akkm
-
Posts
1,386 -
Joined
-
Last visited
-
Days Won
1
Content Type
Forums
Articles
FAQs
Online Manual
Support: Blocks
Support: Games
Bug Tracker
SIGames Manual (beta)
Profiles
Posts posted by akkm
-
-
9 hours ago, Svenc said:
From my end, you needed to get Messi to have the amount of finishes (mostly inside the box) he has in real Football. Which is 5 per game (for CR7 it's actually 7). That usually doesn't neeed an exploit, however it needs him to be the focal point of attacks. E.g. Tuning Things so that more of the finishes fall to his, which doesn't happen by Default. Hence the AI struggles here also, as it can't make a Player the "focal Point of attacks". This is not within the AI's "match Management Arsenal".
An admittedly old example of this
I don't believe in the 1 Goal from 100 1vs1s, however what I believe in is that the lack of scoring from 1vs1 currently may be made up by the conversion from range (as also shown here), which (as in FM19) seems multiple times higher than in actual Football. Somebody actually did a Research here. SI never intentionally beefed up anything so that scorelines would remain realistic. Hence, the reported 3-6 hockey scorelines every week on FM15, depending on Formation, roles and duties, gifting unrealistic space all over the pitch back to front until patched.
Clearly one of the root causes of there being this many one on ones is exemplified by the CB behavior in the above. Both by a) shifting inside to too closely follow his Partner he makes the distance to Morata too big in the first place. And b) his movement is all wrong, gifting Morata tons of acceleration space by heading headfirst straight into Morata's direction, as opposed to anticipating his (logical) movement goalwards. That said, the above Kind is pretty much usually a 1 in 4 Chance give or take (1 Goal from 4 of such Long-term average). Mind, the volumes may come down anyway once the above is and should be fixed.
wait so you consider that morata chance a one on one...I certainly wouldn't...a one on one in my view is a player clear in on goal with time and space to finish it
0 -
33 minutes ago, Svenc said:
I never claimed sitting Deep, keeping ten men behind the ball and hoofing the ball upfield, usually gifting the Opposition attack after attack, was a good idea. Which is where the many Goals conceded come from making for the minus Goal difference. All I claimed that on FM15 you could do such, and if that guy upfield was Messi, and due the space available in that ME, he could still bang in enough Goals being by himself alone to somewhat Keep you afloat in the table.
(I personally doubt that'd be viable in actual Football -- or in any current release. Worth a try though…)
Yeah but how many times without match engine exploits did FM enable messi scoring 50 goals a season...so messi was doing that in actual football but no proper football style in FM was allowing messi to play like messi...only exploits allowed messi to score like in actual football
0 -
39 minutes ago, Rashidi said:
I agree that video was just an example of how terrible defending was on FM17 it made attacking look so good.
There's absolutely awful defending there but that is more an example of how bad defending should be punished...ie if defending is that bad then decision making employed and simulated in FM17 means that it gets exposed as it should. Movement, decision making and pass execution destroys bad defending as it should be so that shows attacking decision making was superior then. Yes defending was inferior then but the latest engine has inferior attacking decision making and movement...so really defending has advanced and attacking declined so current engine you could flip the dynamic...terrible attacking makes defending look good...yes defending is improved but bad decision making/movement makes it seem better.
What should be upheld in that video and what FM17 offers is the superior attacking decision making and movement and were current engine at that level then the repeated attacking patterns currently wouldn't be a thing and the overall balance would be better.
Pointing out bad defending on FM17 is valid but the slant on attacking in it is a limited view and frankly a regressive one to development of the match engine
6 -
Interesting discussion above
one thing on the issue of perception...I'd extend this to bias and it's a pertinent issue of current times with the entrenchment of people's bias/perceptions all over the world being engineered and exploited by populists/propaganda/fake news giving rise to the likes of brexit/election of trump and populists all over and social media enabling the propagation of this through filter bubbles/echo chambers. What we see in situations of economic downturns and social disaffection peoples perceptions and biases are more easily 'manipulated'/played upon with them becoming crystallise in minds to the detriment of reality, facts and evidence. In essence this shows that humans have a tendency to inuldge bias/perception to make themselves feel good about themselves and have their opinions 'confirmed' even though they can sometimes be demonstrably false...or basically people when they want to believe something will believe it no matter what.
This leads to football...one of the reasons for it's popularity is it lends itself to varying opinions allowing people to hold theirs up against another and ultimately as it's not scientific can walk away and still feel good about themselves and say 'it's a game of opinions' therefore who's to say who is right or wrong. Extending this further to football when people say there's no right or wrong way to play the game...this is broadly a fair comment...but when you start to get deep in the weeds there are BETTER ways to play than others...once one starts to argue against that then one is going down the road of indulging bias.
