PDA

View Full Version : in-depth analysis of CCC's



baaf
13-11-2010, 13:26
When i started my first save of FM11, i was quite shocked by the amount of CCC's that my team missed. It was even more mind boggling when u take into consideration that i played as man city.
About 30 games (overall not just PRM) into the season I really wanted to come here and whine about it but i thought that would be just sad, so i decided to actually crunch some numbers so i can back my self up.

Through my 1st 30 games i generated about 3.3 CCC's per game, and had a conversion rate of less than 25%, which is pathetic to say the least even if you are not man city. But then i thought about how it may be bad for both the AI and the user so i checked the stats of every opp. that i had played against and compiled their stats in the games they played against me. The AI generated 0.83 CCC's per game and had a conversion rate of 34%.

I was really mad at the game and was on the verge of opening a rage thread, BUT, i just had to compile numbers through the whole season to make it not look as a fluke run.

And to my surprise, things started to get a little bit better and my team clicked. By the end of the season (63 games played) i had the following numbers:

user: CCC's = 187 (2.96 per game) , HC's (half chances)= 170 (2.69 per game) , goals = 124 (1.96 per game) ,
(CCC's converted , HC's converted) = (61 , 28)


AI: CCC's = 50 (0.79 per game) , HC's (half chances)= 88 (1.39 per game) , goals = 42 (0.66 per game) ,
(CCC's converted , HC's converted) = (15 , 5)



So..... to the conclusions:

1) 32.62% of CCC's were converted by the user, while the AI teams converted 30%.
this still bugs me. my team is miles ahead of 90% of the league in terms of talent, yet they still barely out-perform them. One in three one on ones converted is kinda "meh" for a top team, imo it should be more around 50% for a team of this quality.


2) 16.47% of HC's were converted by the user, while AI teams converted 5.6%.
the AI is horrendous when it comes to HC's, which in most cases is still a kind of one on one but with tougher angles or with defenders around you. I think the conversion rate of those chances should be bumped a bit.

3) 24.92% of overall chances (CCC's and HC's) were converted by the user, while the AI teams converted 14.49%.
the AI teams suffer because of the HC's, but all in all, if the conversion rates of CCC's and HC's are bumped up a bit it should be ok. I mean really, top teams aren't this bad and bad teams aren't as crappy as this rate shows.

4) 28.23% of user goals were not through "real chances" (ie CCC's and HC's), while AI teams had a 52.39%.
this annoys me the most, it basically means that the AI score through improbable shots/screamers/tough headers more than half of his goals. and its not like my team doesn't shoot long or lacks good headers, even if i put long shots to rarely they still shot around 8 long shots a game, which is MUCH more than the AI teams shoot from the same distance, yet i can't come even close to their conversion rate.
long shots really need some tuning down for the sake of everyone.


I really wanted to do the same analysis on an AI team like arsenal or manU, but i really don't have the time :(
that info can be really helpful to make better conclusions and eliminate any doubt about ME bias towards AI.

this is NOT a rant post, it was going to be a couple of days ago but not any more. Im really interested what you guys have to say about those stats :)

baaf
13-11-2010, 23:37
***bump***

Shaka Sangoma
14-11-2010, 00:27
This is really interesting stuff, and a lot better than the "my striker missed 2 shots, game is broken rant." But have you considered contrasting this data with RL Premier League data, simply to find out if Chelsea, who are in front at the moment, convert 32% of their chances? Your argument would have a lot more credibility that way

Tubey84
14-11-2010, 00:33
I actually agree with your points about HCs and the long shot tuning from the AI.

However, I disagree with everything else. The CCC conversion rate is literally spot on at about a third. Additionally, the overall chance completion is accurate if you consider that as the user you're creating a lot more chances from short distant with the AI getting the odd few; if the AI wasn't scoring so many from range then the overall chance ratio would be about equal. Similarly, the opposition only have one or two headed opportunities per game compared to ALOT more for the user, so again you'd expect the conversion rate to be higher for the user.

