Home Field Advantage in EL

[elkjiaer is back]

Hi All,

I was recently performing some analysis of the Home Field advantage in Europa League taking into account all matches played since 2009/2010 to the current season (1440 group stage matches in total).

Then I had this idea to look at the % of home and away wins per Match Day just to see if there was any difference. I have realized that there is something about MD2 which I can not really explain..
50.4% of Home wins (second highest is MD5 with 47.9%), lowest % of Away Wins (24.5%)
Second highest average number of Home goal scored and lowest number of away goal scored

And all those values are statistically significant (meaning that it is not just by chance...)

When I look at the UCL the behaviour is more or less as I would expect, in other words there are more home wins in the crucial matches (MD4 and MD5) compared to the others and the same goes with gols scored and conceded.

So In other words my question for you is the following: how can we explain this "weird" behaviour in EL for MD2? is there any contributing factor I am overlooking? to my knowledge the calendar is also compiled at random, so we can not say that MD2 is when for example all teams from Pót1 face teams from Pot4 at home.

[Dragonite]

The Easternmost teams always play MD2 at home (to play the MD6 away since it’s too cold to play there in December)?

They also always play MD5 at home, which according to you is the second highest.

[elkjiaer is back]

The Easternmost teams always play MD2 at home (to play the MD6 away since it’s too cold to play there in December)?

They also always play MD5 at home, which according to you is the second highest.

Thanks this is actually a very smart and interesting hypothesis indeed . But since teams for cold climate countries aren’t indeed the strongest ones how do we explain such a gap in performances ? In MD5 it could indeed be pretty cold over there but in MD2? Anyway this could be the reason why we don’t see this in CL since only Russian teams play MD5 at home ( not many other regular teams from other cold countries )

[Lorric]

Well, you did all this work, so I'm sure you could also check the % of home wins by these Eastern teams, and see if it is higher than the MD2 %. That will tell you if they're the ones moving the needle.

Trips to Eastern teams are generally regarded as tough fixtures. Not all of them, but more often than not.

[elkjiaer is back]

Thank you all for your inputs.
I have digged a little more into this just to find out if there is any patter. For now I am just trying to see if Uefa has some rules for allocating teams from specific countries to play Home/away in MD6.
In the period 2009/10 to 2018/19 this is the percentage of matches played away for some countries:
100% Belarus, Finland, Kazakhstan, Latvia, Norway, Russia, Sweden
75% Moldova and Scotland

We should also remember that, in the case of Scotland for example, Celtic and Rangers when both playing in EL like the current season, may not play at home on the same matchday since they are teams from the same city. So in this case we have two rules contracting each other.

If we now compare this list of countries with the one used still by Uefa for the draw of the Euro 2020 qualification:
Winter venues:Belarus, Estonia, Faroe Islands, Finland, Iceland, Latvia, Lithuania, Norway, Russia, Ukraine.

So when comparing the two lists (neglecting those countries which never played in a group stage in the last 10 years) we can see the following differences:
Kazakhstan and Sweden are considered "cold winter venues" for CL and EL but not for Euro qualifications
Ukraine is only considered "cold winter venue" for Europ qualification but not for CL and EL , even if when you look at average temperatures there in December they are not much different than other countries in this list.

[Jens]

I can confirm this observation.
To check if it might be expected, I compared it with result "probabilities" I calculated from bookmaker odds.
The differences in expected values are small from matchday to matchday, while the actual results show a clear deviation.
However, 240 matches is not THAT much.
I will try to find out which countries are "overperforming" on MD2.
By the way, last matchday is the one with the most surprises indicated by a low SPCR (lower value means more underdog wins).

Code: Select all

     │ exp.HW  exp.D  exp.AW │ HW   D   AW │  SPCR  │   M
─────┼───────────────────────┼─────────────┼────────┼───────
 MD1 │ 114.50  60.75   64.75 │ 100  69  71 │ 101.17 │  240
 MD2 │ 109.52  60.74   69.74 │ 121  60  59 │ 105.32 │  240
 MD3 │ 111.32  59.76   68.93 │ 106  65  69 │ 103.70 │  240
 MD4 │ 114.30  58.89   66.81 │ 112  60  68 │ 106.75 │  240
 MD5 │ 111.33  60.14   68.52 │ 115  55  70 │ 105.03 │  240
 MD6 │ 114.91  59.95   65.15 │ 107  55  78 │  98.84 │  240

exp.HW │ expected home wins
 exp.D │ expected draws
exp.AW │ expected away wins
    HW │ actual home wins
     D │ actual draws
    AW │ actual away wins
  SPCR │ sum of probabilities of correct results
     M │ total matches

[Todor]

By the way, last matchday is the one with the most surprises indicated by a low SPCR (lower value means more underdog wins).

This can be expected as many teams which have nothing to play for, field their B or C squads.

[elkjiaer is back]

I can confirm this observation.
To check if it might be expected, I compared it with result "probabilities" I calculated from bookmaker odds.
The differences in expected values are small from matchday to matchday, while the actual results show a clear deviation.
However, 240 matches is not THAT much.
I will try to find out which countries are "overperforming" on MD2.
By the way, last matchday is the one with the most surprises indicated by a low SPCR (lower value means more underdog wins).