This then starts to feed into point below
3 hours ago, optimusprimal82 said:Perception is a powerful influence when analysing this stuff remember, and for every person unhappy there's another who seems to be generally ok with how things are; doesn't mean that either side is 'right' - just as the game itself is all about opinions, so too will a simulation of said game be, but one thing that we can surely all agree on is that improvements can be made.
I agree with all that...but it's also absolutely the case that one side can actually be right in certain instances and it's not a 'the game itself and therefore the simulation is all about opinions' yes it is of course about opinions but not all those opinions are correct. I'm not suggesting that you're saying that it's not an issue but what I will say is that approach or an 'agree to disagree' can actually compromise the significance of issue itself in terms of 'well people can have different opinions therefore it's not really an issue'...again I'm not saying you're saying it's not an issue but perhaps arguing more the scale/extent of the issue but the bottom line is it is a major issue in terms of how quality players and teams operate from an attacking point of view and how they go about breaking down teams in the real world being undersimulated in FM.
Passing decision making/execution and player movement are the biggest deficiencies currently in FM in attacking third to enable quality attacking incision/patterns which we see in real world footballing meaning the way/the how teams create/score is displaced towards the wings/wider play disproportionately over the years.
few points of note
Firstly is that those who defend number assists from crosses cite FMs broader definition of what constitutes a cross to defend it's overuse of crosses as a means to create openings/goals yet at the same time cite real world through ball assist stats to uphold a notion that real world assists from through balls are what they are therefore FM assists from through balls are ok and therefore there's not really an issue even real world stats taking whoscored definition of a through ball is a rather narrow one and likely much narrower than FM (have seen that on other threads)...so that's inconsistent in itself
secondly the reference to through balls (which is an issue for sure) in any discussion of undersimulated central play should be broadened to threaded passes into feet/tighter spaces/pen area and other subtle/guile/incisive passes one twos etc as well as through balls as a means to open up defences of all sorts...deep lying/normal/high lined ones, set defensive units/transitioning units etc...it shouldn't be limited to how one defines a through ball as there are other creative passes which occur in all phases of play which need bumping up and as also mentioned above...a through ball may lead to an initial opening to set up a goal further on in a move not necessarily just an assist.
@MBarbaric makes a key point here
2 hours ago, MBarbaric said:i think nobody argues there are no through balls on counter attacks. the real problem arises in positional attacks where team faces a prepared defensive unit.
Again we have many examples of FM producing through balls on counter attacks but through balls/threaded passes/subtle passes when playing against prepared defensive unit/deeper lying defences are an undeniable issue.
Just to go back to perceptions about real world football. One may regularly see/hear opinions that playing against deep lying/prepared defensive units you will see more goals from crosses/long shots/set pieces and more instances of long shots/crosses because of the difficulty of breaking down these defences. Also one should be cognisant of deep lying/set defences are difficult to break down in themselves and certainly are generally harder to break down than normal/higher defensive set ups/lines but the perception of the difficulty of breaking these defences and the how is also misplaced.
Take man city for example
Man City
goals
assists
goals
shots
set pieces
crosses
open play
inside box
outside box
2017/18
20%
17%
80%
88%
38%
2018/19
10%
18%
90%
87%
32%
Prem
goals
assists
goals
shots
set pieces
crosses
open play
inside box
outside box
2017/18
27%
19%
73%
87%
40%
2018/19
29%
18%
71%
87%
38%
What this shows is that despite playing against deep lying defences more than (arguably in prem anyway) all other teams Man City score less than average number of goals from set pieces and more from open play, assists from crosses are broadly in line with other teams and city actually take shots from outside the box LESS than the average...so all this is the opposite which I've often seen held up as a way to breakdown deep lying defences...the opposite is in effect in the real world using City as example and this is what is missing from FMs engine. In FM breaking down deep lying defences is overly reliant on patterns at odds with the real world...so basically what this shows is that more crossing/more long shots/more goals from set pieces to breakdown defensive sides just isn't a thing in real world football so it shouldn't be a thing in FM.
Taking this back to FM and the issue of central play and extending the conversation beyond through balls (though they are a big deal) to other types of incisive passes then one can see they are definitely undersimulated. Of course no one should expect any team to consistently open up deep lying defences with ease game in game out as that's not a real world thing either but the frequency and tendencies that quality teams try things is not being replicated in FM.
Other things such as frequency(even within a match)/tendency of AI teams to set up lower defensive blocks is definitely a contributing factor but it essentially masks underlying pass selection/decision making not identifying openings and executing passes when opportunities arise...so key elements of how teams pass and move to create openings are missing in FM.
