HTs on Global/Possession

From Hattrick

Keywords: (Possession)

By: HT-Hasse 2237636.13 as reply to 2237636.9
To: Jazz23 25.12.2004 at 22:31
No I want 56% to be 56%!!!

That should be 5 chances for one and 4 for other team... not always but average in 100 or 1000 games should be that! 'Cause now team with 56% gets about 75% of chances...

No. On average a team with 56% possession does get 56% of the chances. But you forget that not all chances are reported. The chance has to be good enough to be presented in the match report.

Keywords: (Possession)

By: HT-Hasse 2237636.16 as reply to 2237636.14
To: Jazz23 25.12.2004 at 22:41
their defence is to weak to stop opponents attack so if they won 5:0 we can say that they got all chances.

Sorry, but you can't. It does not work that way.

Keywords: (Possession)

By: HT-Hasse 2237636.20 as reply to 2237636.19
To: Jazz23 25.12.2004 at 22:44
And your point is?

Besides, you cannot tell that it is 6:2 in chances. It is 6:2 in reported chances, which is a completely different thing.

Keywords: (Possession)

By: HT-Hasse 2237636.31 as reply to 2237636.30
To: Jazz23 25.12.2004 at 23:17
If it is a normal chance a team with 54% posession has 54% chance of getting it.

Keywords: (Possession)

By: HT-Hasse 2237636.48 as reply to 2237636.47
To: el-insaciable 25.12.2004 at 23:55
Actually your statement in this thread made me wonder... When it's calculated which team will get a chance, is there a randomfactor which decides how great the chance is? Or is it only the ratings (with an amount of random too) which has an influence?

I don't know exactly how that works, and I don't want to know too much. It would spoil the fun of the game for me.

Keywords: (Possession)

By: HT-Tjecken 2942648.123 as reply to 2942648.109
To: redrobert 19.04.2005, at 10:19
the possession that is mentioned is from the 45th/90th minute. that's what ht-bjorn is saying and everything i observed so far proves this right.

That's correct.

Keywords: (Possession)

By: HT-Tjecken 2942648.125 as reply to 2942648.124
To: Ratsia 19.04.2005 at 11:03
And it's looking more and more like it is too non-linear, so that even a small difference in possession averages a huge difference in how the chances are allocated.

For example, the difference in that particular match was by no means small. Elkabong's opponent had roughly 35% higher midfield rating. This is roughly the same as difference between average div III and div I teams in relatively established countries (difference measured by Hatstats, which is linear function of ratings and thus comparable to studying the difference of individual ratings). Yet, like you said, Elkabong was a favourite in that match, even though his midfield was in a sense two division levels below the opponent's midfield and the only part of the ratings that was better for him was attack.

So, the question is "what is a small difference". I would never call 30-50% difference small, regardless of the rating in question. Getting 30-50% advantage on some area requires significant investments. Assuming identical formations (and other variables) this means roughly 30-50% better players. For example, "excellents instead of passables" or "supernaturals instead of outstandings". IMO those are huge differences.

A 10% difference indeed is quite small, and in case of midfield it will give 52% possession and thus probably on average slightly less than 6 chances for the dominating team. This will not be a decisive thing in the match. Even 20% difference, which is not that small anymore, will only give 55% possession and probably 6-7 chances. It's still relatively easy to counter that by dominating other ratings.

Fix it up so that it is linear or at least close to it.

It could be tweaked a bit, but a linear system would be quite catastrophical. If the scoring probabilities would be linear as well, there really wouldn't be anything for us to play, because it would not matter how you position your players (strong side versus weak or vice versa and so on). Linearizing midfield alone would not cause exactly the same effect, but it would make midfield have almost no value at all if scoring formula remained the same as it is now.

Very nice post in a very nice thread I must say. I love to read those threads like this where users really try to discover new stuff instead of just stating the same old myths again. All users participating and reading this thread will/can actually even learn a lot from it (apparently even including elkabong this time :)).

Keywords: (Possession)

By: HT-Tjecken 7831771.115 as reply to 7831771.1
To: Everyone 24.09.2007 at 21:10
Reply to (8995830.883)

If i remember correctly, Hasse once said that chances are distributed proportionally to possession.

If that's true, then Hasse was most probably hallucinating.