Code: Select all

     │ exp.HW  exp.D  exp.AW │ HW   D   AW │  SPCR  │   M
─────┼───────────────────────┼─────────────┼────────┼───────
 MD1 │ 114.50  60.75   64.75 │ 100  69  71 │ 101.17 │  240
 MD2 │ 109.52  60.74   69.74 │ 121  60  59 │ 105.32 │  240
 MD3 │ 111.32  59.76   68.93 │ 106  65  69 │ 103.70 │  240
 MD4 │ 114.30  58.89   66.81 │ 112  60  68 │ 106.75 │  240
 MD5 │ 111.33  60.14   68.52 │ 115  55  70 │ 105.03 │  240
 MD6 │ 114.91  59.95   65.15 │ 107  55  78 │  98.84 │  240

exp.HW │ expected home wins
 exp.D │ expected draws
exp.AW │ expected away wins
    HW │ actual home wins
     D │ actual draws
    AW │ actual away wins
  SPCR │ sum of probabilities of correct results
     M │ total matches

Thanks Jens this is a great analysis. So I have performed my own calculations only for MD2 using Elo ratings instead of bookies and removing countries which have played less than 3 times in the EL since 2009/2010.
The results are somehow quite surprising, but I am looking forward to compare those with yours.
It basically looks like the countries which are overperforming on MD2 are:
Cro, Mol, Ned, Ita, Kaz, Su, Sco, Bel, Cze etc.. in this exact order

The ones underperforming (from the worst on ):
Bls, Rom, Fra, Ukr, Cyp, Den, Azb, Ger

based on those data I honestly do not see any clear pattern..and defintely we can not conclude that the significant different we see on MD2 is the to countries defined as winter venues.

@Jens: how did you collect all bookmakers odds? is there any easy way to do it from a website or so? If you have any database could you please send me a Private message?`I would be ver interested to dig more into this

[Ricardo]

How do you define 'expected home wins'
On what rule do you expect a homewin? The better uefaCoefficient at that time, the bookmaker odds at that time(odds in what country), the predictions in Lyonnais prediction game?
Did any of these also shows this behaviour of rising in MD2?

I see there is a SPCR column, but how is this related to the expected..

[Jens]

Btw, another thing you see is that draws seem more likely to happen on MD1 than on MD6 which is rather intuitive to explain.

@Jens: how did you collect all bookmakers odds? is there any easy way to do it from a website or so? If you have any database could you please send me a Private message?`I would be ver interested to dig more into this

Yes, kind of. Since starting to do comparisons of Clubelo and Euroclubindex with bookmakers I have been using Oddsportal. I take the average of all available bookmaker odds for an individual match. If you need more info how to get the right settings, just write me a PM.

I still have my old Excel sheet from 2014/2015/2016 where I did the mentioned analysis. Results I provided on Twitter.

How do you define 'expected home wins'
On what rule do you expect a homewin? The better uefaCoefficient at that time, the bookmaker odds at that time(odds in what country), the predictions in Lyonnais prediction game?
Did any of these also shows this behaviour of rising in MD2?

I see there is a SPCR column, but how is this related to the expected..

I calculate the individual "probability" for an home win from the odds and then simply add up the values of all 24 matches, so I get the "expected" number of home wins. Then, naturally, add up the values of the 10 seasons for the specific matchday.

In that sense, the schedule is balanced. It's not like all favourites play at home on MD2.

From my point of view, it's hardly explainable by any schedule characteristics. After all, maybe just randomness/noise?

About SPCR:
My initial thoughts were wrong. It depends on how heavily there is an imbalance in the outcome probabilities in the individual matches, e.g. 95% for home win. Assumed all matches were like this, the SPCR value will be closer to the total number of matches.
But since the schedule is usually balanced, you can say it's a rough measure if more favourites won (resulting in higher SPCR value) or more underdogs won (or draw) (lower SPCR value).

[greenbay]

The Easternmost teams always play MD2 at home (to play the MD6 away since it’s too cold to play there in December)?

They also always play MD5 at home, which according to you is the second highest.

Thanks this is actually a very smart and interesting hypothesis indeed . But since teams for cold climate countries aren’t indeed the strongest ones how do we explain such a gap in performances ?

Still possible.

Cold climate countries usually come with a huge travel distance for the away side. As they are usually located at the edge of UEFA, never in the center. Maybe four hours of flight instead of one hour cramped into the seat.

Cold climate countries often do have lots of empty real estate between the cities. Maybe no international airport in the vicinity. So a three and a half hour bus ride afterwards, not just a 45 minute transfer to the hotel. Therefore a late arrival, no chance to have a training session on the travel day compared to a short trip.

I would not be surprised that if you compute the average home win percentage compared to the away team travel distance, there will be significant results. I'm sure that you'll play better football if your trip on the day before only lasted a few hours instead of needing a full day.