Another perception issue is use of width/wings/crossing in FM vs real world to break down deeper lying defences a la city do
FM classification of how they see a cross differs from many real world stats collectors and whilst that can nebulize things it's fair enough to explain certain differences of numbers of crosses/assists from crosses...the issue is how play is simulated to get the situation of where the likes of city 'crosses/cut backs' are played from and the play which unfolds to get the likes of sterling/sane into said situation
a lot of play in FM has feeling of pass pass pass and then out wide leading to cross which causes more trouble than everything that preceded it...basically again crosses seem disproportionately dangerous Vs players creativity centrally...i'm talking about crosses from out wide as opposed to low crosses from inside channels in and around edges of pen area you see from city so again this is a misrepresentation of real world football...crosses from out wide are actually very easily defended by deep lying defences...the low crosses/cutbacks from inside channels less so...but there is a subtle yet enormous difference between the two
I generally make above distinction when talking about crosses as its clear city make a lot of use of these inside channels cut backs/square balls/low crosses or whatever terminology one may use. The thing is as I say there's actually a huge difference between them and crosses from wider positions...obviously positionally for starters but then the ones city employ tend to be lower and closer to pen area/edge or in pen area so automatically the proximity to recipient and 'on the ground' pass/cut back/cross is increasing the probability of a goal being scored...its easier for attacker to finish a chance like this than a cross from wider through the air and also very simply and perhaps most pertinently the player playing the cut back/cross is in a position to more accurately pick out the player in the pen area being closer and the play which has preceded it has been played slickly and fluidly to get player like sterling/sane into a highly advantageous position for said situation...all of it is with the intent to increase the probability to score/create a higher quality chance.
the how is also a very important distinction to be made...watching city you will see that how they get the ball into these positions is through incisive play working the ball fluidly through spaces and often the preceding pass into the path of the runner to play cut back/low cross is vertical/diagonal one taking out a FB or CB but there is the incision/penetration pass wise...the preceding pass often isn't from out to in which is an important thing to simulate properly going forward
quick review of man city goals Vs Chelsea and arsenal
vs chelsea
- sterling 1st goal...de bruyne inside to bernardo...bernardo drives deep into pen area for cut back/low cross...almost on edge of 6 yard box...so played from 7 yards out and from edge of 6 yard box so very close to goal
- aguero sitter was cut back/low cross from zinchenko near goal line and maybe 3 yards away from 6yb to bernardo who moved into nearly same position to play ball across the goal from
- aguero 1st goal..zinchenko cuts INSIDE albeit wonder strike
- aguero 2nd goal...zinchenko cuts INSIDE flipped air pass which caused havoc...error strewn from there...were that in FM people would be all over it with bug reports lol
- gundogan...aguero laid it outside alright to sterling in a classic wide position but from there sterling played a threaded lateral pass INSIDE to aguero feet and miscontrolled cleared bang goal
- aguero pen...fernandinho out to sterling is very FMesque pass to out wide as yes it happens for sure (just too often in FM) but from there sterling drives at FB (on azpi outside) but he's driving diagonally towards goal
- sterling 2nd goal...this is classic city play...silva threaded pass (which I personally consider a through ball though that's not how stats providers would class it) into zinchenko between two players INSIDE outer defender for zinc to play low cross from edge of area.
vs arsenal
- laporte gets breaking ball...hits cross from just inside pen area...was chest height
- fernandinho FM type ball out wide to sterling...but from there sterling passes inside to gundogan who dinks it over the top to sterling making run inside and connection he makes is on edge of 6 yard box square ball to aguero
- sterling gets ball on edge of pen area again but drives into box but plays ball across box from 7 yards from byline and between edge of pen area and 6 yard box
Basically all the cut backs/crosses were low and mostly well inside pen area or edge of pen area like laporte assist and generally passes come from centralish positions to inside channels in or around edge of pen area and if passes to outside then sane/sterling will drive inside or try and get into inside channels to receive passes for cut backs/low crosses/laterals to others inside pen area.
so that's the breakdown of city's goals so really while it may perceived that city are using width and wings it's really not the way many people think it's being used...it's almost an illusion...the runs are coming from the wider areas but the ball generally isn't or basically sane and sterling maintain their width (nominally to stretch opposition) but then make runs inside to receive pass from a silva/bernardo/de bruyne/gundogan who are in a central position playing said pass to an optimal position in an inside channel as close to or inside the pen area as possible and as deep (close to goal line) as possible.
in FM I don't see the same situations city create in terms of build up and what precedes the moment where player receives the ball and certainly not with frequency with which these cut backs are played...ie the tendency in FM is for crosses from out wider and more through the air...so really again this Vs real world is at odds a bit and tweaking this (for next year) will be a game changer
with regard to this
6 hours ago, optimusprimal82 said:I think we're all analysing/bashing/hating on the ME (and it's not unfair to say it definitely needs some improvement in the attacking sense - odd decision to improve defensive play and leave attacking for later) when (and not exactly a sherlock-esque deduction) it's more likely a combination of ME+reputation+tactics - there's a reason a fair few of us now fear the bottom 5 sides in a division more than the top teams!
