Forum

xml	No replies possible in the archive

Author: bert.kassies
Date: 19-09-2008, 19:26
Edited by: bert.kassies
at: 19-09-2008, 19:44

The first table of two-leg statistics based on first result was given by JPV in 2004 (see e.g. archived topic).

See below for an updated table based on all European Cup matches of the last 30 years (from 1979 until 2008). Only two-leg results in qualifying and knock-out rounds are used. The total number of matches is 4532. Only results based on a minimum of 10 matches are shown.

The table shows:
a. 1st leg result (home team vs. away team)
b. number of times that the home team advances
c. number of times that the away team advances
d. total number of matches with the same 1st leg result
e. chance for home team to advance
c. chance for away team to advance

 7-0    13-  0 ( 13)  100% /   0%
 6-1    21-  0 ( 21)  100% /   0%
 6-0    28-  0 ( 28)  100% /   0%
 5-0    67-  0 ( 67)  100% /   0%
 4-0   113-  0 (113)  100% /   0%
 5-1    40-  1 ( 41)   98% /   2%
 3-0   236- 20 (256)   92% /   8%
 4-1    72-  8 ( 80)   90% /  10%
 5-2    12-  2 ( 14)   86% /  14%
 2-0   339- 74 (413)   82% /  18%
 3-1   163- 43 (206)   79% /  21%
 4-2    25-  9 ( 34)   74% /  26%
 1-0   352-227 (579)   61% /  39%
 4-3     5-  5 ( 10)   50% /  50%
 2-1   154-159 (313)   49% /  51%
 3-2    45- 57 (102)   44% /  56%
 0-0   141-288 (429)   33% /  67%
 3-3     5- 14 ( 19)   26% /  74%
 1-1   108-316 (424)   25% /  75%
 2-2    29-121 (150)   19% /  81%
 0-1    38-317 (355)   11% /  89%
 1-2    12-177 (189)    6% /  94%
 2-4     1- 20 ( 21)    5% /  95%
 2-3     2- 59 ( 61)    3% /  97%
 1-3     2- 97 ( 99)    2% /  98%
 0-2     3-181 (184)    2% /  98%
 0-3     0- 97 ( 97)    0% / 100%
 1-4     0- 45 ( 45)    0% / 100%
 0-5     0- 36 ( 36)    0% / 100%
 0-4     0- 35 ( 35)    0% / 100%
 1-5     0- 17 ( 17)    0% / 100%
 0-6     0- 10 ( 10)    0% / 100%

The correspondence with the previous JPV table is within 5%.

Author: bert.kassies
Date: 19-09-2008, 19:53

Some statistical noise is visible in the table above. E.g. it is strange that the 1st leg result 4-3 is better for the home team than 3-2. This is probably caused by the low number of matches (10) for the 4-3 result. If only results based on a minimum of 50 matches are shown the table looks better.

 5-0    67-  0 ( 67)  100% /   0%
 4-0   113-  0 (113)  100% /   0%
 3-0   236- 20 (256)   92% /   8%
 4-1    72-  8 ( 80)   90% /  10%
 2-0   339- 74 (413)   82% /  18%
 3-1   163- 43 (206)   79% /  21%
 1-0   352-227 (579)   61% /  39%
---------------------------------
 2-1   154-159 (313)   49% /  51%
 3-2    45- 57 (102)   44% /  56%
 0-0   141-288 (429)   33% /  67%
 1-1   108-316 (424)   25% /  75%
 2-2    29-121 (150)   19% /  81%
 0-1    38-317 (355)   11% /  89%
 1-2    12-177 (189)    6% /  94%
 2-3     2- 59 ( 61)    3% /  97%
 1-3     2- 97 ( 99)    2% /  98%
 0-2     3-181 (184)    2% /  98%
 0-3     0- 97 ( 97)    0% / 100%

Re: Two-leg statistics based on first result (2)
Author: Overgame Date: 19-09-2008, 20:45	Too bad we haven't got the results for the last 10 years, we'd see some differences (as far as i remember, 0-0 is close of 50/50 now)

