FM20 xG - Am I Scoring as Much as I Should Be?

[Joey Numbaz]

FM20 I feel like I am not getting enough goals based on my chances.

First the stats (per game) - through 7 games, 5 EPL, 1 Community Shield, 1 Champions League

Goals: 2.14
Shots: 23.57
On Target: 13.00
Clear Cut: 2.43
Half Chances: 3.57
Chances Created: 4.71 (interestingly this does not equal CCC+HC - it's the players' individual "chances created" added up)
Long Shots: 3.71
Crosses Comp: 15.00
Penalties: 0.29
Penalties Scored: 0.14

Here are the numbers for the rest of the league (average 4.89 league, 1 other game, mostly league Cup, several vs. each other)

Goals: 1.19
Shots: 13.19
On Target: 6.77
Clear Cut: N/A (would have to manually add them from boxscores)
Half Chances: N/A (would have to manually add them from boxscores)
Chances Created: 2.35 (interestingly this does not equal CCC+HC - it's the players' individual "chances created" added up)
Long Shots: N/A (would have to manually add them from boxscores)
Crosses Comp: 6.77

I guess 2.14 goals isn't that bad. But I decided to go crazy. I have been wanting to do this for years. I spent the last 24 hours (I slept for 4 hours and spent a little time with the family) - it was mostly spent doing data entry. NOTE - that was when I started writing this post Friday night - turns out I had a lot more work ahead of me than I realized. I have been working on this for 3 full days almost non-stop at this point.

For all 380 premier league games (760 team games) from the 2027-28 season I recorded:

Goals, CCC, HC, Long Shots, Shots, Shots on Target, Shots off Target, Blocked Shots, Penalties, PenMissed, Own Goals

Let's see if I can make a simple xG model to see how we are finishing.

Expected goals for Penalties are easy. 82 penalties, 60 scored.

xGPen = 0.73

So I subtracted penalties from every game for goals, CCC, and shots on target (I get that I'll mischaracterize the handful of PK that aren't goals or saves).

NOTE: Unfortunately, I can't really pull out set piece goals. There isn't anything in the game match stats to say how many shots, chances and goals game from set pieces, which is a bummer.

OK moving on the good stuff.

 

Those are the average stats for each game where a team scored that number of goals. BoxNChSH are shots from in the box that weren't CCC or HC. SHonTnP are non-penalty shots on target.

I know many people think CCC and HC are meaningless, but CCC are not. Half chances through, well we'll get to that a little later.

Also note getting shots means getting goals. This at least passes the smell test.

Next I ran a regression and it came out really neat. Better than I expected.

The Coefficients on the bottom left are what we care about the most. It’s a decent R-square too, .63, and the standard error on any game is 1.057 goals which isn't bad. Obviously there is a ton of variance in any one game.

Notice, shots in the box that aren't chances end up being more valuable than half chances! I am starting to think half chances really are pretty meaningless.

You can use Tolerance and VIF to test for colinearity. Tolerance ranges from 0-1, higher is better. Anything below .2 is a flag and below .1 is a giant flag. VIF is just the reciprocal of tolerance, you want that to be below 10 for sure and preferably well below 5.

 

So we run into a problem here - there's a bunch of colinearity. Specifically non-chance shots from the box, shots on target, and shots off target, to a lesser but important extent long shots too. This breaks the regression. You've got all of these things that are basically overlapping each other and a linear regression can't really figure out how to weigh them.

So I decided to do two separate regressions. One is CCC, HC, non-chance shots from the box, and long shots.

Much better!

Next thing you can do is a LASSO regression, which can tell you if any of the variables aren't pulling their weight. In that case you might want to drop those, or combine it with something else. In a Lasso regression all variables will eventually get to zero as you increase Lambda. It's an iterative thing. 

Or in line chart form

 

That's good, they all drop off around the same time. Note Lambda is based on the number of samples and they are all still there at N*.01 which is 7.59 (we have 760 samples, one for each team game).

However, it turns out half chances really weren't any different than other shots from the box that weren't CCC.

Note non-chance shots in the box are actually slightly higher than half chances! This could not be the case in this big of a sample if they were significantly better shots.

So I tried it another way. Surely if you treat all shots in the box as one variable, and then throw in CCC and HC, a half chance should at least be indicative of a better shot in the box right? So if a shot in the box is worth say .069 xG, then a CC would make that shot .328 better - which makes a CCC actually worth .397 (since they are all also shots in the box).