Clearly all together compound issues but this in itself suggests the undersimulation of repeated quality attacking patterns and more incisive/pentrative play is a real and far from perceived issue...and indeed a big issue. If it weren't then despite playing increased reputation and tactics of deeper lying defences on their own would be ok really...so increased rep causes teams to employ more defensive/deep lying defences against said increased/high rep...nothing untoward there...that happens in real world football so fair enough in FM.
However having it trickier playing against bottom 5 sides than top 5 ones...then that of course is indicative of an issue plain and simple. What the issue is...even more plainly and more simply...is the simulation of creativity/passing incision/movement/how the ball gets used and moved around centrally and in and around inside channels in open play to breakdown these defences is at odds with the real world...after all teams playing against lower based teams tend to average more goals per game than against top based teams...so in essence it IS the attacking play not up to scratch is the crux of it. Its undeniable. Were it more in line with real world play the high rep teams in FM should be able to go about things as high rep teams in real world teams do...as you suggest the opposite can be in effect in FM sometimes...so that's actually a very big deal really. After all it is the Man Citys/Barcas/PSGs who play most deep lying defences yet they're the ones scoring the most goals over the course of a season in the real world...should that not be happening and most importantly reflecting the 'how' that's happening is an obvious shortfall in FM. Essentially FM isn't simulating the tools to unlock these types of defences to the level or attempted frequency in real world. Of course there will be struggles and tight matches of odd goals victories/losses and draws but the difficulty of breaking them down and the 'how' isn't currently simulated as it should be.
So all in all central play IS a big issue and perceptions underestimating this are misplaced. SI is aware of issues here and have tried to improve certain aspects to enhance enjoyment for this year. Whether all that lands this year remains to be seen but identifying the extent of this is important so it can be addressed as required to produce a more realistic engine which produces attacking creativity/incision/patterns/movement as we see in the real world.
7 -
1 hour ago, KlaaZ said:
Regarding the long shots everyone seems to be suffering from, could you also provide this screenshot to compare? Because I really fail to see the problem.
Out of my last 50 games, I scored 100 goals from inside the box, 29 from outside and 3 really long range screamers. So that means more than three out of four goals are being scored from inside the sixteen yard box. That seems like a very reasonable statistic.
do you have assist locations there as well
0 -
3 hours ago, Svenc said:
And the "FM 17 fix" for that, e.g. wide midfielders sitting out wide on defending, caused third division sides to be all over first division ones when they had a man Advantage in the centre of the park (and knap's, TFFs et all multitudes Keep it central, stupid, exploits of that to turn into Comedy Gold whenever an AI happend to sit Deep and plug the middle of the pitch) .
Not sayingn that any of this may repeat, mind.
For my tastes, there's too much talk About perceived weaknesses of the attacking Phase and movement in General. That is a natural bias naturally, as similar to football Viewers, FMers tend to prefer the spectacle. Whilst both attackind/defending phases need to be balanced off one another -- the key to developing the most robust ME ever imo is FM simulating zonal defending "proper" first, that is modeling the defending as "realistic" as possible first. This means including the more physical aspects of defending better with an actual collision between Players/bodies. But also teams defending as Units ever more intelligently. Likely easier posted than put into action, no less as the game still also caters to really really ancient PCs, which means there's likely a finite headroom as to actual calculations. Which also likely won't last forever.
This is the wrong way to look at it svenc...until attacking patterns and movements are simulated better then that's what will impede the development of the engine.
Until you do that you will be stuck with circular balancing exercises which we see every year. You talk of bias towards a tendency to prefer the spectacle but there has been a pervasive bias towards the defensive side on FM to the detriment of the attacking side of it. I've actually just made a couple of posts about this here but some users of FM tend to want to get the feeling that it's there tactical choices that make the difference to the outcome of matches and are unaware of how the undersimulation of the attacking side of the game has 'reverse exploited' things to their benefit
also the classic nerfing of central attacking play
Central play has actually been undersimulated every year bar FM17 since the rewrite to varying degrees it just hasn't been as noticeable it to the same extent as this year. it's come in various guises
- Too many shots – as a result of underdeveloped ball circulation passing options/movement lack of through ball tendencies when there means when central players shoot meaning shot counts see serious spikes
- Too many crosses – as a result of propensity to use players out wide several iterations have seen numbers of crosses per game significantly exceeding real world. Mitigation of this may have taken form of improved defending positionally/restricting forays of full backs/wingers but the long term solution is to programme decision making to see recycling the ball of circulating inside as the better option/decision
- Poor finishing/lack of variation in finishing...generally in play through the middle. over the years some of the finishing from crosses/diagonal passes from out to win has been remarkably good
- Poor ‘one on ones’...again players slotted in through the middle have low finishing rates in FM (I know svenc you question that but big chance conversion rates suggest higher rates than even you like to admit
)
- Play blocked through the middle forcing passes too frequently out to wider players
- offside frequency high most often preventing players slotted in through the middle
- If you consider the options for wide players cross from x2, cross frequency x2, aim crosses at x4 it offers an ability to have a more concentrated and specified means to attack and as a result may imply an increased frequency by which this will be utilised
- Crossing overpowered effectiveness wise...FM18 check out tactics uploaded on forums...many favoured wing backs some getting double digit assists which isn't a thing in the real world.
