Hello all, and welcome to the first of Stigsby’s Statistics!
We all have an exciting month ahead of us and with a win against Sporting we can finally focus on it! Lately a subject has been raised by Ramzi that has caught my interest, the effect of wins and losses against rivals, something which is very relevant for the upcoming games. We are all aware of the effect of winning, meaning that you will get 3 points while the opponent will get 0, very easy to see and grasp, but what is more interesting to look at, is the after-effects. Will there be an effect on the next games? Will it have a definite result on the final outcome of the season or will it just apply to the normal point pool?
For most of you out there, I know that you prefer to keep numbers out of football, but stay with me here. Football is a game of 11 vs. 11, the same terms for both teams, but yet one team comes out on top. What is more relevant is, that it is usually the same teams coming out on top and the same teams losing. What I’m trying to say is that there must be some sort of truth in the numbers, after all, everyone predicts Barcelona and Real Madrid to finish in the top 3, and mostly they do. This prediction is a result based on previous seasons, strengths of the teams and a whole lot of other variables, meaning, that your brain basically does a whole calculation for you without you even thinking about it!
To look at the effects of the rival-matches we will first need some statistical data. As we’re not even halfway, the current season will not give us adequate data to perform a precise and accurate calculation, so we will focus on the season of 07/08. During this post I will focus on the top 5 teams, continuing Ramzi’s initial post. One must keep in mind here, that it is not the true rivals we’re focusing on here, but the top title contestants.
Real Madrid 85 2,2 17,9 21
Villareal 77 2,0 16,2 15
FC Barcelona 67 1,8 14,1 7
FC Sevilla 64 1,7 13,5 9
Atletico Madrid 64 1,7 13,5 7
Points Point average Points expected against rivals Points achieved against rivals
The first thing we can look at is the strength of the title contestants within, meaning how well do they perform against each other. To make a fair comparison we will have to do some simple calculations. To begin with we first calculate the point average pr game, this will give us a picture of the general form of the team during the season, by multiplying that with 8, the 8 matches against the other four top 5 contestants, we will get the expected amount of points against rivals. As shown on the first table we can see, not surprisingly, that Real Madrid last season had a very high bottom-level with an average of 2,2, and on top of that, they “over-performed” against rivals, meaning that they in the end took the title. The rest of the top 4 teams have in fact “under-performed” against rivals last season, giving them fever points achieved than statistically expected.
Now that we have looked at the performance of the top teams, let’s look at the effects of the results of the matches. What is normally put forward is the effect on the team moral after a win/loss. It is very difficult to put a number on the feelings of the specific players, so we will have to find another way to calculate the effect. To do that we can look at the results achieved in the first, the second and the third match after the rival match and see how the team performs compared to the seasonal average. In this calculation I have excluded the results after a draw, as a draw will leave both of the teams at point zero compared to each other, and should therefore be expected to have no substantial influence.
Our initial instinct tells us, that a loss against a rival would lead to a loss in team moral and thereby have a negative influence on our future short-term performances, meaning losing a match against a rival, would lead to a greater loss in points in the next three matches than the seasonal average. So let’s look what the actual numbers tells us:
We all have an exciting month ahead of us and with a win against Sporting we can finally focus on it! Lately a subject has been raised by Ramzi that has caught my interest, the effect of wins and losses against rivals, something which is very relevant for the upcoming games. We are all aware of the effect of winning, meaning that you will get 3 points while the opponent will get 0, very easy to see and grasp, but what is more interesting to look at, is the after-effects. Will there be an effect on the next games? Will it have a definite result on the final outcome of the season or will it just apply to the normal point pool?
For most of you out there, I know that you prefer to keep numbers out of football, but stay with me here. Football is a game of 11 vs. 11, the same terms for both teams, but yet one team comes out on top. What is more relevant is, that it is usually the same teams coming out on top and the same teams losing. What I’m trying to say is that there must be some sort of truth in the numbers, after all, everyone predicts Barcelona and Real Madrid to finish in the top 3, and mostly they do. This prediction is a result based on previous seasons, strengths of the teams and a whole lot of other variables, meaning, that your brain basically does a whole calculation for you without you even thinking about it!