So the stats are distorted by the problems with the sheer amount of long shots from the AI, meaning fewer CCC's and crosses for HC's from the AI - if that makes sense!

phnompenhandy
14-11-2010, 02:35
This is really interesting stuff, and a lot better than the "my striker missed 2 shots, game is broken rant." But have you considered contrasting this data with RL Premier League data, simply to find out if Chelsea, who are in front at the moment, convert 32% of their chances? Your argument would have a lot more credibility that way


It is a 'good rant', but as said above, SI do these analyses. Miles has spelled out in great detail how the stats for FM correlate very closely to real-life stats, certainly at the top level. Indeed, the main cause of the dreaded "supergoalie" was that certain tactics were creating an unrealistic number of chances so either the opp keeper had to save a lot of there would be cricket scores. Now, instead of superkeepers, we have players missing more chances. Seems balanced to me.

nev147
14-11-2010, 02:44
I guess there are many different CCC's. But the ones that annoy me are the one on one's where the player only has the keeper to beat, no defenders around him and has plenty of time. The % of them need to be increased and maybe other forms of chances decreased.

But nice piece of work baaf, and I very much agree and I have been saying it for years.

Sheva Elite
14-11-2010, 03:38
It's a bit of a joke to be fair.

They miss easy chances way too often. A CCC should be finished 70%+ even by crap players. Its an easy chance. Gaping goal usually. HC should me more like 50/50 to at least hit the target cos it is a tighter angle.

And the AI gets nearly all their goals against me in this way. Highlights start as a throw on. They'll throw it, cross it and either score the header or the scramble. Yes my defenders are good in the air and gelled as a team etc. It's either that or an outrageous 25 yarder where their player doesnt even take a run up. Basically toe pokes it in from 25 yards and bends it in the top corner. with 10 long shots.

Martyr1777
14-11-2010, 03:40
I don't have any hard stats, too lazy for that lol, but just had to add my two cents about the long shots.

The comment about AI scoring about half their goals from improbable/long shots really seem accurate in my main save. My teams isn't losing almost at all, but seems like every time I turn around there is a long shot in the back of my net. Mean while I have every single player on my teams set to rare long shots because they never even get it on target.

Even the player on my side with long shot, finishing, etc stats that are the same (if not better) then the players scoring the long shots has literally not gotten one even on target let alone in. Insanely annoying especially considering he has the 'shoots from distance' ppm and I can't get it off him.

I don't remember seeing this many long shots going in with 10.3 especially difference with mine not being on target... at least in 10.3 my side had 2-3 long range goals. But like I said, no hard stats and not even finished with a full season but it looks odd.

baaf
14-11-2010, 09:32
1 in 3 CCC's going in seems a bit inaccurate for top clubs... take barcelona's game last night against villareal, they had 4 CCC's (one of them was called an offside even though it wasn't), they scored three of those (including the "offside"), and messi had one HC and he put it in.
the only period where barca didn't convert chances at a good rate was when villa was in dreadful form (and that didn't stop messi from scoring everything). here it seems that every striker is in bad form the whole season, i mean if a top class striker botched 2 one on ones in the same game the fans will be all over him.

it might be true though that the user creates too many chances, but then again, real madrid/barcelona/chelsea have been winning with crazy results the past couple of years.

tingting
14-11-2010, 09:52
A new FM game and another CCC thread and once again the same explanation of "look at the real life stats". Last year i remember two long threads about this topic..more importantly about a supposed bible of a research paper conducted by some arm chair football enthusiast who had some time on his hands and whose research paper was bandied about as the truth in this forum by some of the most reputed members who were involved in how the game will shape. I had asked a simple question but they were more interested in a personal sledging match with the OP of that thread. So i ask it here again...

Two points: 1) Did the researcher who conducted that study took all conditions into account when coming up with the numbers or did he just pick a 100/500/1000 chances and based his study on it

2) Do SI really believe that their ME accurately reflects real life?

If you are going to reflect real life stats into the game, then you should make sure that you reflect real life responses/decisions into the game as well. But the fact is that while the ME is the best one around in the market...it is a very very very far away from being realistic. And if you are going to reflect real life stats into the game, then what is the use of attributes. If the player in the game is going to convert just over a third of his CCCs then what is the use of his Pressure, Finishing, Composure, Technique and Balance ratings? What about the ability (or lack of) of the goal keeper? What about the pitch conditions or the atmosphere of the game or the position from where the player is taking the shot? Maybe SI should just display his conversion ratio so that we can buy a player who has the best conversion ratio?

I dont want another long thread about CCCs but for those who are pointing to real life stats...they are applicable only to a realistic ME not an imperfect one