So yet another bias in FM against quality attacking patterns.
Overall though unless you enhance attacking patterns and the complementary attacking movement to offer better passing decisions/options to enable more fluid higher quality attacking play then the defensive side cannot be simulated to its maximum. If you do defending first then you won't actually know if the coded core decision making from an attacking point of view is on point as it will be masked by defensive elements...there's no better example of this than this year's well documented (and very well evidenced) threads on central attacking play and lack of central through balls.
The million dollar question has been why has it been deficient this year particularly as to why there was an obvious lack of central through balls. Was it lack of movement by strikers, was it better simulated defending in terms of narrowness, the AI defending so deep so often, is it the lack of spaces available to AMC playmakers to play them into...these are all factors which affect it but the thread below shows definitively and beyond contestation that the core pass decision making to identify and execute these passes isn't being simulated properly to begin with. So this is a glaring omission from the code to enable quality attacking play and therefore a significant element of a users/managers arsenal to try and help unlock all sorts of defences and all sorts of opposition lineups is missing...so basic default decision making is not in FM to try and attack and creativity is woefully short as a result. So this proves that the much talked about weakness in the attacking phase is far from a perceived one but a very very real one...you might say Svenc, any denial of this having seen attached is indeed one indulging one's bias
. That was just through balls but other quality and deft passes in tight areas/threaded passes into feet/passes into pen area putting player in for a shot given angle of run/one twos etc etc...these are also severely undersimulated and needed to break down tight defences
In terms of development of the match engine as you can see trying to identify the problem in the first place can be masked/hidden by strong defensive elements. Therefore developing the defensive side of things first will only mask core issues further meaning further fixes/workarounds to change things are like the 'proverbial band aid fix' with these causing knock ons causing knocks ons ad infinitum.
The way to develop the engine to enable it to achieve its potential most efficiently and most expeditiously is by maxing out the attacking side first in terms of pass selection/decision making and ensuring it's all working well. In essence by doing that the attack should destroy a defence for all the right footballing reasons. Then at that point defending should be addressed whereby it can be coded to deal with and cope with attack maxed out. Defending is mostly reactive anyway. Then you can model positional play/zonal and man and how that movement will prevent passing lanes/block spaces/deal with player runs/behaves as a block to move as a unit around the pitch within a team framework keeping its structure as required etc etc
5 -
On 04/12/2018 at 04:23, akkm said:
Where are you getting your one on one scoring probability from
attached is a counterpoint to above (big chances). Last season of premiership with big chances detailed for certain players...for 20 players attached...the average conversion rate of big chances is 51%...ranging from 70% to 37%...even the much maligned morata converted 37% of his big chances
The average of the top 5 is 63pc
I know you will say big chances can include tap ins but the vardy is top at 70%...impression wise at least he gets a fair amount of chances from balls over the top and running into space for one on ones rather than excess amounts of tap ins...other players on that list are not all about tap ins either and anyway average for 20 players over the course of the season is actually over 51% conversion rate of big chances is significantly higher than the base of 20% which I've seen in one calc for Xg layered on for a big chance.
In essence...big chances conversion rates can and do vary individual to individual...that's just simple math as well
17 hours ago, Svenc said:Just seen this, sorry late, busy month. Don't know how exactly Opta data constitue "Big Chances" these days, on their website it's become a little wooly. However, statistically, we are entering the topics of "small sample sizes" and "Regression to the mean" here. Not only do all of these numbers on the Big Chance list trigger that potentially issue, as Players on that list average like 20 of such chances over the entire Season (which suggests the Definition is fairly strict and only the higher end of chances apply). If Vardy holds such an edge over the average, he won't keep it for Long. If he would he would be the most desired forward in world football. The smaller the sample size, the more significant the amount of chance. There are also a number of penalty takers in the modern game who have like a 100% record -- however, they all tend to have about 20 penalties or less. Eventually, they will not convert, or they'd be likely aliens. That's also the reason why on every seasonal "top free kick taker in Europe" graphics published there tends to be a player or two who never appear again. Free kicks are additionally special, in that it only takes 2 or 3 Goals to score off them, and you may make it onto such a list.
Likewise to how Salah this term only just about "fulfills" his xG (his about +7 overperformance was the highest he had in recent years). It's still "just" a suggested" added Goal roughly every 6th league match over what you would expect, that is based on the finishes he had. Some of his chances however may have been weaker ones, or not at all happened with a different Forward -- xG doesn't account for positioning skills, after all. It purely tries to gauge a difference in finishing ability. On FM likewise, the chance is far more important than the player. My numbers ultimately don't matter at all. It's SI's numbers that matter. They may hugely contradict much what is broadcast on TV all the time, including many punditry cliches towards forwards/finishers and generally chance conversion. All the more reason to get some added valuable feedback in there...