To look at the effects of the rival-matches we will first need some statistical data. As we’re not even halfway, the current season will not give us adequate data to perform a precise and accurate calculation, so we will focus on the season of 07/08. During this post I will focus on the top 5 teams, continuing Ramzi’s initial post. One must keep in mind here, that it is not the true rivals we’re focusing on here, but the top title contestants.
Real Madrid 85 2,2 17,9 21
Villareal 77 2,0 16,2 15
FC Barcelona 67 1,8 14,1 7
FC Sevilla 64 1,7 13,5 9
Atletico Madrid 64 1,7 13,5 7
Points Point average Points expected against rivals Points achieved against rivals
The first thing we can look at is the strength of the title contestants within, meaning how well do they perform against each other. To make a fair comparison we will have to do some simple calculations. To begin with we first calculate the point average pr game, this will give us a picture of the general form of the team during the season, by multiplying that with 8, the 8 matches against the other four top 5 contestants, we will get the expected amount of points against rivals. As shown on the first table we can see, not surprisingly, that Real Madrid last season had a very high bottom-level with an average of 2,2, and on top of that, they “over-performed” against rivals, meaning that they in the end took the title. The rest of the top 4 teams have in fact “under-performed” against rivals last season, giving them fever points achieved than statistically expected.
Now that we have looked at the performance of the top teams, let’s look at the effects of the results of the matches. What is normally put forward is the effect on the team moral after a win/loss. It is very difficult to put a number on the feelings of the specific players, so we will have to find another way to calculate the effect. To do that we can look at the results achieved in the first, the second and the third match after the rival match and see how the team performs compared to the seasonal average. In this calculation I have excluded the results after a draw, as a draw will leave both of the teams at point zero compared to each other, and should therefore be expected to have no substantial influence.
Our initial instinct tells us, that a loss against a rival would lead to a loss in team moral and thereby have a negative influence on our future short-term performances, meaning losing a match against a rival, would lead to a greater loss in points in the next three matches than the seasonal average. So let’s look what the actual numbers tells us:
After win
Real Madrid 2 2,3 2,0 2,1 2,2
Villareal 1,4 1,4 1,8 1,5 2,0
FC Barcelona 1,5 3,0 2,0 2,2 1,8
FC Sevilla 1,5 0,0 0,5 0,7 1,7
Atletico Madrid 0 2,0 3,0 1,7 1,7
Overall 1,3 1,7 1,9 1,6 1,9
First match Second match Third match Three game average Season average
After loss
Real Madrid 3,0 1,0 3,0 2,3 2,2
Villareal 1,3 3,0 2,3 2,2 2,0
FC Barcelona 1,4 2,2 2,0 1,9 1,8
FC Sevilla 1,2 1,2 1,0 1,1 1,7
Atletico Madrid 1,6 2,2 3,0 2,3 1,7
Overall 1,7 1,9 2,3 2,0 1,9
First match Second match Third match Three game average Season average
On the first of the two tables we can see the results after a win and on the bottom one the results after a loss.
What is most surprisingly by these results are, that it seems that the teams actually perform better after a loss than after a win. There seems to be a negative effect on the first match right after a rival match, for both a win and a loss. This can easily be explained by the extra effort put into the match by the players, meaning that they will be worn out in the next match. What is interesting here though, is that the effect seems to be bigger after a win rather than after a loss, there actually seems to be a dive in form right after a win, whereas after a loss there seems to be no real effect. What is even more interesting is, that there actually seems to be quite a positive effect after a loss, the team actually performs better over the next three matches after a loss than during the rest of the season.