You would think, but it doesn't. Trying it this way I get the same thing - Half chances come out worse than shots in the box. -.0045 worse. Which I'm sure is just a sample size thing I can't imagine they are actually worse than any other shot even as negligibly as -.0045 (that would mean you'd lose one goal for every 222 half chances, as opposed to shots in the box). But I am also pretty sure there aren't any better either.

Also note the T-Stat becomes garbage (higher shows more significance). I am pretty convinced at this point that half chances are just noise.

So I'm just going to combine them into one making the regression CCC, non-CCC shots from the box, and long shots.

That gets us here:

So the first xG formula is:

(CCC-Pen) * .396 + (Shots-CCC-LongSH-Pen) * .068 + LongSH * .054 + Pen * .73

This one answers the question - am I getting good shots?

Also note, this implies that about 1 in every 18 long shots is scored. That sounds about right and passes the smell test.

****************

On to the next one. This will look at the quality of the shooting itself.

For this one I am planning on CCC, Shots On Target, Shots Off Target, and Shots Blocked - the last two may end up being dropped. CCC and Shots on Target will overlap some, I have no way of splitting those out. So we may run into some colinearity. Let's check.

NOTE IN THE ORIGINAL POST, I HAD A DATA ERROR in this section - I was double counting the shots off target, there was a mistake from when I was combining columns earlier. It did not change much, but I have fixed it, as of 9:20 PM EST January 31, 2021.

Good to go.

Slight digression. If you had colinearity, and you have no way to combine the offending variables, and you also cannot remove them, you could do something called a Ridge Regression. I did those too along the way when I tried to get them all into one formula, but then I realized it'd be better to split these out into two separate formulas so here we are.

Now lets do the LASSO

While it's running - it iterates 10,000 times for each square in the chart - if you add the "Real Statistics" analysis pack to excel you can run this kind of stuff. It's pretty slick. And very free. Some links:

https://www.real-statistics.com/multiple-regression/ridge-and-lasso-regression/lasso-regression/

https://www.real-statistics.com/multiple-regression/collinearity/

https://www.real-statistics.com/multiple-regression/ridge-and-lasso-regression/ridge-regression-basic-concepts/

There's other stuff there. It's an awesome website.

Back to the numbers:

They all look useful. Shots off target and blocked a little less so maybe we end up cutting those. We could probably remove them without much of an issue.

Here's the regression. It's not bad. Shots off target and blocked have pretty low T-Stats though. Note that shots blocked do kind of nudge your xG up a little bit. Especially relative to shots off target. This would imply that taking an off target shot is worse than not taking any shot at all, since it's negative. I wonder if that's because you always lose possession with an off target shot, and they a typically from further out, whereas with a blocked shot you don't necessarily lose possession. Blocked shots are indicative of having the defense pinned in their area also which is generally a good thing.

You want the absolute value of the t Stat to be at least 1.96 for the variable to be meaningful with this sample size.

So the shots off target and blocked don't really help.

If I combine them, the t Stat becomes 0.295.

Looks like it is best to just leave them out.

I am going to try adding half chances back in though. Maybe this is where it might show something?

No it does not. Half chances are still hot garbage.

So here we go, this one is simple:

The second xG formula is:

(CCC - Pen) * .276 + (Shots on Target - Pen) * .140 + Penalties * .73

If you get the CCC on target, you'd get credit for both, total .416

**********

So back to the original question - has my team been unlucky, bad a shooting or some such thing? Let's look at formula 1:

xG1 = (CCC-Pen) * .396 + (Shots-CCC-LongSH-Pen) * .068 + LongSH * .054 + Pen * .73

Goals: 2.14
Clear Cut: (2.43 - 0.29 [pen]) * .396 = .849 xG
Half Chances: 3.57 = MEANINGLESS
Shots: 23.57 = (removing CCC, LongSH, Pen) we get 17.142 * .068 = 1.166 xG
On Target: 13.00 = Not applicable for this formula
Long Shots: 3.71 * .054 = .201 xG
Penalties: 0.29 * .73 = .209 xG
Penalties Scored: 0.14

Add it all up, based this shot quality, you'd expect 2.42 goals. So yes, we have been underperforming.

Now let's do formula two:

xG2 = (CCC-Pen) * .276 + (Shots on Target-Pen) * .140 + Pen * .73

Clear Cut: (2.43 - .29 [pen]) * .276 = .591 xG
On Target: (13.00 - .29 [pen]) *.140 = 1.780 xG
Penalties: .29 * .73 = .209 xG

Based on a combination of great chances and getting shots on target, it looks like 2.58 xG

***********

So now onto the next question - WHY? WHY IS THIS HAPPENING! Why am I not scoring as much as I should. Could it be the tactic. Mainly the footedness of the players?