My point wasn't about vardy tho. There was the quote that big chances start as probability of being scored at 20pc but average of top twenty players over course of a season is significantly higher than that. I cited vardy more as an example of one on ones which see repeatedly quoted as 1 in 3 chances are scored from but cannot see that anywhere on the net lol. Absoutely he will more than likely revert to the mean but his average conversion of 3 seasons before this is 53% so again that is significantly better than 20pc or 1 in 3 one on ones...interestingly last two seasons is 66pc...so is he improving. Also his ability beyond conversion rate of big chances would be factored into being the most desired forward in world football...after all whilst big teams create big chances a forwards ability to operate in tight spaces to link, move and finish would be more important factors to consider...no point signing a vardy no matter how high his conversion of big chances is if his other attributes don't contribute against deeper lying defences which they would be required to...big teams are not assessing a forward exclusively for their big chance conversion rate so wouldn't make him the most desired forward in the world especially as big teams aren't hanging around waiting for an opportunity to play a player with brilliant big chance conversion a la vardy over the top just to utilise that conversion rate...going down that road would mean they wouldn't be very successful at the top level and no longer a big team
the average for the 20 players over the course of the season was 51% which is significantly higher than but I read on squawka average big chance conversion rate for premier league is 43.4pc so again that's a far cry from the base 20pc. Also the 43pc is an average so you would actually have players achieving rates north of that and some may consistently do so.
Saw another thing where teams top ten wise with most big chances missed (in absolute numbers) was quoted for last 5 years (cant find article since but I had copied the table over, may have been 3 years but think 5)...the average conversion rate for them was 44%...so that's team wide for 5 years so takes the anomalous elements out of it and is sufficient sample size to start framing expectations from...at least ballpark wise
You can't start to compare deviations with free kicks to pens either as you say...the percentage conversion of free kicks lends itself to random spikes either way but conversion of big chances is far higher than free kicks so it's a redundant comparison to make.
Bottom line is the big chances ARE scored at higher rates than so they stand alone so when one expects these chances to be converted at higher rates than normal chances in FM there is real world statistical justification for this premise. Some may well have expectations higher than real world but suggesting a 20pc base for big chances it's misleading and underestimating how these chances are scored and again as an average with poor chance conversion dragging it down instances/runs of these chances being converted at higher rates will also occur...I havent seen many on these forums talk of their strikers converting 60/70pc of their one on ones even in the short term !!!!
0 -
1 hour ago, Svenc said:
Posts like this still make puppies weep. The frustrating thing is that there may be a great point in there. But possession and shot counts won't ever, ever show it (no less as for as long as an AI Manager Drops Deep during a match, their opponents are gonna dominate those simpleton stats by default, rather than inherent awesomness). Sorry for the repeat. But you still cannot stress this quite oftenly enough.
Let's still tackle this from the numbers/stats perspective. We know that in Football Analytics, that is the real one, "big chances" start at about a goal probabilty of roughly 20%. That's pretty good. The average shot conversion is about 10%. Ranged attempts typically oft are in the <=5% range. A one on one is about a 33% Chance. Why this "low? The key is the term "one on one". Unlike many other chances, the keeper is fully game -- and unlike the Forward trying to hit a specific target, is also allowed his Hands, to deflect it somewhere off that target - cheater). A Penalty ranks at about a 75% of scoring. Generally, outside of tap-ins and penalties, no Forward in competitive football is considered to have the Edge to score (hint, hint, prime reason why football is such a low scoring sports right there). Also, the Players on their respective Level don't make a league difference. Clubs spendn gazillions in an attempt to gain any small edge they can. And oft fail. F'r instance, according to various analytics, Salah last term scored about 7 goals more over the Prem Season than an average Prem forward would be expected to score. That would be an added goal roughly every sixth match average which Bobby Premierleaguestriker may not score. Some of those Goals may have been crucial ones though -- which is exactly the small edges that any competitive sports is About.So, let's do some Maths. We do not know exactly, how things work in-game. But let's assume SI are following such stuff, rather than dreaming up a completely fantasy detached from real sports. Also, let's consider these 6 clear cuts to fall mostly in the 20% range. In other words, each of them may have a 80% of being missed. Whilst we know that FM usually flags forwards as "frustrated/nervous" upon missing somtimes, let's discard that for a minute too. It's not the point of this post. Therefore, the roughly Maths for not converting a CCC would look like this: 80% * 80% * 80% * 80% * 80% * 80%. Mathematically thus, that's a 26% Chance of seeing all of them missed. Or roughly one times in four. Even if all the CCCs would fall in the "one on one" range, it would be a roughly 10% chance. Or one times out of ten. It may be slightly lower with the likes of Lukaku --
unless he's frustrated as **** for missing previous. and still not getting subbed.However probabably not that much.