Author: bert.kassies
Date: 19-09-2008, 21:34

And here are the results of the last 10 years:

Total number of matches: 1884

 5-0    20-  0 ( 20)  100% /   0%
 4-0    48-  0 ( 48)  100% /   0%
 3-0    95-  4 ( 99)   96% /   4%
 2-0   128- 14 (142)   90% /  10%
 4-1    33-  4 ( 37)   89% /  11%
 3-1    72- 13 ( 85)   85% /  15%
 1-0   140- 73 (213)   66% /  34%
 2-1    70- 59 (129)   54% /  46%
 3-2    19- 18 ( 37)   51% /  49%
---------------------------------
 0-0    68-121 (189)   36% /  64%
 1-1    51-140 (191)   27% /  73%
 2-2    14- 56 ( 70)   20% /  80%
 0-1    20-136 (156)   13% /  87%
 1-2     7- 80 ( 87)    8% /  92%
 2-3     1- 27 ( 28)    4% /  96%
 0-2     2- 88 ( 90)    2% /  98%
 1-3     1- 47 ( 48)    2% /  98%
 0-3     0- 50 ( 50)    0% / 100%

Some differences with respect to the 30-year table can be seen. But please note that these results are based on less matches so the noise margin is higher. But e.g. the 2-1 result (based on quite some matches) gives now a significantly higher chance for the home team.

Re: Two-leg statistics based on first result (2)
Author: bert.kassies Date: 19-09-2008, 21:48	It's remarkable that all chances for the home team are higher in the 10-year table. So, it seems that first playing at home is now less a burden than in the past.

Re: Two-leg statistics based on first result (2)
Author: Lusankya Date: 19-09-2008, 23:20	I think it doesn't really matter if you play you second game at home or not.

Re: Two-leg statistics based on first result (2)
Author: Kaiser Date: 20-09-2008, 00:02	It matters. According to statistics 60% teams which play 2nd leg at home proceed to the next round (2006/07 season).

Re: Two-leg statistics based on first result (2)
Author: Cirdan Date: 20-09-2008, 00:39	@Kaiser: I would be not that sure about this statistic, since playing the second match at home goes often with being seeded (at the moment in the CLR2 and the UCR3). I have no idea if there are enough matches where the seeded team plays away first to completely explain a 60-40 chance, but it definitely influences that.

Re: Two-leg statistics based on first result (2)
Author: Osiris Date: 20-09-2008, 08:00	Thanks, Bert. It is a very useful information. Just today I discussed the probability of a host passing after 1:0 win (on Russian site eurocups.ru ) I thought it is about 55% bur now I see it is more than 60%.

Re: Two-leg statistics based on first result (2)
Author: psw Date: 20-09-2008, 10:14	@Bert >And here are the results of the last 10 years: >Total number of matches: 1884 But statistics presented in this message gives the total number of matches only 1719 (with 789 positive results for home, 0.459 prob.). So near 9% matches (from 1884) missed this statistics. Why this value is so large?

Re: Two-leg statistics based on first result (2)
Author: wisla_krakow_fun Date: 20-09-2008, 10:23	Very interesting statictic. I remember match Wisła Kraków with Lazio Roma in 1/8 UEFA cup in 2002/2003. In Roma was 3:3, but in Cracow Wisła lost 1:2 ...

Author: Integral
Date: 20-09-2008, 10:29

Another possible explanation of the interesting fact that after the same score in the first leg nowadays "first at home" team has more chances to progress than 30 years ago is the improving of european refereeing. It is not a secret that some (not very experienced) referees are more tolerant to the home teams. In 1970s and 1980s I guess there were much more countries where unexperienced referees could be influenced by local authorities (I do not mention bribing, there are many "fair" ways to put a person under pressure). Nowadays the overall refereeing seems to be much more professional and unbiased. That's why in the second leg there is no additional advantage to the home team due to refereeing.