This post is long enough. I'll ask questions in the next one.

 

 

Edited February 1, 2021 by Joey Numbaz

[Joey Numbaz]

OK why?! Why am I not finishing well? It has been very frustrating.

I am playing with a fun tactic that Rashidi created, 3 strikers, an AMC behind and two defensive wingers. The complete forwards stay wider.

I have amazing attackers, Haaland, Lautaro, a world class Regen. The AMC is usually Gonçalo Ramos too. And they know the tactic perfectly and their roles nearly perfectly.

         CFs  AFa  CFs

                 AMs

DWs                       DWs

                 HBs

    BPDd    Ls       BPDd

Could footedness be a problem? Do I want the CFs to be on the side of their strong foot or not? I have gone with this because they are wide they tend to put in a lot of crosses also.

I can't really think of any other reason I am underperforming and I am very open to ideas.

Edited January 31, 2021 by Joey Numbaz

[Joey Numbaz]

The four best teams in the league by far were Tottenham, Liverpool, Manchester City and Everton. There was a huge gap in the standings between 4th and 5th. Manchester United was pretty good and so was Brighton. Both also exceeded their xG.

Let's look at how those xG formulas did for the league. Note, I am Everton.

Look at who outperformed their xG1 and xG2 by miles! It was the 3 other great teams, and Manchester United did pretty well in that respect also. I switched my tactic halfway through the season to the current one. I broke my team out based on each tactic. There were 14 games with the 343, 21 with the 442. So I projected both out to 38 games also.

Tottenham, Liverpool, Manchester City all scored 15-25 more goals than expected looking at the two formulas. Why didn't Everton?! With my world class strike force, I really should be outperforming my xG also. In real life great teams outperform their xG too. Because great teams have players who are really good at kicking a soccer ball and making it go where they want it to go. Significantly better than even most players in the Premier League.

So I am now even more convinced there is a tactical issue with my finishing.

Heck look at how my 442 did compared with the 343. Huge difference there! The 343 would get me 88 more shots in the box, 90 more shots on target, 16 fewer long shots over the course of a season, but also 9 fewer goals. So I really am stumped for why we aren't finishing as well.

 

Edited February 1, 2021 by Joey Numbaz

[crusadertsar]

The tactic looks completely unbalanced actually and way too top heavy. I mean what's the point of playing with a halfback when you are not using aggressive wingbacks? Halfbacks just act like regular DMs when you are not using wingbacks. So your halfbacks will just be blocking the space for your libero. And with that setup there's not much linkup between your attack and defence. And your midfield is non-existent. So depending on your team mentality your three strikers might be quite isolated and dependent on occasional long pass from your libero or Half-back. 

Besides that there's probably alot of pressure on your defensive wingers as they basically are responsible for both defending the flanks and supporting your strikers. Unless you made some big changes, I don't believe Everton has anyone that can do that.

So based on that I doubt the issue is with your finishing. I bet if you switch to a more logical, balanced tactic like a good old 4-4-2 you would start performing much better.

Edited February 1, 2021 by crusadertsar

[Joey Numbaz]

I know it looks unbalanced, but it really doesn’t play that way. It creates tons of chances, more than anyone in the league, and it is very solid defensively. I gave up fewer goals than anyone in the league. It’s a wide diamond midfield, which is fine, nothing crazy revolutionary, though it is a bit weird to look at.

The DWs get up and down and even score some goals, they are basically pushed up WBs.

The HB is almost like a combined Regista/DC-stopper.

It needs great players for sure, but I have them. It’s 2028 now and I’ve won 4 of the last 7 champions leagues. I change my tactic up every 2 or so years so to keep it fun and we’ve won all sorts of ways. Everyone has work rate and teamwork which also helps.

The only thing that’s is an issue is my players are missing too many of the good shots this creates.

Edited February 1, 2021 by Joey Numbaz

[crusadertsar]

11 minutes ago, Joey Numbaz said:

I know it looks unbalanced, but it really doesn’t play that way. It creates tons of chances, more than anyone in the league, and it is very solid defensively. I gave up fewer goals than anyone in the league. It’s a wide diamond midfield, which is fine, nothing crazy revolutionary, though it is a bit weird to look at.

The DWs get up and down and even score some goals, they are basically pushed up WBs.

