Key metrics (KPI) for each position

vonTrips · September 3, 2023

Hey. Hey,
first of all, I'm sorry if there's a similar topic here somewhere, but I couldn't find it. The fact is that I've been playing without attributes for a while now (thanks to @GIMN and his great skin), which requires looking more at player stats, data analysis etc.

I've looked at a lot of sources:

Spoiler

There are many more sources, and everyone approaches this primarily according to their own tactics (which goes without saying).

Based on all this, I have compiled the following table of positions/roles and their KPIs, which are divided into primary, secondary and other. The primary ones are the most important for the position (I use them to find players in the game), the secondary ones need to be tracked as well. Others can provide additional insight into a player's performance, but may not be very relevant (sometimes I ignore them).

Also, when possible, I don't use hard data (e.g. Pass Att) as it can be skewed by role and tactics. I think ratios, e.g. Pass Ratio, etc. are better to telling about value regarding player quality/performance. And unless otherwise stated, I use per 90 metrics exclusively.

Bold - primary KPI
Normal - secondary KPI
Purple - other KPI

GK

Spoiler

Save Ratio - absolute KPI;
xG Prevented - how many chances it can defuse;
Shots Per Conceded - we want as many shots as possible on the goalkeeper for every goal scored (not possible to find out in game, need to calculate or have multiple columns in view);
Pass Ratio - accuracy of passes is important in a modern goalie (GK-Libero), in my opinion more important than the number of passes;
Clean Sheets - the more is better (but can be skewed by the team);
Goals Concerned - the fewer is better (but can be skewed by the team);
Progressive pass - a modern goalkeeper should be able to set up actions, that's why I watch progressive pass;
Mistakes - ideally the goalkeeper should not make mistakes that result in goals

CB

Spoiler

Headers Ratio - the ratio is more important than the number of Headers Won;
Tackle Ratio - for me the Tackle Won ratio is more important than the absolute numbers;
Tackle Quality - Tackle Att / Fouls Made; the ideal player can tackle without fouling, yet must be counted manually;
Interceptions - this falls into the "last minute" category and I'm happy for a CB to control this; (can be combined with Tackles per Game to determine hidden qualities);
Poss Ratio - Poss Won / Poss Lost ratio (to be calculated manually), defenders need to gain balls, not lose them;
Key Headers - Key Headers are dependent on player performance, not on role/tactics;
Key Tackles - Key Tackles are dependent on player performance, not on role/tactics;
Mistakes - CB should not make mistakes (it's better to avoid those who make a lot of them);
Blocks, Clearances - they can be important indicators, but I don't pay much attention to them (Is it a my mistake?);

BPD

Spoiler

Headers Ratio - the ratio is more important than the number of Headers Won;
Pass Ratio - a CB must be able to pass, so I want a high percentage of successful passes;
Tackle Quality - Tackle Att / Fouls Made; the ideal player can tackle without fouling, yet must be counted manually;
Interceptions - this falls into the "last minute" category and I like a tracker to master this; (can be combined with Tackles per Game to determine hidden qualities);
Poss Ratio - Poss Won / Poss Lost ratio (to be calculated manually), even more important parameter if I want the defender to play constructively;
Progressive Pass - since it is an absolute number, dependent on tactics, it is a secondary parameter (but it is more valuable than the absolute number of passes);
Key Headers - Key Headers are dependent on player performance, not on role/tactics;
Key Tackles - Key Tackles are dependent on player performance, not on role/tactics;
Tackle Ratio - To me, the Tackle Won ratio is more important than absolute numbers;
Mistakes - if I want a CB to run the ball, he can probably make the occasional mistake (it's better to avoid the ones that make a lot of them);
Blocks, Clearances - they can be important indicators, but I don't pay much attention to them (Is it a mistake?);

FB

Spoiler

Tackle Ratio - to me, the Tackle Won ratio is more important than absolute tackling numbers;
Tackle Quality - Tackle Att / Fouls Made; the ideal player can tackle without fouling, yet must be counted manually;
Pass Ratio - since FB passes more from behind, he needs to have the highest pass completion rate;
Interceptions, Blocks - FB primarily defends, this should be a good indicator of his ability;
Poss Ratio - Poss Won / Poss Lost ratio (calculate manually), defenders need to gain balls, not lose them;
Key Tackles - Key Tackles are dependent on player performance, not on role/tactics;
Cross Ratio - FB's won't center much, but if they do I want centers to be addressable;
xA, Chance Created - Centers should create chances, but it's not an important indicator for FB;
Distance, Sprints - I think FB will "sit back" more so his numbers won't be too high (Or am I wrong?);

WB

Spoiler

Distance, Sprints - I think both are important for WB, WB should be constantly running up and down;
Tackle Quality - Tackle Att / Fouls Made; the ideal player can tackle without fouling, yet must be counted manually (tackle quality is important for any defensive player);
Dribbles Made, Cross Ratio - Running "up and down" with no results is a waste; for WB, dribles and centers are important;
Open Play Key Pass - not sure whether to address absolute numbers here, or just have them as a secondary parameter;
Progressive Pass, Pass Ratio - WB should set up chances on the ground and through the air, that's why I address passing;
xA, Chance Created - I definitely want higher numbers than FB, so this is a secondary parameter, but it can be very dependent on the role/tactics of the player;
Key Tackles - Key Tackles are dependent on the player's performance, not on the role/tactic;
Tackle Ratio - to me, the Tackle Won ratio is more important than absolute numbers;
Blocks - can be an important indicator, but I don't pay much attention to it (Is this a mistake?);

CM - defender

Spoiler

Tackle Quality - Tackle Att / Fouls Made; the ideal player can tackles without fouls, yet must be counted manually (quality of tackles is important for every defensive player);
Tackle Ratio - "DM" is mainly concerned with defense, so I want the highest tackling ratio possible (it's more important than on the edge of the field);
Key Tackles - Key Tackles are dependent on the player's performance, not on the role/tactic; (if it's a box-to-box midfielder Key Tackles are a sketchy parameter)
Interceptions - probably the most important "DM" characteristic and an indicator of player quality;
Distance - primarily for a box-to-box midfielder (he should cover a large part of the pitch), otherwise it can be a secondary indicator;
Pass Ratio - depends on tactics and role, but I think even the "DM" must be able to pass; (for a box-to-box midfielder I would consider this a primary indicator);
Progressive Pass - depends on tactics and role;
Dribbles Made - only for the box-to-box midfielder (who should be able to dribbles);
Poss Ratio - Poss Won / Poss Lost ratio (counted manually), the "DM" must win balls, not lose them;
OPkP, Chance Created - depending on the role this can be a monitored parameter;
Passes Completed - depending on the role this may be a monitored parameter;