Author: psw
Date: 20-09-2008, 11:33

No, data presented show no obvious time dependencies for the probability of positive result for the 1st leg home team.
Really based on 30 years stat we have the probability
p1 = 2026 / 4461 = 0.454
The same prob. based on last 10 years stat is
p2 = 789 / 1719 = 0.459
So p1 = p2 within stat bounds.
BTW estimated deviation d for probability estimate p for a given statistics N is
d = sqrt(p * (1 - p) / N)
So we can not say about any differences when these differences are too small and are within this deviation interval.

Author: bert.kassies
Date: 20-09-2008, 12:14

psw said: "But statistics presented in this message gives the total number of matches only 1719 (with 789 positive results for home, 0.459 prob.). So near 9% matches (from 1884) missed this statistics. Why this value is so large?"

The 1st leg results for which the 10-year statistics are given are the same as those in the 30-year list based on a minimum of 50 matches. Effectively for the 10 year list that implies a minimum of 20 matches with same 1st leg result.

The remaining 165 1st leg results in the last 10 years are:

 7-2     1-  0 (  1)  100% /   0%
 6-2     2-  0 (  2)  100% /   0%
 8-1     2-  0 (  2)  100% /   0%
 8-0     3-  0 (  3)  100% /   0%
 7-0     4-  0 (  4)  100% /   0%
 5-2     6-  0 (  6)  100% /   0%
 6-1     8-  0 (  8)  100% /   0%
 6-0    12-  0 ( 12)  100% /   0%
 5-1    15-  0 ( 15)  100% /   0%
 4-2     9-  4 ( 13)   69% /  31%
 4-3     2-  1 (  3)   67% /  33%
 3-3     2- 10 ( 12)   17% /  83%
 0-4     0- 16 ( 16)    0% / 100%
 0-5     0- 15 ( 15)    0% / 100%
 1-4     0- 13 ( 13)    0% / 100%
 2-4     0-  8 (  8)    0% / 100%
 1-5     0-  8 (  8)    0% / 100%
 3-4     0-  4 (  4)    0% / 100%
 1-6     0-  3 (  3)    0% / 100%
 0-6     0-  3 (  3)    0% / 100%
 5-3     0-  2 (  2)    0% / 100%
 2-5     0-  2 (  2)    0% / 100%
 2-6     0-  2 (  2)    0% / 100%
 0-7     0-  2 (  2)    0% / 100%
 4-4     0-  1 (  1)    0% / 100%
 3-5     0-  1 (  1)    0% / 100%
 1-7     0-  1 (  1)    0% / 100%
 2-8     0-  1 (  1)    0% / 100%
 0-8     0-  1 (  1)    0% / 100%
 0-10    0-  1 (  1)    0% / 100%

Author: bert.kassies
Date: 20-09-2008, 12:18

Cirdan said: "since playing the second match at home goes often with being seeded (at the moment in the CLR2 and the UCR3)"

That indeed can influence the statistics. Buth not too much. The majority of two-leg home/away matches are played in the qualifying rounds. Currently in CL there are 60 matches random versus 8 matches where the seeded team is the away team in the first match. And in UC there are 125 matches random versus 16 matches where the seeded team is the away team in the first match. So in both cases around 12% of the total number of matches have the seeded team as the away team in the first match. Before 2004 the UC matches were 100% random.

Re: Two-leg statistics based on first result (2)
Author: psw Date: 20-09-2008, 12:35	@bert.kassies Thanks. As a workaround the statistics combining can be proposed when the results with a fixed goal difference (4-2, 5-3, 6-4 etc) are placed in the same slot (named e.g. others with diff. +2 etc.)

Author: Ricardo
Date: 22-09-2008, 10:56

Bert, thanks for this, it is really interesting. I already saw in dome topic on forum 2 that 4-3 in first leg is better for the home team then 2-1 and wondered how this could be. An explanation can be that if there are so many goals in the first leg, then it is more likely that in the second match also many goals will be, 7 goals means that the away team was there to make goals, to win, but still they couldn't do that.
On the other hand a 2-1 is something a team can 'accept' in an away game, it is more a result of a defensive strategy: make (at least) 1 goal, and do not lose by more than 1 goal. A loss is acceptable..