Keep in mind that this does not have anywhere near enough data included, as I would have to include several seasons and a blend of different leagues, to give a scientifically viable result. This is only a result of the last seasons 32 intra-rival games, however I think it’s safe to draw one vital conclusion: The wins and losses against rivals does NOT have such a vital effect that they are often put out to be. The matches against rivals are merely a battle of prestige and bragging-rights, and are most likely not having a ground-breaking effect on the final season outcome.
Ps: This is merely an attempt to clarify the effects of rival matches. I have chosen to keep the calculations as simple and open as possible, as I do not wish to influence the result of this paper. If you disagree with any of the calculations or have any doubts as to have I made them, please post your criticism.
Pps: Even though my results tends to prove that there is no beneficial effect by winning over the rivals, let’s hope the Barca players never see this article and beat the hell out of them on the 14th of December ;)
Stigsby
Villareal 1,4 1,4 1,8 1,5 2,0
FC Barcelona 1,5 3,0 2,0 2,2 1,8
FC Sevilla 1,5 0,0 0,5 0,7 1,7
Atletico Madrid 0 2,0 3,0 1,7 1,7
Overall 1,3 1,7 1,9 1,6 1,9
First match Second match Third match Three game average Season average
After loss
Real Madrid 3,0 1,0 3,0 2,3 2,2
Villareal 1,3 3,0 2,3 2,2 2,0
FC Barcelona 1,4 2,2 2,0 1,9 1,8
FC Sevilla 1,2 1,2 1,0 1,1 1,7
Atletico Madrid 1,6 2,2 3,0 2,3 1,7
Overall 1,7 1,9 2,3 2,0 1,9
First match Second match Third match Three game average Season average
On the first of the two tables we can see the results after a win and on the bottom one the results after a loss.
What is most surprisingly by these results are, that it seems that the teams actually perform better after a loss than after a win. There seems to be a negative effect on the first match right after a rival match, for both a win and a loss. This can easily be explained by the extra effort put into the match by the players, meaning that they will be worn out in the next match. What is interesting here though, is that the effect seems to be bigger after a win rather than after a loss, there actually seems to be a dive in form right after a win, whereas after a loss there seems to be no real effect. What is even more interesting is, that there actually seems to be quite a positive effect after a loss, the team actually performs better over the next three matches after a loss than during the rest of the season.
Keep in mind that this does not have anywhere near enough data included, as I would have to include several seasons and a blend of different leagues, to give a scientifically viable result. This is only a result of the last seasons 32 intra-rival games, however I think it’s safe to draw one vital conclusion: The wins and losses against rivals does NOT have such a vital effect that they are often put out to be. The matches against rivals are merely a battle of prestige and bragging-rights, and are most likely not having a ground-breaking effect on the final season outcome.
Ps: This is merely an attempt to clarify the effects of rival matches. I have chosen to keep the calculations as simple and open as possible, as I do not wish to influence the result of this paper. If you disagree with any of the calculations or have any doubts as to have I made them, please post your criticism.
Pps: Even though my results tends to prove that there is no beneficial effect by winning over the rivals, let’s hope the Barca players never see this article and beat the hell out of them on the 14th of December ;)
Stigsby
Posts
11 comments:
This is the first contribution to this blog by Stigsby. I want to thank him for that and welcome him to the blog team!
P.S. We'll work on the graphics for next time...
Interesting indeed... Thanks, Stigsby!
Amazing, Stigsby!
Cant wait the rest, you know, its unfair that pep read all the analysis and we here have to wait, i think thats the good thing for being the boss:))
At last moment I limited my statistical analysis in my following preview, I am lucky for doing this or else I will look naive in statistics compared to what you are doing here;)
Looking forward for the rest!
Thanks for the comments fcbee and Ramzi!
Actually Pep doesn't have anymore analysis, just writing it as we go! I thought i'd see how people reacted to this one before i wrote another one, numbers isn't for everyone ;)
I can reveal though that I already have a second one almost done, it's going to be shorter, but not less suprising!