CM - playmaker

Spoiler

Open Play Key Pass, xA, Chances Created - the playmaker has to make key passes and get players into chances;
Progressive Pass - creative players have to make penetrating passes;
Poss Lost - a player who creates but loses a high amount of balls is a "possible problem";
Pass Ratio - the pass ratio is an important parameter, linked to Poss Lost;
Chance Quality - xA/OPkP (count manually), determining how good the key pass players are;
Dribbles Made - a good indicator for Advanced Playmaker;
Poss Ratio - Poss Won / Poss Lost ratio (counted manually), I don't only track the number of lost balls, but also how many a player gets;
Passes Completed - depending on the role this can be a monitored parameter;
Tackle Quality, Tackle Ratio, Interceptions - even a playmaker can defend, so I track these parameters, but they are not important (I prefer to reach for a complex player who creates and defends);

CM - attack midfielder

Spoiler

Dribbles Made - An offensive midfielder must be able to dribble the ball;
Open Play Key Pass, xA, Chances Created - an important indicator for AM to pass and create chances;
Chance quality - xA/OPkP (counted manually), determining how good a player's key passes are;
Goals - in some roles this can be a primary indicator, in others a secondary one (I like AMs who score goals, for example);
NPxG - I want the AM to not only create chances, but to get them;
Conversion Rate - 1 goal in 10 shots? Thanks, I don't want...
Pass Ratio - like any player, I want them to have the best pass ratio;
Progressive Pass - "AM" probably shouldn't completely mess them up, maybe that could be other indicator;
Shot Ratio - if a player shoots, I don't want them to shoot into the stands (depends on the role, who you want on the AM finisher, will definitely address this indicator more);
Sprints - I don't know if this is a rather useless indicator?
Poss Lost - "AM" can already afford some ball losses, but there shouldn't be too many;

Winger

Spoiler

Dribbles Made - Winger needs to dribble and create pressure on the defense (somehow link this to Fouls Against ????);
Sprints - rushes into open space, is this a primary parameter for Winger?
Cross completed - important parameter, Winger has to center and I want a lot of successful centers (one of the few absolute metrics);
Cross Ratio - if he centers a lot, I want to know how successful he is;
xA, Chance Created - I don't know if these are supposed to be primary indicators or just secondary ?
Distance - to me this is just a secondary indicator, a good winger doesn't need to cover a lot of distance, just sprints;
Open Play Key Pass - maybe I'm not understanding this correctly, but to me a winger creates the most chances with centres, that's why OPkP is just a secondary indicator;
Pass Ratio - as with any player, I want them to have the best successful pass ratio;
NPxG, Shot Ratio - it's nice if Winger gets a chance and hits the net;
Poss Lost - he can afford some ball losses, but there shouldn't be too many;
Fouls against - is it an indicator of a good Winger that he is fouled often?

Wide Forward - like Inside Forward

Spoiler

Dribbles Made - WF needs to dribble and create pressure on the defense (somehow link this to Fouls Against ????);
Goals - I want this player to score more goals than he creates them;
Shot ratio - shots must end up in the goal, not in the stands;
Open Play Key Pass - he should be able to make a key pass though;
Pass Ratio - as with any player, I want them to have the best ratio of successful passes;
NPxG, Conversion Ratio - player should be getting chances and finishing them, but I don't mind worse numbers than a pure striker;
xA, Chance Created - sometimes a player has a chance to set up, that's why I watch these parameters;
Offsides - depends on the specific setup, but a player should definitely have as few offsides as possible;
Sprints - definitely not as important a parameter as for Winger;
Poss Lost - he can afford some ball losses, but there shouldn't be too many;
Fouls against - is it an indicator of a good Wide forward that he is often fouled?

Target Man

Spoiler

Headers Won, Headers Ratio - TM is mainly there from winning headers, whether finishing on goal or as a pass to a teammate;
xA, Chances Created - TM should create chances for other teammates by winning duels;
Goals - TM is more supportive (as I understand it), so goals can be a secondary parameter
OPkP, Pass Ratio - TM doesn't have to be a playmaker, but must be able to pass the ball to teammates as well;
Offsides, Fouls Made - weaker players are often offside or fouled, such players are better avoided;
Conversion Rate, NPxG, xG Overperformance - TM is also an attacker, so it is good to watch attacking parameters, but they are not important;
Poss Lost - TM should not lose many balls;

Striker

Spoiler

Goals - this striker is on the pitch mainly to score goals;
Conversion Rate, Shots on Target - we want him to take a lot of shots on goal, but at the same time efficiently (1 goal out of 10 shots on goal is really bad);
NPxG - the striker needs to get chances every game;
xG Overperformance - the ideal striker is one who scores more goals than the number of chances he gets;
Shots on Trg Ratio - maybe a useless indicator when I'm watching Conversion Rate?
Offsides - a player who has a lot of offsides can unnecessarily spoil the game for the whole team;
Pass Ratio - every player has to be involved in the game and therefore pass well;
Sprints - depending on the role, this can be a secondary indicator;
Headers Ratio - depends on the role and tactics, some people want a striker who will also header, some people don't;
xA, OPkP, Chance Created - depending on the specific role it may be desirable to have a striker who not only finishes but can create chances;

The goal is not to create another spreadsheet or application. Nor is it to create new views for the game (there are many available, e.g. from @FMStag, and they are really great). I just want to have a discussion regarding specific metrics and their relevance to the position. I would also like to have feedback on the choice of primary or secondary KPIs. Like for example, shouldn't I address Blocks, Clearances and Interceptions for each defensive position? What about pressures, should I add them, too? ... Thanks for any feedback.

Big thanks to @GIMN for his skin, and for the insights I gained earlier:

Chance quality - counting chance quality as xA/OPkP was his idea and I like it
xA before Asissts - using xA instead of Assist as an indicator for assists, because xA is not dependent on teammate quality; i.e. whether a teammate converts a chance or not

PS: I apologize for my English (sometimes DeepL had to help me).

Edited September 3, 2023 by vonTrips

milenec11 · September 6, 2023

Excellent article I am surprising that no one participate sport interactive should add PPDA also :

What does PPDA stand for?

PPDA stands for passes per defensive action.

What is PPDA? What is passes per defensive action?

As the high press became a more prevalent part of football, and the use of data increased, attempts were made to quantify pressing. While simple data such as tackles and interceptions in the attacking third, or the opposition’s pass completion rate in their own half, can give some indication of how well a team presses, they have severe and understandable limitations. PPDA is an advanced metric that attempts to quantify the act of pressing in football in a more comprehensive (though, it’s important to say, not full) manner.

How is PPDA calculated?