The HB is almost like a combined Regista/DC-stopper.

It needs great players for sure, but I have them. It’s 2028 now and I’ve won 4 of the last 7 champions leagues. I change my tactic up every 2 or so years so to keep it fun and we’ve won all sorts of ways. Everyone has work rate and teamwork which also helps.

The only thing that’s is an issue is my players are missing too many of the good shots this creates.

That might be all well. Although its still wouldn't be a formation I would use simply because it looks like something created with a very specific team in mind and not exactly a plug in tactic. I would be curious to see the instructions, especially the defensive line and line of engagement. I'm guessing you are running into some very defensive opponents, parking the bus. And with three strikers you are probably contesting the final third with not enough late runners to exploit space better. That's why I've never been a fan of such three striker formations that try to batter the opposition into submission.

[Guest]

You're racking up your XG through sheer amount of shots, you have almost had 100 more than the next highest team. What makes you think the chances you have are quality? You're saying people are wrong to suggest CCC's are meaningless but CCC does have a fundamental flaw to the point I have turned it off my stats because it is so misleading. 

[HUNT3R]

You're asking questions that only really you can answer. "Could footedness be a problem?" Yes, it could. Is it in this case? You'd have to analyse your chances. You have done a fair amount of work to look at the stats and conclude that you're not converting enough. That's half the job. You've identified a possible issue, but now it's down to analysing the type of chances you create and who is taking them. 

Looking at your total shots / goals in the table you posted, it's around 10%, which is a little low, but not crazily so. In your OP, you're at around 9%, so you haven't started the season well compared to the other season. In FM20, I usually see between 7% and 13% conversion rates. In my last season in FM20, I had 13% (tied best) and 2nd placed PSG had only 9% (tied for 3rd). The bottom team were as low as 6%. In my last 6 or 7 seasons, I was consistently scoring at around 11-12% (2 seasons ago, I had 11% which was 2nd best and won the league). 

In this season I created 50% (and it was the same 2 seasons ago) more 'chances' (not sure what counts as a chance) than any other team. My tactic analysis (I haven't manually counted) shows I created 52 CCCs in the last 30 league matches and converted 41 of them. I've scored 116 goals in 33 matches. 2 seasons ago, it was worse.  52 CCCs and 37 scored in the last 30 league matches. 111 goals in 38 matches. It was similar 2 seasons before that as well.

Maybe some of those stats helps in a way to compare to what you're getting. I'm was predicted 1st in all 3 seasons though. 

If I go back to the season before my first CL win, I was predicted 3rd. 14% conversion rate which enabled me to score 108 in 37 matches.

In all those seasons, I averaged around 20-21 shots per match.

If I go back another season, one before I won the league (and 7 seasons ago in total), I was predicted 2nd and finished 3rd. Conversion rate was only 11% and both the teams (PSG and Lyon) who finished above me had better rates. Oddly, even though we scored a similar amount of goals etc, my 'chances created' was again about 50% higher than all but 1 team and they finished 9th. CCC-wise, 109 created and 82 scored. I had a different tactic (and obviously different players) to 7 years in the future, but shots/game, conversion rates, CCC conversions etc were all similar. I was able to afford better and better players as the years went on. Although it didn't necessarily show in the stats, we dropped fewer points over the years. That could be down to a tactic change, but perhaps also my players being more consistent and/or better when it's really needed.

A big thing to look at in a formation like yours is whether you have a lot of headed chances instead of shots. Headers have a much lower conversion rate in general. Blocked shots would also be an indicator that you're not creating enough space or you're taking too long to use the space you have.

Maybe have a look at hidden attributes like Consistency and Big Matches, both of which would show up in a scout report if it's either good or bad. If it's 'in the middle' it wouldn't be mentioned.

I never really look at CCCs only. At the very least, if I do look at stats and at that specifically, include HCs. In general, I rely on my own eyes. Shot conversion rates can be a bit useful, but only as a broad overview to give you an idea over the medium/long term history.

[Joey Numbaz]

Sure. Full credit, this is an @Rashidicreated tactic, I may have made a tweak or two, I recreated it from pictures, it wasn't a download. It's fun and way more solid defensively than I ever would have figured. It's not like we don't score any goals. We just score a lot fewer than we should.