Author: psw
Date: 22-09-2008, 15:13

@Ricardo
The problem is that we can not state (based on current statictics only) that 4-3 or 3-2 is better (or worth) that 2-1 after the 1st leg.
Really let's consider 30 year statictics as the most full one.
The probability of positive 2-legs result after 2-1 score will be
0.49 +/- 0.028. The same probability after 3-2 score is 0.45 +/- 0.05
So statictically these values are the same (we should reject the hypothesis about different statistics for these variants).

Author: Ricardo
Date: 22-09-2008, 16:33

@psw,
OK, I don't know enough about statistics to challenge your statement. I don't know where you deviations of 0.028 and 0.05 come from. You state it as the truth and won't argue that.
What I tried to explain is why it could be true that 3-2 is better for a home team then 2-1.

Could you tell me in how many years the deviation has been reduced to a 'certain' statement about if 3-2 is better than 2-1, assuming that the same ratio of 3-2 and 2-1 matches appear and the same ratio of home/away qualification continue..?

Author: 5UCLGSteams
Date: 23-09-2008, 00:45

@ Bert "It's remarkable that all chances for the home team are higher in the 10-year table. So, it seems that first playing at home is now less a burden than in the past."

It may be less a burden than in the past but it is still disastrous stats for home teams in 1st leg I guess.

After drawing at home in the first leg an average of almost 25% (1/4) of the teams do manage to go through after the away second leg game. To me this shows that in case of a draw it is more likely the away team was better and the draw is just also a very good result for them rather than the home team was unlucky with the draw.

Also a 1 goal defeat at home in 1st leg is critical for more than 80% of the teams meanwhile a 1 goal defeat away in 1st leg is critical for an average of 60 % of them...despite it is "the same thing" ...a 1- goal defeat. And yes a "4-3 is better than a 3-2" for the away team looks ok for me because scoring away once or twice could be more of being also lucky meanwhile when you score three times it certainly shows they can very easy score against their opponents from the very beginning of the tie even under "away pressure"...That's also the reason why 2-0 is weaker than 4-2 for the visitors.If you cannot score away there is a bigger chance that you won't be able to defeat that strong defence on your home ground too,meanwhile scoring 2 away goals in the first leg under away pressure shows it is more likely you will have no problems scoring also at your home ground and you can focus more on how to concede less.
After 2-0 away, the main problem for the visitors is how to break that solid defence, after 4-2 it is rather how to concede less,as they will more likely be able to score again .

The most impressive for me are the stats for "2-2" draw with "81%" chances for the visitors to go through after the 1st leg.It is amazing really, it's still a draw , but then 4 out of 5 home teams are already out...

...and this also shows some very interesting things i guess...even if the visitors were let's say "2-0" down , they were able to have an immediately (and not "postponed"

) comeback, meanwhile after leading "2-0" away you both relax easily and send the home team into "all or nothing " until the end, that;s why I also guess 1-2 is so strong despite only 1 goal advantage for the visitors after the first leg.

I would conclude that all these stats show more of the Power of Reaction after the 1st goal of the tie is scored and it might be usually faster for the better team of the tie. I guess in the future it would be very interesting studying not necessary the final scores of the games, but the power of the opponents reaction after the first goal(s) is scored .. let's call it "PAFG" ... what do you think about this?

Anyway , very nice and useful stats. 10x Bert for all your work.

Re: Two-leg statistics based on first result (2)
Author: bert.kassies Date: 23-09-2008, 10:22	Ricardo said: "Could you tell me in how many years the deviation has been reduced to a 'certain' statement about if 3-2 is better than 2-1, assuming that the same ratio of 3-2 and 2-1 matches appear and the same ratio of home/away qualification continue..?" The deviation used by psw d = sqrt(p * (1 - p) / N) is explained in the following wikipedia articles Margin of error and 68-95-99.7 rule.