PPDA focuses only on actions in the area of the pitch in which a team might reasonably execute a high press. The people that came up with PPDA decided on three-fifths (or 60%) of the pitch nearest the opposition’s goal. This is the whole of the opposition’s half, plus a fifth of their own half (below).

vonTrips · September 7, 2023

On 06/09/2023 at 10:17, milenec11 said:

Excellent article I am surprising that no one participate sport interactive should add PPDA also :

What does PPDA stand for?

PPDA stands for passes per defensive action.

What is PPDA? What is passes per defensive action?

As the high press became a more prevalent part of football, and the use of data increased, attempts were made to quantify pressing. While simple data such as tackles and interceptions in the attacking third, or the opposition’s pass completion rate in their own half, can give some indication of how well a team presses, they have severe and understandable limitations. PPDA is an advanced metric that attempts to quantify the act of pressing in football in a more comprehensive (though, it’s important to say, not full) manner.

How is PPDA calculated?

PPDA focuses only on actions in the area of the pitch in which a team might reasonably execute a high press. The people that came up with PPDA decided on three-fifths (or 60%) of the pitch nearest the opposition’s goal. This is the whole of the opposition’s half, plus a fifth of their own half (below).

Oh, good point. That looks interesting. But my post is about things that I can get from the game.

My idea was to start a discussion about KPIs for individual positions. So that the resulting list would then make sense across roles/tactics (maybe not too different from what I have here). Ideally having 4 primary indicators, 4 secondary indicators and 4 other indicators for each position. Unfortunately, no one wants to discuss. 🤔🙄

But thanks to all who reacted 😉

rich10 · September 8, 2023

Hey @vonTrips you might find more discourse in the Good Player forum for this topic.

I've always been interested in playing without being able to see the player attributes so I think I'll give it a go with the skin you've recommended here.

I do have a couple of questions for you:

1. I presume we can save these positional stats in a template for each position when searching for players?

2. Wouldn't you need to build up a history of performance for players? And so how does this work when you first load up a save? Or would you typically disable the 1st transfer window, or run the game on for a season so this data accumulates?

3. Also, this data is lost when a new season starts, correct? Is this when you'd need to make a note against potential transfer targets in a spreadsheet outside of the game?

Thanks in advance, and thanks for the quality post.

milenec11 · September 8, 2023

I believe ( read it again and again ) that is a good idea for some skin creator to include this KPI but Unfortunately the game does not give the stats in order to create them.

the most important for exp KPI for the position of GKs stats is this:

GAA

Goals Allowed Average

Number of goals allowed by game on average.
Formula: (GA*90)/MIN

and Unfortunately, will have to calculate it because the game does not include it ,

This also happens and in other KPI positions and this idea became not easy applied .

Edited September 8, 2023 by milenec11

vonTrips · September 8, 2023

13 hours ago, rich10 said:

I presume we can save these positional stats in a template for each position when searching for players

Yes, you can use all stats in views or filters.

13 hours ago, rich10 said:

Wouldn't you need to build up a history of performance for players? And so how does this work when you first load up a save? Or would you typically disable the 1st transfer window, or run the game on for a season so this data accumulates?

Yes, history is a problem. The data is in the game, but the game can't provide it. At the end of each season you need to backup your file.

And I always disable the 1st transfer window. After half of season the data become relevant.

14 hours ago, rich10 said:

Also, this data is lost when a new season starts, correct? Is this when you'd need to make a note against potential transfer targets in a spreadsheet outside of the game?

For data history, I have a small application where I always save the outputs and can easily browse and compare them.

14 hours ago, rich10 said:

I've always been interested in playing without being able to see the player attributes so I think I'll give it a go with the skin you've recommended here.

Try it, it's fun 😉

14 hours ago, rich10 said:

Thanks in advance, and thanks for the quality post.

You're welcome and thank you

zois92 · December 7, 2023

Hey @vonTrips, just came across this wonderful article of yours. I’m curious to see if you used it in any of your saves and how it helped you recruit players.

I have one more question.

Is there a way to calculate the open play involvement of a player?

Edited December 7, 2023 by zois92

vonTrips · December 8, 2023

Hi @zois92,

yes, I played with data analysis in 3 saves.

1. Moneyball in Italy - this is my first, I learned it here. https://www.fmseries.cz/viewtopic.php?t=11987

2. Baník Ostrava save - all on my blog https://medium.com/@fmpanda, first post https://medium.com/@FMPanda/fm23-zvedáme-baník-ze-dna-1-4bfc3b5912b

3. C.F. Os Belenenses - the True ones of Belém. This is my last save which I start played in FM23 and transfer the save to FM24. On my blog, too. First post here https://medium.com/@FMPanda/fm23-c-f-os-belenenses-ti-praví-z-beléma-1-b0b992083d98

A na blogu najdete další příspěvky o analýze dat a Moneyballu. Stejně jako moje poslední verze o KPI - https://medium.com/@FMPanda/moneyball-klíčové-metriky-2-8a14274887fa But this is for FM23. I have an application for summaring player performance, but it's not debugged. For FM24 I haven't plan create an application. I want use only data in game (probably) and all what Mustermann Iconic Skin give me.

Maybe will be a new version of KPI, where I'll use Open Play Crosses etc. Follow my blog to find out. 😉

milenec11 · December 8, 2023

Hahaha myslis ze on mluvi czesky? Ja myslim ze on je Rek Jako ja.

Edited December 8, 2023 by milenec11

zois92 · December 8, 2023

6 hours ago, vonTrips said:

Hi @zois92,

yes, I played with data analysis in 3 saves.

1. Moneyball in Italy - this is my first, I learned it here. https://www.fmseries.cz/viewtopic.php?t=11987

2. Baník Ostrava save - all on my blog https://medium.com/@fmpanda, first post https://medium.com/@FMPanda/fm23-zvedáme-baník-ze-dna-1-4bfc3b5912b

3. C.F. Os Belenenses - the True ones of Belém. This is my last save which I start played in FM23 and transfer the save to FM24. On my blog, too. First post here https://medium.com/@FMPanda/fm23-c-f-os-belenenses-ti-praví-z-beléma-1-b0b992083d98

A na blogu najdete další příspěvky o analýze dat a Moneyballu. Stejně jako moje poslední verze o KPI - https://medium.com/@FMPanda/moneyball-klíčové-metriky-2-8a14274887fa But this is for FM23. I have an application for summaring player performance, but it's not debugged. For FM24 I haven't plan create an application. I want use only data in game (probably) and all what Mustermann Iconic Skin give me.

Maybe will be a new version of KPI, where I'll use Open Play Crosses etc. Follow my blog to find out. 😉

anything in English? :lol:

milenec11 · December 8, 2023

1 hour ago, zois92 said:

anything in English?