The PIs I have are stay wider for both CFs. The front three all have close down more, tackle harder, mark tighter. Haaland is left footed, the other 3 up front are all right footed. Spahic is a monster - possibly the best player I've ever had. He was 3rd in the Ballon d'Or

 

Mael Gaudin is also a world class player on the left

Here is the tactic, with the star ratings so you can see how good the players are. Basically I rotate Gaudin, Camavinga, Tonali, and Florentino for the 3 spots, with Camavinga playing DW/HB, Tonali playing L/HB and Florentino as L, depending on who is in the lineup.

Th

Edited February 1, 2021 by Joey Numbaz

[Joey Numbaz]

Deleted, the missing post is back. Weird.

 

Edited February 1, 2021 by Joey Numbaz

[Joey Numbaz]

9 minutes ago, Crazy_Ivan said:

You're racking up your XG through sheer amount of shots, you have almost had 100 more than the next highest team. What makes you think the chances you have are quality? You're saying people are wrong to suggest CCC's are meaningless but CCC does have a fundamental flaw to the point I have turned it off my stats because it is so misleading. 

I don't find any evidence of that. CCC very clearly across the entire league is strongly associated with goals. There is no way around that based on the analysis I did in the first post. I think the onus is on people who say they are flawed to prove it.

Half chances on the other hand - yes, they don't mean much of anything, other than any other shot from the box.

[Guest]

This is listed as a CCC and it's not a CCC imo when a attacker bangs a shot off of the defenders legs who is stood right in front of him, there are other issues too.

I literally put so little stock in CCC's that I had no idea that I am in the bottom half for them.

Edited February 1, 2021 by Crazy_Ivan

[Joey Numbaz]

28 minutes ago, HUNT3R said:

Looking at your total shots / goals in the table you posted, it's around 10%, which is a little low, but not crazily so. In your OP, you're at around 9%, so you haven't started the season well compared to the other season. In FM20, I usually see between 7% and 13% conversion rates. In my last season in FM20, I had 13% (tied best) and 2nd placed PSG had only 9% (tied for 3rd). The bottom team were as low as 6%. In my last 6 or 7 seasons, I was consistently scoring at around 11-12% (2 seasons ago, I had 11% which was 2nd best and won the league). 

In this season I created 50% (and it was the same 2 seasons ago) more 'chances' (not sure what counts as a chance) than any other team. My tactic analysis (I haven't manually counted) shows I created 52 CCCs in the last 30 league matches and converted 41 of them. I've scored 116 goals in 33 matches. 2 seasons ago, it was worse.  52 CCCs and 37 scored in the last 30 league matches. 111 goals in 38 matches. It was similar 2 seasons before that as well.

I think the big thing is that the other great teams in the league, all way outperform their xG compared to the average. Which makes sense since they have they superstars up front. I feel like I should be doing that also since I too have some of the best players in the world, but I am not. I should be converting chances much higher than the average - which tells me something is broken.

32 minutes ago, HUNT3R said:

A big thing to look at in a formation like yours is whether you have a lot of headed chances instead of shots. Headers have a much lower conversion rate in general. Blocked shots would also be an indicator that you're not creating enough space or you're taking too long to use the space you have.

I do have a lot of headed chances. For sure. That could be part of it. I try to get tall players with good heading but this makes a lot of sense. Even with low crosses ticked. Nothing too crazy in here re: blocked shots. I highlighted the other great teams plus Manchester United, just to show how they are all better than xG, as expected. My blocked shots don't look out of line with them, and while I have a lot more shots off target, I also have a lot more on target too.

Maybe I could turn off the hit early crosses and see how that works?

[Guest]

Here is another btw....listed as a CCC, XG listed it as a 0.14 chance

 

 

[Joey Numbaz]

While they are not perfect, and I am sure there are cases where things like a defender in the way are an issue, this is tough to argue with (CCC-P is clear cut chances not counting penalties, and goals also does not include penalties).

That's a pretty good association there. They definitely are not meaningless, even if they are not perfect. It's a good rule of thumb, at least on FM20, that every 5 CCC should result in 2 goals (just from the CCC - you will get other goals too of course) for an average team, depending on the quality of the players shooting, of course.

Edited February 1, 2021 by Joey Numbaz

[Guest]

10 minutes ago, Joey Numbaz said:

While they are not perfect, and I am sure there are cases where things like a defender in the way are an issue, this is tough to argue with (CCC-P is clear cut chances not counting penalties, and goals also does not include penalties).

That's a pretty good association there. They definitely are not meaningless, even if they are not perfect. It's a good rule of thumb, at least on FM20, that every 5 CCC should result in 2 goals (just from the CCC - you will get other goals too of course) for an average team, depending on the quality of the players shooting, of course.

