Luketa

I stated in my original post that I more that usually make tweaks in my tactics. I make the correct decisions most of the time, because, generally, when I make this "in game" tweaks, I neutralize the opposition domination, if this is the case. This game is not so mysterious at all. I CAN see the effects of my tweaks. But here the point is that it is not enough with this in FM, the game punishes the player with ridiculously often undeserved looses.

Chand

The engine does tend to overcompensate too much. The only reason I can't bring myself to accept this as a fact is simply that I have no conclusive proof of this. Only the feeling we all get, and is common when these things happen, and some bits of circumstantial evidence.

The problem with identifying this overcompensation is that it manifests itself a little in several areas of the game. As stated above and elsewhere many times:

- Teams with numerical inferiority coming back against really hard odds.
- Otherwise normal keepers that manage to play blinders against human teams.
- Teams of lesser quality having little problems to take the game to better teams.

... and so on. We all know the symptoms but we can't point at the disease. The most sensible explanation would be that the game does tend to overcompensate in its favor. Whether this is a mechanic to nullify supertactics or if it has another purpose, no one really knows and it's impossible to tell without having trained eyes look at the code itself.

The only area in which I did a little bit of digging (don't have much free time anymore, I'm afraid) was regarding AI keepers. Collecting stats along a season or two. The issue of normal, unspectacular AI keepers earning MoM awards after several of my games caught my eye for some reason, so I started to collect their numbers throughout a couple of seasons.

My intent was to find out if there was anything there, in their seasonal stats, that would show if these few keepers were just having good seasons despite their stats, or if the AI was simply raising its game and keepers against me. If MoM were any indication (and they are, to a degree), these keepers would not be strangers to receive these awards. That is, if they were excellent performers themselves, or if they were having good seasons, they would earn MoM against many other AI teams, not just my human team. However, if at the end of the season it turned out that these keepers few (or only) MoM award was against me, and I was somehow meeting these keepers all the time, then it'd be logical to conclude there was something else behind it.

After collecting a couple of seasonal averages, I came up with a final average of 9.2. That is, all conditions being equal, any keeper had a 9.2% chance of earning a MoM award at any given game. What this means was that statistically, on average, any team should encounter a team with a MoM keeper roughly 3.5 times after 38 games.

So far, so good. However, when I went to check my team's history, turns out that I had faced keepers that went on to be awarded MoM at the end of the game five times per season. 42% over the seasonal average.

In other words, whereas all AI teams (as per their own seasonal stats) faced a MoM keeper about three times per season, my team faced a MoM keeper five times (as per my team's own seasonal stats at the end).

Of course this is not conclusive of anything, but then again the numbers don't lie. Human teams seem to be around 40% more likely to encounter a MoM keeper than AI teams. That number will change up and down, of course, as keepers, teams, leagues and tactics change, but I doubt it would move too far from that.

And that's where the overcompensation might be. Or at least one of the most evident symptoms. It'd be great if this was retuned and fixed, because if this is the case, it makes no sense at all. Good keepers are good keepers, regardless of the opposition. Sure they can have good days and bad days, but 40% more good days against human teams just because they're human teams is a tad unfair, not to mention unrealistic.

One thing is making a game challenging, another thing is being a jerk.

2ndchance

Good post Chand!.

I have been supporting my claims with evidence all along, although for a different reason.

I was sick of being told that those games when you dominate with a hatful of chances and the AI wins by scoring with its only shot or two on target, evens itself up over the course of a season or two.

So i started posting screenshots of every games final stats and so far, noone has come up with a sensible explanation, in fact more and more people seem to agree, even if they were totally against what i was trying to prove and i'm continuing to do so with the faint hope that it just may finally start to be taken seriously?.

Basically the stats show that the most i can usually expect from games that i have been outplayed in is about 3 or 4 points a season(usually draws), whilst the AI take a whopping 20-25 points!.

At the end of the day the stats show that if i get outplayed even slightly i will lose, whilst i can outplay the AI sometimes 30 shots to 1 and still lose.

However much anyone wants to stick up for SI and FM07 the stats dont lie, unless of course i was somehow doctoring them?, then all i can say to people is "do the test yourself" and i'm sure many more will be suprised with the results!.

Marseille07

Chand,

The # of MoM keepers do not occur often in the first place. I'd suggest using average ratings instead, which gives you more accurate information with regard to keeper's performance.

My feeling right now though, is that superkeepers do not exist. I do not really get 1-shot-wins-game situations much in this version.

Of course, lack of superkeepers does not mean AI is not getting other forms of boosts though.

Chand

No, I didn't use the word 'superkeepers' because I didn't want to come across as wanting to get the discussion down that road.

To your point, when I ran everything discarding MoM and focusing on each keeper's seasonal average, all my results told me those keepers had always been within 5% (on average) from their seasonal average. In fact, I should say closer to 3%.

And that's one of the other things that keep me from just stating categorically there's something going, because my numbers regarding these keepers are giving me two different outcomes, so to speak.

On one hand, when I look at the seasonal averages, all keepers were within 3-5% of their average when facing me. So, no cause for concern. And yet, from the same numbers, I realize that season after season, regardless of what team I'm managing, tactics used, squad, league, etc... my human team ends up facing more MoM keepers than the AI teams do. If this had been a localized, precise fluke I would have discounted it. But the only reason I decided to get the numbers was that it was happening often enough for me to notice it.

I still don't know how to compatibilize on one hand all keepers being within 5% of their seasonal average (which is negligible), and on the other the increased amount of keepers I face that end up being MoM. The only thing I can think of is that for some reason, my human team manages to catch the 'spikes' in these AI keepers seasons. I catch keepers in more good days than the AI does, but the chances of that to happen all the time, season after season, regardless of all other factors are ridiculous. I should be playing the lottery with this kind of luck.

I have to conclude, not seriously, that the engine overcompensates against me, even if it's just a little bit every time. However, I have no solid proof of this other than these numbers. Either the AI overcompensates when it faces me sometimes, or I'm catching keepers on good days way more often than the AI does.

Luketa

Thats exsctly what I am trying to say, if not a cheat, there is some kind of compensation because many, many people are experiencing the same. I posted examples that show the unreal "luck" the AI has. If there was some kind of logic, the times that the human player and AI grab undeserved victories would be more balanced.

Marseille07

Chand,

Three MoMs vs Five MoMs doesn't really tell much statistically. That was the reason why I proposed the seasonal average, which seems to be within the reasonable range.

Now it may be the case that AI are indeed overcompensated; but we do not have statistical data to prove it beyond reasonable doubt, at least not from the GK ratings analysis.

x42bn6

I've seen worse.

Marseille07

Would you like to elaborate more on that?

2ndchance

Take a look at the Screenshots!