χρεισημοποιησε το google translator θα βοηθησει καπως

dzek · December 8, 2023

I think the stats and overall average player ratings are more about roles than positions.

The general stats for each position can show a picture of the team but the more specific stats are towards the roles.

For example BPD would be good to consider key passes rather than from a CD. Goals per game would be more relevant for an IF than for W/IW roles etc.

vonTrips · December 8, 2023

5 hours ago, zois92 said:

anything in English?

Nothing, sorry. Please use DeepL to translate 😉

3 hours ago, dzek said:

For example BPD would be good to consider key passes rather than from a CD. Goals per game would be more relevant for an IF than for W/IW roles etc.

Thanks for the feedback, I'll definitely think about it

BL1TZ_GT · January 28

Loving this. But curious if you have a range that's ok, good & poor like FMStag does in one of the links you got inspiration from.

Love the explanation,but I never know if the number in the KPI is OK, good or poor, its just a number to me haha

@vonTrips

st4lz · January 29

I did a lot of research into this and haven't come up with a definitive solution yet.

Main issues:

* Some of the stats are favoring players heavily in teams that have an advantage - it is almost impossible to beat the average xG or Pas % from the underdog team

* Opposite, some of the stats are much better in the underdogs, especially the defensive ones - all of the Tackles Completed, Key Tackles, Interceptions, Clearances, Blocks, Headers, etc are higher for teams that defend often

* Lower efficiency doesn't always mean worse performance - ex. Headers Won Ratio for two CB, one is the air superiority corner target, and the other stays back during Attacking Set Pieces. The first one may have 60% HWR with 12 Aerial Attempts per 90, the other may have 80% with 3 Aerial Attempts per 90. HWR is as useless as Conversion Rate or Shots on Target % (vs. xG), you can't just aggregate them and treat all situations equally (you discriminate low probability scorers, like players who score a lot of headers or long shots will have lower).

* Some of the stats are dependent on team tactics or playing roles, and can't be compared objectively - it is hard to compare a No-Nonsense CB with a CB when it comes to Pass %, the first one is encouraged to make safe clearances. Or to compare a Pressing Forward with a Trequartista when it comes to Pressures Attempted.

* Some of the stats are negatively correlated - there is often a low Pas % that goes along a high Key Passes. Creative players do take more risks, so their mainline Pas % is lower, which doesn't mean their passing is weaker. You need to identify those relations, and not try to put too much weight on Pas % when looking for a playmaker.

I generally like to create indicators that are constructed based on available statistics, I like the Tackle Quality (Tackles Attempted / Fouls Made) that you created.

I have completed my evaluation model for forwards, and I think the fewer areas you compare the better results you may have. My model assumes 3 main abilities to evaluate a forward: winning space, finishing, and success to waste ratio.

Winning space (favors big teams): simply Np-xG/90

Finishing (favors underdogs): This is xG overperformance, but calculated as a ratio, not as a difference - Np_Goals/Np-xG

Success to waste ratio (neutral for team strength or tactical bias): Shots on Target, Crosses Completed, Key Passes, Critical Chances Created, Headers Won divided by Shots off Target, Missed Crosses, Wasted Passes, Offsides, Headers Lost

I have quite a success in finding great overperforming strikers with this model.

Edited January 29 by st4lz

vonTrips · February 3

On 28/01/2024 at 21:55, BL1TZ_GT said:

Loving this. But curious if you have a range that's ok, good & poor like FMStag does in one of the links you got inspiration from.

Love the explanation,but I never know if the number in the KPI is OK, good or poor, its just a number to me haha

@vonTrips

To find out if a number is good or bad, I need to run simulations. Everyone does the simulations differently. For example, I found that in FM24 some KPIs are lower in the top 5 European competitions than in the others. And the numbers from South America, for example, are on average up to 10% higher...

I'm waiting for the latest game update, then I'll run a new simulation and make the data available.

My plan is to simulate (full details) the top league in the top 15 countries in Europe (by country club coefficient) + Brazil, Argentina, Chile, Mexico and USA.

Then it will be possible to say if the numbers are good or not... but in general we can say that it is possible to base on FM Stag's material or on what Mustermann Skin makes available. The numbers don't vary much. They just have to be numbers from the current version of the game. You can't use FM Stag's numbers for FM23 if you play FM24 yourself.

vonTrips · February 3

On 29/01/2024 at 14:06, st4lz said:

Main issues:

* Some of the stats are favoring players heavily in teams that have an advantage - it is almost impossible to beat the average xG or Pas % from the underdog team

* Opposite, some of the stats are much better in the underdogs, especially the defensive ones - all of the Tackles Completed, Key Tackles, Interceptions, Clearances, Blocks, Headers, etc are higher for teams that defend often

* Lower efficiency doesn't always mean worse performance - ex. Headers Won Ratio for two CB, one is the air superiority corner target, and the other stays back during Attacking Set Pieces. The first one may have 60% HWR with 12 Aerial Attempts per 90, the other may have 80% with 3 Aerial Attempts per 90. HWR is as useless as Conversion Rate or Shots on Target % (vs. xG), you can't just aggregate them and treat all situations equally (you discriminate low probability scorers, like players who score a lot of headers or long shots will have lower).

* Some of the stats are dependent on team tactics or playing roles, and can't be compared objectively - it is hard to compare a No-Nonsense CB with a CB when it comes to Pass %, the first one is encouraged to make safe clearances. Or to compare a Pressing Forward with a Trequartista when it comes to Pressures Attempted.

* Some of the stats are negatively correlated - there is often a low Pas % that goes along a high Key Passes. Creative players do take more risks, so their mainline Pas % is lower, which doesn't mean their passing is weaker. You need to identify those relations, and not try to put too much weight on Pas % when looking for a playmaker.

It's certainly an interesting note worth exploring. On the other hand, I've been picking players based on these KPI's for the third time in my career (3 diff saves) and so far I've been pretty successful (but I don't play for top teams). Sure there are mistakes, but not major ones.

If you look at my blog and find my career with Baník Ostrava, you will see that I bought a left-back (I guess David Schnegg) who had bad numbers this season because he played in the wrong team. But I also had historical data stored in DB, so I saw that he played the last two seasons well, but not the current one (he was on loan). So I bought him and he played great for me again.

Of course, if this was his first season (which I have dates from), I certainly wouldn't have given him a second glance. But that's how it works in real life, sometimes you just miss a good player. I don't want to have a magic formula, but I want to evaluate players and make my own decisions accordingly. One more important thing - only the primary KPI is used to filter players. The secondary and other KPIs are for deciding who could play better.

On 29/01/2024 at 14:06, st4lz said:

I generally like to create indicators that are constructed based on available statistics, I like the Tackle Quality (Tackles Attempted / Fouls Made) that you created.