Very next game, played the first half I have one CCC and this is it. XG rates it 0.10, so I really beg to differ. If i was doing what you have and well done btw on the top notch effort put in I would take the CCCs on a case by cases basis when working your XG out.

[HUNT3R]

34 minutes ago, Joey Numbaz said:

I do have a lot of headed chances. For sure. That could be part of it. I try to get tall players with good heading but this makes a lot of sense. Even with low crosses ticked. Nothing too crazy in here re: blocked shots. I highlighted the other great teams plus Manchester United, just to show how they are all better than xG, as expected. My blocked shots don't look out of line with them, and while I have a lot more shots off target, I also have a lot more on target too.

Maybe I could turn off the hit early crosses and see how that works?

You've highlighted a possible issue with conversion rates, so again I would say that it's probably going to be best to analyse the chances and look for issues and in particular, patterns. It could be the type of chances or that it falls to a certain player or against certain teams or certain times of the match. The stats help paint a broad picture, but that's it.

If you're still stat focused, an idea is to maybe start with your front players and see if any of them have poor conversion rates? The top players in the league usually average a goal for every 4th or 5th shot, so that's maybe a starting point in terms of identifying either a player or position that's causing you issues.

[Joey Numbaz]

47 minutes ago, HUNT3R said:

The top players in the league usually average a goal for every 4th or 5th shot, so that's maybe a starting point in terms of identifying either a player or position that's causing you issues.

Oh wow - yes, we are low there. This is combining both tactics, I can't split them out, but (stats don't included penalties):

Player     Goals     Shots   %
Spahic       16         169     9.5%
Haaland    16         182     8.8%     (he played Wa in the 442)
Lautaro     18         126    14.3%
Ramos       14         113    12.4%   (he played striker in the 442)

Part of the reason I changed the tactics was that I wanted to find a way to get all four of these guys on the field at the same time. In the 442 I was limited to 3 at a time.

Ah wait! I can find a save from around the time I switched tactics and subtract those out.

Before the switch 442 (3 games with other formations) prorated to 90 minutes
Player       Goals     Shots    %
Spahic        0.48       4.46    10.9%
Haaland     0.40       5.04      8.0%   (as a Wa from the left)
Lautaro      0.42       2.69    15.5%
Ramos       0.50       3.59     13.9%
Total          1.80      15.64    11.6%

After the switch (343) protated to 90 minutes
Player      Goals     Shots    %
Spahic       0.35       4.51      7.8%   (usually CFs on the R and he's right footed)
Haaland    0.46       4.68      9.8%   (usually CFs on the L and he's left footed
Lautaro     0.46       3.46    13.2%  (usually AFa in the middle and he's right footed)
Ramos       0.19       2.10      8.8%  (he's an AMC mostly now only occassionally a CF on the right when rotating)
Total          1.48      14.82   10.0%

 

 

Edited February 1, 2021 by Joey Numbaz

[HUNT3R]

Was just mentioning it as a possibility and as I said, that's mentioning the top players over a league season. 15% isn't bad, for instance; just that the top guys in the league are usually above or around 20%. Not sure how closely FM20 followed that rate. My own team in my final season had fairly low rates there, but a) I didn't use Work Ball Into Box or Shoot Less Often (only 1 of the front 4 had this), so long shots weren't specifically discouraged and b) goals are usually spread around my attacking players, not that that would change conversion rates.

My 2 best players in terms of goals scored were around 17.8-18%. Perhaps it's lower in FM20 than it should be - I don't know. 3rd best was 17% and 4th best only 14%.

Might be worth looking into Spahic and Haaland (in the 343) and their chances at least as those do seem low.

[Joey Numbaz]

1 hour ago, HUNT3R said:

You've highlighted a possible issue with conversion rates, so again I would say that it's probably going to be best to analyse the chances and look for issues and in particular, patterns. It could be the type of chances or that it falls to a certain player or against certain teams or certain times of the match. The stats help paint a broad picture, but that's it.

Definitely understand what you are saying here. I think it's probably the headed attempts that's the biggest issue. Any ideas for how to reduce that? Obviously cutting down crosses would help. We had the most headers won in the league 1213, and the highest percentage won 67%. I try to get tall players everywhere.

We had by far the most crosses in the league. 487 completed (at 32%). Second place was 386 at 24%, and 3rd was 311 at 22%. So we are great at getting on the other end of crosses.