If anybody considers this to be defending the game, they're missing the point entirelly.Maybe one day even SI are gonna wake up and listen -- no less as football may not be like, you know, Maths. But a computer code trying to ape that Football surely is. Right down to its last bit.
Where are you getting your one on one scoring probability from
attached is a counterpoint to above (big chances). Last season of premiership with big chances detailed for certain players...for 20 players attached...the average conversion rate of big chances is 51%...ranging from 70% to 37%...even the much maligned morata converted 37% of his big chances
The average of the top 5 is 63pc
I know you will say big chances can include tap ins but the vardy is top at 70%...impression wise at least he gets a fair amount of chances from balls over the top and running into space for one on ones rather than excess amounts of tap ins...other players on that list are not all about tap ins either and anyway average for 20 players over the course of the season is actually over 51% conversion rate of big chances is significantly higher than the base of 20% which I've seen in one calc for Xg layered on for a big chance.
In essence...big chances conversion rates can and do vary individual to individual...that's just simple math as well
0 -
- Popular Post
32 minutes ago, Mensell76 said:No play out of defense is a whoopiedaisies on my part. Thanks mate, I overlooked it
Next to that, I am not asking people to give my tactic a go. I am asking people to try out the balanced mentality which has a result of more careful weighing of when to pass short and when to pass more direct. And it sort of just releases the attackers, they move more freely.
The no playmaker was intentional because on positive and attacking mentalities they went too much for the killer ball. I am now testing with an mc advanced playmaker. However I do not necessarily want 1 player to receive the ball too much. Playmaking can be done in any role but then divided over more players, if you know what I mean.
The work ball into box is intentional. I do not want to get into the handball defense situation too often. Instead I want to be careful in possession in the first phases of play but in the final third have more room for my IF's to run into. Just like PSV plays in real life
Thing is you're entering the realm of gaming the engine to make stuff work...it shouldn't have to be that way. I'm not saying you are...just trying to get the team to play the way you want...few years back I was using contain to counter the directness which the match engine was producing lol. But basically you're doing things to 'stop' the engine from producing certain predetermined actions rather than doing things to 'make' your team play the way you want it do by tapping into fundamentally sound basic decision making being simulated properly
Conceptually those roles shouldn't be producing free flowing football...a Box to box mid is more of an off the ball impact midfielder getting on the end of things up in top third and up and down the pitch. DM is just that...a defensive midfielder more in an off the ball positional screen and water carrier to provide short passes to the more creative and break up play. A CM can conceivably do additional things being a generic role in a sense
IFs can do things themselves but are also reliant on being slotted in etc and be provided with the ammunition to do so
WBs are more reliant on ball being channeled out wide.
Are you seeing through balls being played in the top third from central positions along the ground and if so what is the frequency of them. These are what is missing from the match engine and essentially is what makes the AP role in the attacking midfield strata redundant from a creative point of view...which ties back into @fidelitywarspost from earlier that this is key to changing the 'feel' of the engine and indeed adding to the actual realism of matching closer back to creative and a more sophisticated way of playing football
The bottom line is the decision making triggers in the match engine are currently producing too much direct football...there will always be ways to mitigate this given the levers SI allows but the underlying decision making isn't simulating high level and even at times base decision making you see in the real world. Players in the real world don't have a trigger point of space = pass over the top...or anywhere close to that.
Stiker/IF off the ball movement isn't dynamic enough or even active laterally, vertically and diagonally to provide options for the more creative players to slot in and even to drag defenders out of position to open up passing lanes for others to be slotted in
Basically the match engine is not simulating pass decision making to move the ball more efficiently into central areas in the top third to have creative players utilise their skillset from those positions to slot players in with threaded passes along the ground...that is the starting point. The trigger point is too early and too deep on the pitch. This prevents quality play from unfolding as it should
The off the ball movement isn't helping that either. Layer on top of correction of pass decision making better movement in the top third by strikers, IFs and AMs to offer passing lanes and passing options more dynamically and not just immediate passing options but consideration of passes after initial passes. One of the premises of Guardiolas methods is a constant flow of movement to create openings by dragging players/defenders out of position to get players in...the immediate pass is seldom the intent of the ball carrier. CIty/Barca are essentially trying to work the ball into the most advantageous position to slot a player into a scoring position with the highest scoring probability. No matter how many ball players or skilled players they have most moves will break down without much defensive effort unless there's a steady stream of fluid movement off the ball
Both of these being addressed should create the best match engine we've seen giving a more balanced and realistic representation of how football can be played in terms of a slicker fluid flow.