Thanks, there are more of those custom (computational) indicators. And I have ideas for more, but those stats would have to be separated in-game - e.g. separate defensive and offensive headers, separate open-play xA from xA, etc.

On 29/01/2024 at 14:06, st4lz said:

I have quite a success in finding great overperforming strikers with this model.

I trust you on this and I like your approach, I may try it out myself accordingly. And I would be happy if you post your research somewhere here on the forum. I'm not saying that my way is the only right way, quite the opposite. I'm constantly improving it myself, I now have version 4 of the KPI for example. But I still need to validate it against the data after the game update.

st4lz · February 3

2 hours ago, vonTrips said:

It's certainly an interesting note worth exploring. On the other hand, I've been picking players based on these KPI's for the third time in my career (3 diff saves) and so far I've been pretty successful (but I don't play for top teams). Sure there are mistakes, but not major ones.

I am pretty sure you have success. People are playing attributeless and have success, even though it looks like a huge disadvantage. Paying attention to the data brings a completely different level of understanding to this game. Attributes are often misleading, there are a lot of hidden factors that play a role as well, the game is exploiting very well the biased judgment of our minds. The stats are much more reliable.

Also, if you know how to play, the game is not too hard, so I would be amazed if anyone who puts in a certain level of cognitive effort is unsuccessful.

3 hours ago, vonTrips said:

One more important thing - only the primary KPI is used to filter players. The secondary and other KPIs are for deciding who could play better.

This is huge, I found out the same. There is too much data to look at everything. Usually, you want a player to be great in just 1, 2, or max 3 areas, the other areas are either unimportant or he just needs to be around average. I saw people playing "Moneyball" saves and trying to search the replacements with 8-10 stat filters, the random process of the stats distribution and variety of player types, league quality, team performance, tactics, etc makes it very random who is going to match the criteria.

3 hours ago, vonTrips said:

I trust you on this and I like your approach, I may try it out myself accordingly. And I would be happy if you post your research somewhere here on the forum. I'm not saying that my way is the only right way, quite the opposite. I'm constantly improving it myself, I now have version 4 of the KPI for example. But I still need to validate it against the data after the game update.

I don't trust it myself, I constantly improve it as I see the flaws from time to time. The point is, if there is a wonderkid striker in the league or a really good one well above the others, he is always somewhere on top.

You can try it by yourself: https://moneyballfm.com/Strikers

I'd like to eventually create a page for every position, but while forwards, keepers, or center-backs seem easy, there is a lot of variance in central midfield. Maybe I will implement the KPIs that are composable instead. I would be happy to hear some ideas

Rodrigogc · February 3

I haven't read the whole thread, so ignore if this has been adressed already: Do you guys take into consideration that players might be being used in an underperforming role by the AI ? For instance, this:

Dribbles Made - a good indicator for Advanced Playmaker;

Indeed that is correct, but what if a player who's got good dribbling is being used in a role where he won't dribble at all ? How do you know he is good at it ? The AI often misuses players, playing St Pauli I've seen bayern using Wirtz as a winger, Marcos Leonardo in my save was being used as a CM even though he is a striker..

I've played attributeless for a while and adored it, but turned on attributes again because I didn't know how to evaluate my own newgens. Would like to play attributeless again though.

Edited February 3 by Rodrigogc

st4lz · February 3

@Rodrigogc Fair point, the stats are useful for finding overperformers, but they won't let you find an excellent player who is for various reasons not performing or simply sitting on the bench. The thing is, you skip them as 1 out of 1000 is useful compared to 1 out of 10 which produces good stats.

For the newgens, I would manage the youth team as well. I would play them in the same position for a long time to have comparable stats. Also, I'd use generic CB, DM, CM, AM, etc for youth team tactics, to not emphasize any specific role, so their stats should show what they are good at.

But youth stats are quite useless, they are not exportable and it is all about how well they perform in the senior team. That's what friendlies are for, you should watch the games and make your guesses.

janrzm · February 3

Just seen this, will read through later, looks really interesting.

There was a time when I'd happily create spreadsheets, export data, even creating charts. Although the geek in me enjoys this it does present me with two issues.

Firstly, its a little time consuming and for a game that already has a heavy time demand it's hard to justify for me.

Secondly. and probably a gripe as much as an issue is I'm employing staff in game to do this......... :lol: I was so excited when the data module was added but if you scratch below the surface it's not really doing the job you'd want it to. I want my staff to proactively present me with analysis in ways that relate to my tactic, player roles, opposition etc. No manager in any club with a data analyst is left to pour over raw data, make their own visualisations etc. The Data Hub, whilst necessary to the game is like a few other recent additions, a nice idea that hasn't been implemented anywhere near its potential.

Edited February 3 by janrzm

Andros · February 4

I wonder if this website could help out a bit here: https://themastermindsite.com/2022/04/13/evaluating-players-based-on-role-continuity/

It attempts to break down positions into roles and some stats that the creator believes indicate good players in those roles (see down the bottom for a link to specific roles). I guess there is some subjectivity here but maybe worth a look.

Those with a subscription to the athletic can see they looked at 18 player roles: https://theathletic.com/3473297/2022/08/10/player-roles-the-athletic/ Again there is subjectivity, EG a wide attacker unlocker could be inverted or a crosser, which is not really separated in the article. The article does not give specific stats on how each label was obtained.

Sorry if i am overcomplicating things, but I thought it was potentially useful!

vonTrips · February 4

20 hours ago, Rodrigogc said:

I haven't read the whole thread, so ignore if this has been adressed already: Do you guys take into consideration that players might be being used in an underperforming role by the AI ? For instance, this:

Dribbles Made - a good indicator for Advanced Playmaker;

Indeed that is correct, but what if a player who's got good dribbling is being used in a role where he won't dribble at all ? How do you know he is good at it ? The AI often misuses players, playing St Pauli I've seen bayern using Wirtz as a winger, Marcos Leonardo in my save was being used as a CM even though he is a striker..

I've played attributeless for a while and adored it, but turned on attributes again because I didn't know how to evaluate my own newgens. Would like to play attributeless again though.

You can get youth team stats out of the game. It's laborious, you have to take a special look at each competition, but it can be done. Check out my blog https://medium.com/@FMPanda/fm23-statistiky-rezervních-týmů-3bdb8b2ab7ec

But as @st4lz wrote, when a player plays out of position, or on the wrong team, you just overlook him - it can happen. Then it comes down to having historical data. Because you can look at how he's played over multiple seasons, or how performance has changed over the course of a season as well.

vonTrips · February 4

17 hours ago, janrzm said:

I was so excited when the data module was added but if you scratch below the surface it's not really doing the job you'd want it to. I want my staff to proactively present me with analysis in ways that relate to my tactic, player roles, opposition etc. No manager in any club with a data analyst is left to pour over raw data, make their own visualisations etc. The Data Hub, whilst necessary to the game is like a few other recent additions, a nice idea that hasn't been implemented anywhere near its potential.