Breaking it down

before the 343 formation - 274 crosses completed, which prorates to 434 per 38 games at 31%

using the 343 formation - 213 crosses completed, which prorates to 578(!) per 38 games at 33%

That is a ton of crosses wow. The 442 did not have hit early crosses, though the winger on the left and the fullback on the right have cross more often by default. In the 343 both defensive wingers have cross more often by default. And since the complete forwards are staying wider they cross a lot too.

before 343 cross attempts per 90:

Spahic        1.3
Haaland     4.8  (as a Wa from the left)
Lautaro      1.3
Ramos       1.4
Gaudin      5.5  (as a FBs left)
Princigalli  14.4 (as a FBa right mostly)

343 cross attempts per 90:

Spahic        5.3(!) (CFs staying wider)
Haaland     3.5    (CFs staying wider)
Lautaro      0.8    (AFa)
Ramos       0.6    (AMs)
Gaudin      9.3  (as a DWs left)
Princigalli  15.4 (as a DWs right)

Edited February 1, 2021 by Joey Numbaz

[Joey Numbaz]

2 hours ago, HUNT3R said:

Maybe have a look at hidden attributes like Consistency and Big Matches, both of which would show up in a scout report if it's either good or bad. If it's 'in the middle' it wouldn't be mentioned.

These are two things I actually look for! I pretty much never have a player who is bad in big matches, and I am pretty sure everyone is at worst not inconsistent. I have decided not to buy a 5-star superstar ratings wise, because he's fairly inconsistent.

Spahic - dark green consistent, nothing for big matches. But he's still only 21, and I've never seen that on a scouting report before a player is 22

Haaland - dark green consistent, relishes big matches

Lautaro - dark green consistent, enjoys big matches

Ramos - dark green very consistent, nothing for big matches

Gauidn - dark green consistent, nothing for big matches

Princigalli - light green fairly consistent, thrives on playing in big matches. I don't know that I've ever noticed that as a thing before, sounds even better than relishes!

[Joey Numbaz]

32 minutes ago, HUNT3R said:

Was just mentioning it as a possibility and as I said, that's mentioning the top players over a league season. 15% isn't bad, for instance; just that the top guys in the league are usually above or around 20%. Not sure how closely FM20 followed that rate. My own team in my final season had fairly low rates there, but a) I didn't use Work Ball Into Box or Shoot Less Often (only 1 of the front 4 had this), so long shots weren't specifically discouraged and b) goals are usually spread around my attacking players, not that that would change conversion rates.

My 2 best players in terms of goals scored were around 17.8-18%. Perhaps it's lower in FM20 than it should be - I don't know. 3rd best was 17% and 4th best only 14%.

Might be worth looking into Spahic and Haaland (in the 343) and their chances at least as those do seem low.

Thanks for all of the insight, it's definitely helping point me in the right direction.

Filtering on attackers and attacking mids for the leaderboard I get this - it stinks that they don't remove penalties for this. Or even show you on the page how many penalties a guy had.

%       Goals Shots
21%     18       82
21%     13       61
20%     16       79   inflated by 2-for-3 on penalties actually 18.4%
17%     15       88
16%     25     151
16%     15       91   inflated by 3 penalties actually 13.6%
16%     12       73   this is Lautaro inflated by 1 penalty actually 15.3%
16%       9       56
15%       7       44   this is the corpse of Harry Kane at age 34, inflated by 3 penalties actually 9.8%
15%     10       66
15%     11       73   inflated by 4-for-6 on penalties actually 10.4%

It looks like anything 15% or better is really good in league play. Only 8 attackers/attacking mids in the league hit 15% if you discount penalties.

Edited February 1, 2021 by Joey Numbaz

[lfds89]

About headers: I've been playing a tactic that heavily relies on headers. I have to say some attributes are very important to get a decent win rate (which then leads to a decent goal rate). Bravery, Jumping, Strength are probably the most important. I've noticed some smaller players winning by anticipation and off the ball. Then, definitely heading to direct it. So, I would say if your tactic depends more on crosses than others, then you should look into individual's more than tactics alone.

Regarding CCCs : I've no data gathered, but from observation I feel crosses generate more CCCs than possession plays. I played possession with Tottenham and I had plenty of shots inside the box without near opposition that never counted, not even as HC. With crosses I get 1/2 CCCs every game, minimum. In both my saves with Porto and Carlisle my players with highest chances created / 90 are the ones that cross the most and set-piece takers. This means : your CCCs might be different to other team's CCCs.

