (without having to export to Excel)
- PPDA (passes per defensive action) would go hand in hand with your pressing overhaul, as it gauges the intensity and effectiveness of a team's press
- Post-shot xG would give more insight into the quality of an attempt on goal (and the potential save) rather than just the quality of the opportunity before a shot is attempted (which xG does)
- Players Packed or Progressive Passes would give an indication of how forward-thinking a player's passing stats were. Not just how many passes were completed, or how many resulted in a shot, but how many travelled forwards and bypassed how many opposition players.
- xAssists serves a similar purpose to the above, but just assigns a value to every pass a player makes to see how effective their overall passing is at creating dangerous opportunities
- Number of high intensity runs or sprints would paint a clearer picture of a player's style and intent of their movement more so than just distance covered or dribbles attempted/completed. I know SciSports do this one already for in-real-life football

[danielgear]

16 minutes ago, FMStag said:

I love that player combination suggestion! Especially since in Analytical Data after a match you can see the passing combinations counted between each of the players. Simplifying that to just 'assist provider' and 'goalscorer' and making it an available chart for a team or competition seems like a nice lift and drop of similar fields of information.

I've made some suggestions to SI regarding future statistics, so I think it's a case of trusting that our feedback is taken under consideration going forward. Guess we just need to 'watch this space!' There will be a plan for how the Data Hub and available metrics are expanded upon, I'm sure.

My very first and most critical recommendation was a request to have key passes/chances created split into 'from open play' and 'from set pieces,' exactly like you've mentioned above @Shrewnaldo. Great minds, etc.

The others were:

- Every performance metric to have a 'per 90' variant included so they can be more functionally compared between players and teams who may have played a different number of games (without having to export to Excel)
- PPDA (passes per defensive action) would go hand in hand with your pressing overhaul, as it gauges the intensity and effectiveness of a team's press
- Post-shot xG would give more insight into the quality of an attempt on goal (and the potential save) rather than just the quality of the opportunity before a shot is attempted (which xG does)
- Players Packed or Progressive Passes would give an indication of how forward-thinking a player's passing stats were. Not just how many passes were completed, or how many resulted in a shot, but how many travelled forwards and bypassed how many opposition players.
- xAssists serves a similar purpose to the above, but just assigns a value to every pass a player makes to see how effective their overall passing is at creating dangerous opportunities
- Number of high intensity runs or sprints would paint a clearer picture of a player's style and intent of their movement more so than just distance covered or dribbles attempted/completed. I know SciSports do this one already for in-real-life football

As discussed before Adam and with how the majority of FM tactics seem to be gegenpress more data from presses would be handy. Frustrates me that all the info I can get, yet I can’t get how many presses per 90, how many were successful by winning ball back etc. 

 

Alas that’s not possible at the moment. I’ve been doing my own equation for offensive players which is distance covered / minutes played. Theory is as we’re higher up the pitch majority of the game the more distance covered = more presses. Could be totally off though  (Adams informed me that QME distance covered is broken so only possible in ME leagues) 

If that gets fixed for FM23 could look at adding distance per 90 etc going forward @_Ben_

 

[Draakon]

I like statistics and comparison, but what bothers me a little in FM is that the game doesn't take into account what's our opponents like.

 

For example, I have a first team player that I prefer in tough league games, away games and Champions League games. And another player in the same position (youngster or rotation/backup) who plays in (early) cup rounds and against smaller teams in the league (starts mostly in home games). Statistical comparison between those two would be completely flawed - weaker player could have better or similar rating, less "quiet games" or "poor performances" just because he never starts against big teams. He never has to go up against Barca, Liverpool or Real. 

What I'd like to see more is the option to compare, for example "performances against top half teams" or some option like that which would take the strength of opponents into account.

[SixPointer]

Great work as always Ben. Something I’ve been saying for a while is that the likes of yourself and others should be making examples so that SI can really catch up in the game, the data hub is great addition to the game but I can’t help feel like it could be so much better with a lot more depth.

Personally Im going to echo Daniel and say that I would like to see more data based on pressing.

more specifically 

Passes Allowed Per Defensive Action (PPDA)

To measure the pressure that the defending team puts on the opposition players when they are in possession of the ball. The PPDA metric is calculated by dividing the number of passes allowed by the defending team by the total number of defensive action

[_Ben_]

Thank you so much to everyone who has liked this and offered some feedback! I'm going to keep working on things and am really open to ideas for what else could be done with the data we have at hand!