Totally agree! DATA HUB doesn't work, just like exporting data from the game doesn't work. For example, in the Data Hub you will find that there is also Open-play xA, but there is nothing like that in the views.

That's actually the reason why I solve the KPIs, why I export everything and process it through a script. I have a primary KPI for player filtering and nothing else in the game. I keep a history of the data by making a backup of the file at the end of the season, and then possibly going back to it.

vonTrips · February 4

12 hours ago, Andros said:

https://theathletic.com/3473297/2022/08/10/player-roles-the-athletic/

This is what I read and partly based on. The problem is that this is a real world condition. In game it's different unfortunately, there's the Match Engine and therefore every role needs to be mapped towards the ME. Sometimes it goes against the grain.

12 hours ago, Andros said:

https://themastermindsite.com/2022/04/13/evaluating-players-based-on-role-continuity/

I don't know, but I'll definitely check it out. Wouldn't there be the same problem though, that it's a description of the real world, not a description of the game?

vonTrips · February 4

20 hours ago, st4lz said:

You can try it by yourself: https://moneyballfm.com/Strikers

Thanks so much, I'll definitely check it out as soon as I have time.

dunk105 · February 4

Not sure if mentioned but the SAS 24 skin has a FMStagStats tab on each players profile- its very, very useful for comparing relevant stats for players!

st4lz · February 4

@dunk105 The raw stats are nice to have, but they are not too useful when used as is.

Let's consider the scenario of looking for a new Pressing Forward.

The natural process of selecting a good PF is to take stats and order by Pressures Attempted per 90 or Pressures Completed per 90. But, I can guarantee, if you do that you will come up with poor results. It is really hard to say if the number, let's say 8.4, is much better than 7.6. The amount of pressures is highly correlated with the amount of time the team is in possession, meaning the teams that have only 35% average possession have much more chances to make pressures, than teams that have 65% possession. You may find out that players on top of this list are not even the PF, but other roles that low-possession teams are playing.

To balance that, you could discount the pressures amount by the average team possession stats. The problem is, the only way the data is aggregated in FM is per 90 minutes. There are other approaches in football analytics, and one of them is to present data per touch, which takes into account possession.

touches = passess_attempted + crosses_attempted + headers_attempted

You can aggregate touches per team per 90 to know the average team possession. Then, you can look at the pressures distribution by possession, and decide how much more pressures should be valued per 1% possession increase.

There are a lot of calculations to get there, but the output might still be not satisfactory, as you could also factor in Opponents PPDA, to make sure the player is really outstanding with pressures, and it is not influenced by the team's tactical focus.

My point is, that when looking for a forward position, we should primarily care about the main KPI which is the ability to score goals (F9 and similar creative forwards should be rather judged as attacking midfielders IMO). Then, if we isolate ~5 best options, we should factor in the pressing tactical requirement.

It is a completely different approach than comparing stats of two strikers in your own team. If they play enough amount of time, the level of opponents is similar and their teammates are the same (if there is the backup team playing only cup games or low tier opponents, it is still hard to compare with first team). You can compare not only the pressures output, but most importantly, what is the impact of the pressures on the overall team performance.

All of the above lead me to conclusion, that recruitment KPIs and performance KPIs should be different.

On 03/02/2024 at 15:04, vonTrips said:

I now have version 4 of the KPI for example. But I still need to validate it against the data after the game update.

It would be awesome if you put a v4 update, I would gladly take a look at what changed.

milenec11 · February 5

13 hours ago, st4lz said:

@dunk105 The raw stats are nice to have, but they are not too useful when used as is.

Let's consider the scenario of looking for a new Pressing Forward.

The natural process of selecting a good PF is to take stats and order by Pressures Attempted per 90 or Pressures Completed per 90. But, I can guarantee, if you do that you will come up with poor results. It is really hard to say if the number, let's say 8.4, is much better than 7.6. The amount of pressures is highly correlated with the amount of time the team is in possession, meaning the teams that have only 35% average possession have much more chances to make pressures, than teams that have 65% possession. You may find out that players on top of this list are not even the PF, but other roles that low-possession teams are playing.

To balance that, you could discount the pressures amount by the average team possession stats. The problem is, the only way the data is aggregated in FM is per 90 minutes. There are other approaches in football analytics, and one of them is to present data per touch, which takes into account possession.

touches = passess_attempted + crosses_attempted + headers_attempted

You can aggregate touches per team per 90 to know the average team possession. Then, you can look at the pressures distribution by possession, and decide how much more pressures should be valued per 1% possession increase.

There are a lot of calculations to get there, but the output might still be not satisfactory, as you could also factor in Opponents PPDA, to make sure the player is really outstanding with pressures, and it is not influenced by the team's tactical focus.

My point is, that when looking for a forward position, we should primarily care about the main KPI which is the ability to score goals (F9 and similar creative forwards should be rather judged as attacking midfielders IMO). Then, if we isolate ~5 best options, we should factor in the pressing tactical requirement.

It is a completely different approach than comparing stats of two strikers in your own team. If they play enough amount of time, the level of opponents is similar and their teammates are the same (if there is the backup team playing only cup games or low tier opponents, it is still hard to compare with first team). You can compare not only the pressures output, but most importantly, what is the impact of the pressures on the overall team performance.

All of the above lead me to conclusion, that recruitment KPIs and performance KPIs should be different.

It would be awesome if you put a v4 update, I would gladly take a look at what changed.

In touches you should add also dribble attempts and shots

milenec11 · February 5

@vonTrips

do you believe that it will possible to share with us the excel sheet script that mansion above ?

st4lz · February 5

2 hours ago, milenec11 said:

In touches you should add also dribble attempts and shots

Dribble failures to be precise, successful dribble may lead to other actions, like a shot.

Would be awesome if some kind of passes received and passes aimed were added to the game, we could analyze how good a player is as a pass recipient (the first touch impact). It would be very useful to counter gegenpress. We should add possession won to it and have touches. But I doubt possession won/lost is a reliable stat now, I think it may be broken.

The same for crosses received/aimed for separating offensive headers.

vonTrips · February 5

17 hours ago, st4lz said:

All of the above lead me to conclusion, that recruitment KPIs and performance KPIs should be different.

This is a perfect idea! 😉 I've been toying with the idea of not evaluating my own players this way for a while now. So far I'm doing it because it seems easy, but...