[Joey Numbaz]

27 minutes ago, lfds89 said:

Regarding CCCs : I've no data gathered, but from observation I feel crosses generate more CCCs than possession plays. I played possession with Tottenham and I had plenty of shots inside the box without near opposition that never counted, not even as HC. With crosses I get 1/2 CCCs every game, minimum. In both my saves with Porto and Carlisle my players with highest chances created / 90 are the ones that cross the most and set-piece takers. This means : your CCCs might be different to other team's CCCs.

Very interesting about headers being a disproportionate share of key chances. I'll try to keep an eye on this.

Player     BRA   STR   JMP  HEA  ANT  HDR%  HDRwon90  Key
Spahic      10      15     16    14     16      63%       4.67            89
Haaland   15      17     16    14     17      71%       5.90          117
Lautaro    18      14     11    14     17      41%       2.23            54
Ramos      13      14     15    15     15      59%       3.25            62

Spahic and Haaland get most of the headers. i think they are all pretty good here. Martinez is short, but he's decent otherwise. He only wins 41% of headers. Of course winning a header and making a headed shot are a little different. Haaland was 4th in the league amongst attackers/AM in headers won and second in key headers with 87 in league. For whatever that is worth.

Edited February 1, 2021 by Joey Numbaz

[lfds89]

This is from End of Season Porto Statistics. Surely the correlation isn't that high, but it suggests a trend. Crosses Attempted / 90 x Chances Created / 90 for each player (even players who barely cross)

[Experienced Defender]

On 31/01/2021 at 22:54, Joey Numbaz said:

CFs  AFa  CFs

                 AMs

DWs                       DWs

                 HBs

    BPDd    Ls       BPDd

 

10 hours ago, crusadertsar said:

The tactic looks completely unbalanced actually and way too top heavy

Honestly, I don't think that this tactic - or rather, setup of roles and duties - is unbalanced. The formation is top-heavy and obviously very adventurous per se, but the setup itself has good overall balance IMHO.

In fact, what it lacks is not tactical balance but penetration. In that regard, I would just switch the AMC's duty to attack and that's it.

[crusadertsar]

39 minutes ago, Experienced Defender said:

 

Honestly, I don't think that this tactic - or rather, setup of roles and duties - is unbalanced. The formation is top-heavy and obviously very adventurous per se, but the setup itself has good overall balance IMHO.

In fact, what it lacks is not tactical balance but penetration. In that regard, I would just switch the AMC's duty to attack and that's it.

But if you set AMC on Attack then you only really have one midfielder in central midfield. I don't understand how that's balanced. But whatever I don't really care to be honest. If it works that's what matters.

Edited February 1, 2021 by crusadertsar

[Rashidi]

17 hours ago, Joey Numbaz said:

Sure. Full credit, this is an @Rashidicreated tactic,

Thanks I completely forgot that i made a 343Diamond, the halfback is one of my favourite roles for these 'centrally empty' formations. I tend to make things easy for myself when i evaluate chances as well. To be honest it can be overwhelming trying to nail down good xg chances that dont come from crosses etc. The effort you are putting into this is pretty astounding

[Experienced Defender]

14 hours ago, crusadertsar said:

But if you set AMC on Attack then you only really have one midfielder in central midfield. I don't understand how that's balanced

Yes, but you also have a 3-men back line, one of which is the libero, who virtually acts as a quasi-midfielder in possession 

[Joey Numbaz]

 

5 hours ago, Rashidi said:

The effort you are putting into this is pretty astounding

I can go a little overboard when inspired :-) That and I had a 4-day weekend from work, needed a project!

Any thoughts on why I am missing so many chances? Did you notice the footedness of the strikers being important? Any ideas on how I can reduce the headers? I'm thinking turn off early crosses and maybe even turn on work ball into the box. But open to any ideas before I start to expirement.

1 hour ago, Experienced Defender said:

15 hours ago, crusadertsar said:

But if you set AMC on Attack then you only really have one midfielder in central midfield. I don't understand how that's balanced

Yes, but you also have a 3-men back line, one of which is the libero, who virtually acts as a quasi-midfielder in possession 

Yeah you still have a 3-1 or 2-2 defense when attacking, depending on how involved the libero chooses to be. Often he stays back with the other defenders, but he does provide support for the half back when he's being pressured as well. And with good work rates the DWers do a pretty good just getting back too. You can turn them to defend duty if the other team is exposing you also. It is a very fun tactic to watch, outside of all the missed headers :-)

Edited February 2, 2021 by Joey Numbaz