----

8 hours ago, danielgear said:

is it compatible with Google sheets too?

Not sure, sorry!

6 hours ago, FMStag said:

I also LOVE that you've managed to incorporate the possession adjusted variants in your sheet. That's something I should really figure out how to do. I assume the team's average possession so far this season needs aggregated manually and input somewhere for each individual player analysis?

7 hours ago, Shrewnaldo said:

Maybe comparing defensive actions to passes per 90 is a helpful, if imperfect, proxy? 

I may have been too simple with it, using a formula I found on the internet that balances the possession back to that 50/50 to create that possession adjusted metric.

"50%/TEAM POSSESSION x Defensive Actions."

As for the team possession - it's manually inputted once and then an Index formula:

Matches it with the team name (which is why it appears on the defensive metrics and not offensive metrics), which shows up in my raw data, like so:

6 hours ago, FMStag said:

I am writing a guide just now for my Excel bit, but my main frustration is wishing I could automate multiple 'find and replace' commands to occur the moment the user hits 'paste.' I don't know VB though I wish I did! Does your automation involve any clever macros? If so, teach me!

I've used macros, a long time ago, but do want to go back in - if only to just replace the '-' with '0' although I do use the IFERROR(A1-A2,"0") for example, which would just show a 0 if there was an invalid string or text in either A1 or A2. I think macros are the next step, for me in this creation.

6 hours ago, FMStag said:

What's your Twitter @_Ben_, so I can follow?

5 hours ago, Jimbokav1971 said:

I would love to follow you on Twitter too if you tweet about FM. 

I just have a private twitter and peruse the #FM22 hashtag from time to time! Sorry.

5 hours ago, Jimbokav1971 said:

Lets be honest here, (without being too critical of FM), but that's sort of where I expected the Data Hub to be long before now. If anything I expect the FM World to lead the real World in terms of data analysis, because that's basically what it is at it's core. Instead it seems to lag badly behind. It's an area that SI could literally explode and improve, but for whatever reason things seem to happen in cycles and by the time they are introduced they are already out of date. 

I 100% agree with this. I know that the Hub is in it's infancy but, like you, expected a bit more. I was even able to create these scatter plots last year with a custom skin, so it's not moved on a lot. However, FM tends to be like this as it embeds new features over the course of years. I live in hope.

4 hours ago, Shrewnaldo said:

Would absolutely love this assist-scorer combination to be available somewhere

When you've still got things like this:

A wealth of stats just not been exported or this:

Several buttons to click through to see progressive passes (a stat that has become more important of late) but is not accessible easily or readily available for more than just your team, your idea is somewhat of a pipedream: sadly!

3 hours ago, danielgear said:

I’ve been doing my own equation for offensive players which is distance covered / minutes played. Theory is as we’re higher up the pitch majority of the game the more distance covered = more presses. Could be totally off though 

I saw this but, as you say, saw midfielders rating 7.5 over 20 games but running 4km per game just because the Belgian league isn't modelled. A real shame as I like the idea of defensive actions per metre ran or vice versa!

2 hours ago, Draakon said:

What I'd like to see more is the option to compare, for example "performances against top half teams" or some option like that which would take the strength of opponents into account.

You can just about see that in team metrics - but, again, it's just scorelines and not a lot more. Certainly an area that I would love to explore but it's never a possibility on the scale I want - where I sometimes compare European top flight forwards under 21 - which is a massive criteria range!

41 minutes ago, SixPointer said:

Passes Allowed Per Defensive Action (PPDA)

I've done this in the past. Here is a graph I made at the start of my save:

But it's super time consuming as you have to collate all the info for team losses of possession and opponents losses in opposition halves etc. Definitely a doable metric and a really useful one!

[_Ben_]

I've been having a play this evening with a thought in my head that this is all quite nice for recruiting individuals, but what about looking at leagues or teams for top players. I might want to have a look at teams who have been relegated and see if they have any standout players, for example.

So I have made this:

It's collating their advanced metrics and then ranking them, not percentile this time, out of those in the same data set - in this case, the league.

Lots of work to be done here including adding the top three from the league so you could, in theory, input a whole league's worth of players and unpick an analyst report from there. Is this the kind of thing that would be usable for people?

EDIT:

Here is what a 'league table' of advanced stats might look like...

Edited June 8, 2022 by _Ben_

[QP]

Man that's impressive, would be amazing to use these expecially while doing a moneyball save.