The second important truth that emerged in your post is the influence of tactics/role and team quality on metrics values. I address this mainly with CBs, where high clearences and blocks numbers are found in a lot of material. It's just that my CBs have an AR of 7.00+ and those numbers are extremely low. That's why, for example, I don't want to use CLR and BLK specifically when looking for CBs. Because high numbers say nothing about a player's performance. For the same reason I don't address Press Att, Press Com and Press Ratio at all.

17 hours ago, st4lz said:

It would be awesome if you put a v4 update, I would gladly take a look at what changed.

Of course I will post it here, at least again as a table. A more comprehensive text will be on my blog (in Czech lang), but this time I want to publish the values that I measure myself. From the whole spectrum I take the 85th percentile as the ideal (100%), or the 15th percentile (e.g. for goals conc.).

Edited February 5 by vonTrips
add words

vonTrips · February 5

3 hours ago, milenec11 said:

do you believe that it will possible to share with us the excel sheet script that mansion above ?

I have it in a spreadsheet just for the format. There are no formulas/scripts. I used to build some sort of tool in excel, but it proved to be unnecessarily complex/necessary to extend/maintain. So I don't see any reason to share it, if I can, I can put it here as plain text.

vonTrips · February 5

52 minutes ago, st4lz said:

But I doubt possession won/lost is a reliable stat now, I think it may be broken.

Again all I can say is - great idea. This is exactly what I addressed in our Discord discussion. That no one really knows how Poss Won/Lost is made. That I actually only know how successful the passes are, but I can only tell which ones were unsuccessful (forward? sideways? backwards?) from the graphs. Most importantly, I have no idea if the loss is from centers, dribbles, etc. Knowing this data is important for me to be able to evaluate players correctly.

vonTrips · March 5

Hi everybody!

Especially @st4lz @milenec11 @Rodrigogc @dunk105 @BL1TZ_GT

Here is my new version of KPI:

Full post about it with an explanation is on my blog - https://medium.com/@FMPanda/moneyball-klíčové-metriky-4-8f2e4bb4e3e5 - and yes, it's in Czech language. But Czech language is beautiful, trust me 😉😂 Alternatively, use DeepL for translation.

milenec11 · March 5

@vonTrips

You have absolutely right on that but I believe that we need a little bit to narrow them down and to make them comparable.

what I mean is that in modern football , and of course this is a game but we try to make it sim not arcade I believe, we do not have position or roles but we have covering areas . A good player to be good must playing at least 3 position good enough so I believe that we need to categorize those KPIs differently.

and the second is that we must make this raw data comparable . Ok we find a player that plays as we want and has the stats we need in our tactic to perform but where he achieve those numbers ? Will perform in my team in my league the same ?
so we need to calculate them as plain numbers or value to compare with the average values of our current league . Am I wrong ???

Edited March 5 by milenec11

vonTrips · March 5

2 hours ago, milenec11 said:

You have absolutely right on that but I believe that we need a little bit to narrow them down and to make them comparable.

That's your point of view. What I have posted here is my perspective. If you want to make any changes, make them 😉 That's all I can tell you

And of course you're right, the numbers are not easily comparable. You can't compare 0.2 xA achieved in the Premier League with the same number achieved in the Austrian Bundesliga. But what I've listed here is just a guide. The final decision is on each manager (like as in real life).

st4lz · March 5

13 hours ago, vonTrips said:

Full post about it with an explanation is on my blog - https://medium.com/@FMPanda/moneyball-klíčové-metriky-4-8f2e4bb4e3e5 - and yes, it's in Czech language. But Czech language is beautiful, trust me 😉😂 Alternatively, use DeepL for translation.

I speak Polish, so for me Czech is not only beautiful, but also funny

Isn't Save Ratio the same as Shots per Conceded, but inverted?

Overall, I think what you picked is very reasonable and well thought. My question is, where to go from here?

Do you think it is possible to create a ranking for each stat per position and combine all the results together using weights for [P, S, O] category? We could end up having a list of best performers for each position per league. Would the results make sense and you trust the order?

vonTrips · March 6

9 hours ago, st4lz said:

I speak Polish, so for me Czech is not only beautiful, but also funny

To moze spotkamy sie w Biedronce 😉😂😂😂

9 hours ago, st4lz said:

Isn't Save Ratio the same as Shots per Conceded, but inverted?

Good point. The problem is that I didn't edit the GK this time, I just added Clearances into others metric. But if I remember, I wanted to differentiate that the goalie has 75% SV and it's 1 goal from 4 shots. But then there's GK who got 5 goals, but from 20 shots. The expected save ratio and xG faced metric would be a great solution for this, but unfortunately you can't get them from the game.

I'm afraid I didn't get the result right. Maybe it should have been the Shots Faced Per 90 metric. I'll have to take another look at it.

9 hours ago, st4lz said:

Do you think it is possible to create a ranking for each stat per position and combine all the results together using weights for [P, S, O] category? We could end up having a list of best performers for each position per league. Would the results make sense and you trust the order?

I would like to publish a tool that will process an HTML file and then say these numbers. And from what I've tested (so far), those numbers can be trusted. But as I've written before, the final decision is always on manager

Edited March 6 by vonTrips

milenec11 · March 6

@st4lz @vonTrips

the most safe (if is safe enough at scouting in soccer ) for GKs raw stats are GAA (goal against allowed) and PshtsTPRV ( POST SHOTS ON TARGET PREVENT)

The first is a custom stat which has a formula that the stats in game allow us to create

the second bohuzel is imposimble because we know the formula but the game doesnot gives us the shots on target that a GK faced

we can ofcourse to add rhe shots that he conceded plus tipped helded or parried but we do not have info for those that hit on the bar ..

ZacSr · March 6

On 05/03/2024 at 09:45, vonTrips said:

Hi everybody!

Especially @st4lz @milenec11 @Rodrigogc @dunk105 @BL1TZ_GT

Here is my new version of KPI:

Full post about it with an explanation is on my blog - https://medium.com/@FMPanda/moneyball-klíčové-metriky-4-8f2e4bb4e3e5 - and yes, it's in Czech language. But Czech language is beautiful, trust me 😉😂 Alternatively, use DeepL for translation.

I really liked your distinction between primary and secondary metrics. Although some points are debatable, but in general they represent the key characteristics of players.
Despite the Czech language, I was able to read this with my browser

Edited March 6 by ZacSr

ZacSr · March 6

1 hour ago, milenec11 said:

@st4lz @vonTrips

the most safe (if is safe enough at scouting in soccer ) for GKs raw stats are GAA (goal against allowed) and PshtsTPRV ( POST SHOTS ON TARGET PREVENT)

The first is a custom stat which has a formula that the stats in game allow us to create

the second bohuzel is imposimble because we know the formula but the game doesnot gives us the shots on target that a GK faced

we can ofcourse to add rhe shots that he conceded plus tipped helded or parried but we do not have info for those that hit on the bar ..