It feels close but a million miles away at the same time in terms of the long/deep over the top balls needing to be replaced by the more appropriately played along the ground thread the needle/slide rule passes in the top third dependent on settings of course...but the long deep passes are ok with more direct approach and more direct passing etc but not on other settings...before anyone mentions that lol
As with all things in changing stuff in the engine, it's a behemoth but the team at SI will figure it out. Once they do I think everyone will enjoy the game at a level not seen before in terms of what type of football the engine can produce and enable users to see their teams play the way they want them to but until we're there, we're not there lol
10 -
2 minutes ago, RocheBag said:
I think what he's saying is because over half of the assists are classified as 'other' then they could be anything, including crosses. So we can't get a true feel for what percentage really is from crosses.
Clearly that isn't the case, but it's impossible to prove a negative so he just keeps repeating it.
I hear you on the proving a negative thing
However, whoscored has an actual definition of how they define a cross.
I've just checked their crosses per game for this season against premierleague.com website and by team they correspond with every team bar everton and brighton which are just out by 1 each. So 18 of the 20 teams it is precisely the same crosses/game for whoscored and the premierleague.com website
So that suggests whoscored definition of a 'cross' is consistent with the official premier leagues.
One can therefore conclude that with a consistent definition of what a cross is their assists from crossing will stack up as its an assist from how they define a cross.
By having a clear and consistent definition of a cross whoscored data relating to number of crosses and assists from crosses can be accepted. That they have a large number of assists bunged in to an 'other' category is academic to the exercise of establishing assists from crosses...and we know that because there is a very CLEAR definition of a cross by whoscored
1 -
46 minutes ago, themadsheep2001 said:
I'm an analyst in real life. If your stats list more than half your goal sources as "other", there is no way you can build a credible picture. I've already said early on there needs to be more central play. Still doesn't change the fact whoscored's metric do not even paint close to the full picture
6 hours ago, themadsheep2001 said:And we've already pointed out that we can't verify where whoscored are getting their metrics from (because it doesn't look like other) especially when so many goal sources are marked as other.
I've actually looked at whoscored match report and under chalkboard and any 'indicent' within a match of cross/pass/key pass/shot etc corresponded precisely with any match I've watched in real world. So how they capture the data was very accurate.
What are you saying you can't verify where whoscored are getting their metrics from...is the assists thing...is that what you're saying you can't build a credible picture from and is it in relation to the crossing thing
0 -
1 hour ago, MBarbaric said:
@akkm
as far as I know, not all xG's are the same nor they have the same methodology of determining the outcome. However, most of them do include the position on the pitch where the player took the shot from (depending on who makes the calculation, the pitch is divided in up to 34 zones so central/wide is very much accounted for), the foot (stronger/weaker)... Don't know about the pressure being applied although I'd guess that is also accounted for, at least to an extent. That being said, xG isn't the perfect measurement, but it is as good as it gets at the moment. It is mind boggling why FM doesn't include this in stats as all the data is already in the ME and could be far more accurate than anything collected in real.
Yep..they are dependent on distance from goal/angle of shot etc but i think it's only one model that factors in proximity of defenders (or very few currently) or pressure on the ball...it's more based on as you say and yes not all Xgs are the same which was point i was making about one on ones...I wouldn't classify some one on ones as per what Xg may do...so on that basis the one on one depends on how it's classified...if it's grouping together more difficult angles than centrally and not factoring pressure on ball/proximity of defenders then it's diluting the 1 in 3 stat for this.
0 -
Just to give that one on one context @Svenc...re your example before of pedro on .35xg...tho technically he's briefly one on one with the keeper but from a bad angle...that's actually not a chance Id consider a one on one chance...so it depends on how you classify one on ones...personally i consider them player being slotted through centrally enough with plenty of time and space vs keeper...if I saw that pedro chance you showed before in FM i actually wouldn't remotely class it as a one on one chance.
Most Xg calcs don't factor in proximity of defenders either...yes they weight it for a through ball but proximity of defenders and classification of what a one on one chance is may impact perceptions Vs Xg stats. I expect if you look at more central one on ones with certain aspects of consideration like time on ball to finish, distance from defenders and space to run in to would impact Xg of these chances...currently it seems the calculations of Xg for one on ones don't distinguish angle of approach/proximity of defenders or take into account quality of player in that situation...the xg calcs are more from generic examples...and are all pooled into one.
Xg finishing rates show that better forwards consistently outperform Xg rates so one may expect that applies for one on ones as well.
So all that in context could mean that players will convert more than 1 in 3 and if you narrow down classification of one on ones (to more central chances with further level of granularity to introduce proximity of defenders) could also impact positively (from a forwards point of view) an expected 1 in 3 conversion rate
0
Football Manager 2020 Pre-Release Beta *Official* Feedback Thread
in Football Manager General Discussion
Posted
yeah but only the first one is a proper one on one chance there. the second is a cross and he never gets away from the defender for the third one so how on earth is it a one on one lol...it's a one on two basically being pressured by defender. Only the first chance is morata free from defenders one on one with time and space to do his own thing..which in that instance he executes awfully lol