Oh and yeah Pep, we'll just look into the graphics for next one. Maybe adding the tables as pictures instead of text, just to make it easier to understand :)
Stigsby, do u do ur comparison as a bench-marking taking the first team as a reference? or you compare each team with the rest?
because it give me different conclusions for each method than the other
(its rory, i forgot my log in on my other computer!)
I like the addition! but sadly, i work with stats for a day job, so i've gotta add my spin/critique! :)
There's a problem here: there's a natural limit to the number of points a team could possibly score in a game: 3. so with 20 total games between the rivals, there's only 60 possible points to score. and adding up actually achieved points, there were 59 points.
as you have it right now, though, there are a total expected points of 75.2 points. That's not only not the reality, it's impossible! Instead, if you create an average points expected based on the pool of five teams (basically: barca's total points on the season = barca's points against rivals), you'll get a more accurate prediction. that is, create this number:
Barca's total points/ (reals total points +villareals total + barcelona+ sevilla+atletico).
That above number (in barca's case: .201) gives the expected share of the 59 points barca should have gotten based on barcelona's performance against the rest of the league*. In this case, it gives us an expectation of 11.9 total points. Still, the end result is the same: Barcelona grossly underperformed.
I'd love to see the after loss/after win numbers after controlling for the quality of the opponent and for home/away status. but that's probably getting too complicated for the blog. But these are the type of math questions that i enjoy. i know--i'm a nerd.
*this is not an ideal number in the slightest. there are more complicated models, but the different won't be significant and i'm too lazy to do it.
Interesting article. Thnx for your hard work.
Some comments: I don't think you can reasonably say that a team overperforms or underperforms against its rivals unless all things are really equal between teams, and they are most definitely not b/c of player absences, among other things. Your numbers suggest that Barca heavily underperformed against rivals last season, but with Messi, Eto'o, & Ronnie missing for large chunks of the season (and no killer injuries to key players at Madrid & Co.), I might be inclined to say that Barca performed just slightly below par against its rivals.
Rory's suggestion to control for the quality of opponents is a good one. Other variables to control for include time between matches, injuries/suspensions, weather.
Good work, Stigsby, keep it coming. I'm not that well skilledin the matter, so I haven't much to add, but I enjoyed reading it!
Damn, wrote a long long reply to you all but apparently it disappeared, here we go again!
To Ramzi, could you please clarify it abit more? :)
To Rory, great with some constructive critism, keep it coming! Yes i was aware of the 60 total point issue when i wrote this, in the calculation I've done here some matches counts double, the numbers. It was merely meant as a very quick look at the strenght of the teams. The reason the total points exceeds 60 is that this is a calculation done for each team isolated, meaning that if you add the expected points from all the teams you will get the total from 32 matches. It was not meant as a throughtout analysis of that, simple a small look at how the actual points on the table can sometimes cheat. I can hear that you fully know what you're talking about and if you're interested it would be great if you could read through my posts prior to publishing for eventual mistakes? It is easy to make a mistake as the things i'm calculating on is often hard to put numbers on and you can easily lose your head in it. I have a very limited time to write these at the moment so there is the posibility of "too-quick" moments!
If you're interested leave me an email at Christofferfalck@hotmail
To Blaugranafan, this is simply a quick look at how the teams actually perform, not the reasons behind it. The variables you mention there is simply the causes why some teams might over-perform and why some might under-perform. But yes, if we should do a throughtly analysis of the teams, giving them a rating from 1-10 those would some of the first variables to include!
To Refter, thanks for the comment, I'm glad you liked it. Takes alot of work and good comments is what makes it worth it :)
Thanks for all the constructive comments! Keep them coming, it will only make my future posts better and maybe enlighten some new areas!
Ps. A little spoiler here - The next post will look at the away strenght and the final winner of the season!
Post a Comment