Could hits on the bar be counted as shots past the target?

milenec11 · March 6

3 minutes ago, ZacSr said:

Could hits on the bar be counted as shots past the target?

A shot hitting the post does not count as on target but if interference a defender or any other situation then counts .

exp you make an of balance shot and a defender put his foot to block the shot and the ball after that hit the post . Despite the fact that the tranjectory of the ball was of target a player interfered and after hit the post this counts as shot on target .

is complicated that’s way is impossible I believe for the game to include this raw stat .

we can imaginary to simplify the formula by putting the stats that we have in game
Post target shots save % = goals conceded / ( goals conceded + tipped + parried+ saved )

ZacSr · March 6

4 minutes ago, milenec11 said:

A shot hitting the post does not count as on target but if interference a defender or any other situation then counts .

exp you make an of balance shot and a defender put his foot to block the shot and the ball after that hit the post . Despite the fact that the tranjectory of the ball was of target a player interfered and after hit the post this counts as shot on target .

is complicated that’s way is impossible I believe for the game to include this raw stat .

we can imaginary to simplify the formula by putting the stats that we have in game
Post target shots save % = goals conceded / ( goals conceded + tipped + parried+ saved )

Yes, of course, those raw stats should be included. Otherwise it will not be a very correct stats.

milenec11 · March 6

Also the game does not include an other raw stat crucial for a Gk aspect.

the aerial duels ,

we have an attribute such as aerial and command of area or rushing out that give us an idea if a GK took at of his line to cover a cross in air or in ground but missing the fact if this attempt will be successful or not and at what number .

unfortunately the game does not give a big attention at Gks for me the raw data that gives is minimal.

perhaps is the only position in FM that playing the game without attributes the ratio of scouting and luck is present a lot.

milenec11 · March 6

2 hours ago, vonTrips said:

Good point. The problem is that I didn't edit the GK this time, I just added Clearances into others metric. But if I remember, I wanted to differentiate that the goalie has 75% SV and it's 1 goal from 4 shots. But then there's GK who got 5 goals, but from 20 shots. The expected save ratio and xG faced metric would be a great solution for this, but unfortunately you can't get them from the game.

I'm afraid I didn't get the result right. Maybe it should have been the Shots Faced Per 90 metric. I'll have to take another look at it.

Mmmmmm a big conversation on that . As the game is and with the stats we get the SV% is purely dependent from defense performance.

if a GK faced 4 shots and conceded 1 is the some if Face 20 and conceded 5 but the game does not give us the data to knew what shots was that ? 1vs1 ?own goals ?distance shoots? pure chances ? It was the some situations that those 2 competitors GKs faced ?

how we can found out which one perform better ? Is a way , if it was possible to calculate or had the Post-shot expected goals minus Goals allowed per 90 minutes.

the Goals allowed per 90 min is easy to find GAA= ( Goals allowed *90)/ actual minutes played

but the post-shot expected goals go figure…..

so for me the only thing that you can do for GKs is simply to stats which can read them as one

xPV% and SV% if the difference is + the the GK is over performance if is minus then his is underperforming. What else ?!?

Edited March 6 by milenec11

milenec11 · March 6

And to finish this monologue and the KPI or moneyball or cover attributes or only stats scouting type of game that skins or excels or anything else can provide to play is a big in game number that in big rate covering all …..AVERAGE RATING.

an average rating is created by the SI using its one type to calculate the performance of a player within the game using the game facts . For example a goals counts as 0,9 and an assist 0,4 etc etc ….

So search for a player within the bunget you have the role you need the league that can be compared with your’s team actually playing and took the biggest AVR . This player about 90% is the gay you need

vonTrips · March 6

55 minutes ago, milenec11 said:

And to finish this monologue and the KPI or moneyball or cover attributes or only stats scouting type of game that skins or excels or anything else can provide to play is a big in game number that in big rate covering all …..AVERAGE RATING.

an average rating is created by the SI using its one type to calculate the performance of a player within the game using the game facts . For example a goals counts as 0,9 and an assist 0,4 etc etc ….

So search for a player within the bunget you have the role you need the league that can be compared with your’s team actually playing and took the biggest AVR . This player about 90% is the gay you need

This is absolutely not true, within Moneyball you have to look for undervalued players who will perform well and won't cost you a lot of €. And the second thing is that Average Rating is offensively minded! So a CB who scores goals (e.g. from SP) will have a better AR than a CB who defends better. But which CB is better?

I've tried searching for players by AR and you often get a hit, but you pay a big price. But if you're looking for a specific player (good defender/tackler/header), AR won't help so much to you. That is why this post was created and why I am finding KPIs.

As for goaltending, that's a longer discussion, but I also agree that it's a lot about luck. I still need to fine-tune the GK ratings and their KPIs.

st4lz · March 6

3 hours ago, vonTrips said:

I would like to publish a tool that will process an HTML file and then say these numbers. And from what I've tested (so far), those numbers can be trusted. But as I've written before, the final decision is always on manager

I wanted to do it on my website, but if you want to do the app yourself, I won't spoil the fun for you. I'll figure out my own. Just don't forget to pin me like this time when you are done

@milenec11 Save % is indeed irrespective of the chances faced. xG-Prev doesn't say how many chances you had to increase the number. A fair comparison would be xG-Prev/ShotsFaced, as it takes both quality and quantity into account.

The optional stats for me are either Pas % for short distribution (but mostly to eliminate bad passers rather than reward the best ones), or KeyPasses/90 if I would want the goalkeeper to initiate counter-attacks (1-2 per season may be luck, but seeing more would be interesting).

vonTrips · March 6

7 minutes ago, st4lz said:

I wanted to do it on my website, but if you want to do the app yourself, I won't spoil the fun for you. I'll figure out my own. Just don't forget to pin me like this time when you are done

What I have is an online application (PHP - Nette). It will be publicly available. I want to deploy it by the end of this week. Of course, I'll pin you

And if you have an idea how to better evaluate GK, or how to replace ShotPerConc, please help me. Even the discussion itself is important. It helps to find an interesting idea (solution) that I wouldn't have come up with otherwise. I'm not omniscient. I like to read other people's thoughts.

Sign In

Key metrics (KPI) for each position

Recommended Posts

Link to post

Share on other sites

What does PPDA stand for?

What is PPDA? What is passes per defensive action?

How is PPDA calculated?

Link to post

Share on other sites

What does PPDA stand for?

What is PPDA? What is passes per defensive action?

How is PPDA calculated?

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites

Link to post

Share on other sites