Post by QPR Report on Jan 17, 2009 1:08:19 GMT
When get to stats, I get a little confused!
From The Times - Daniel Finkelstein
Red-card myths are dismissed by statistics
My oldest son is not a big football fan. Unlike his brother, goals, and who scores them, are not of much interest to him. He does, however, like a good red card. Games in the garden never last very long because there are only two of us to start with and I always get sent off in the opening minute.
I do, however, see his point. The drama, the remonstrations, the early bath. Is there a finer sight? The only blot on the red-card landscape is the punditry. As the player begins to walk, someone is bound to say that teams are harder to break down with ten men. Can this be true? Well if it was, managers would begin matches by fielding fewer players than their opponents, wouldn’t they? The Fink Tank, however, feels that a little bit of precision is called for. So Joel Minsky, together with Dr Henry Stott and Dr Ian Graham, has been on the case.
A simple way of assessing the impact of a red card is to see whether it changes the destination of the points. Most sendings-off take place late in the game, so looking only at where the points go on average may underestimate their impact on performance.
However, the damage is still considerable. This is where the Fink Tank computer model comes in. We are able to establish at any given point in any game how many goals we would expect a team to score or concede in the time left. We can therefore look at 958 games where one red card was shown and see what impact it had on the scoring and conceding rate of the ten-man team.
On average the team who had a player sent off were expected to obtain 1.25 points, making them the underdogs before they began. By the time they got a red card they were falling behind expectations and had only 1.02 (if the game had been stopped at that moment). By the end, on average, they had only 0.82 points. In other words, a red card produced a 20 per cent fall in the number of points the team could expect.
We then examined four hypotheses. The first was a simple one. When a team are winning and go down to ten men, their scoring rate will decrease. If an attacking player is sent off, they will not replace him.
But if it is a defender, they will take off a striker and bring on a defensive substitute. True? Yes. The average expected number of goals to be scored after a red card drops significantly, from 0.46 to the observed rate of 0.26.
Similarly, the hypothesis that if a team are behind when the red card is shown their conceding rate will increase, is true. The team are chasing the game, so they will ensure that they keep a strike-force on the park and sacrifice defending. The model shows a rise in conceding, again a significant one, from an expected 0.45 to an observed 0.72.
Fascinatingly, however, the remaining two hypotheses turned out to be wrong. We thought that a winning team would protect their goal and a losing team would keep scoring. But it seems that they do not. Winning teams concede significantly more than we would expect and losing teams score significantly fewer.
This leads us to question the strongest part of the red-card myth. The idea is that reduced player numbers are offset to some extent by a boost to the team’s psyche.
The numbers suggests that this might not be true. Teams are doing worse in parts of the field where they have the same number of players as their opponents.
www.timesonline.co.uk/tol/sport/football/fink_tank/article5532755.ece
From The Times - Daniel Finkelstein
Red-card myths are dismissed by statistics
My oldest son is not a big football fan. Unlike his brother, goals, and who scores them, are not of much interest to him. He does, however, like a good red card. Games in the garden never last very long because there are only two of us to start with and I always get sent off in the opening minute.
I do, however, see his point. The drama, the remonstrations, the early bath. Is there a finer sight? The only blot on the red-card landscape is the punditry. As the player begins to walk, someone is bound to say that teams are harder to break down with ten men. Can this be true? Well if it was, managers would begin matches by fielding fewer players than their opponents, wouldn’t they? The Fink Tank, however, feels that a little bit of precision is called for. So Joel Minsky, together with Dr Henry Stott and Dr Ian Graham, has been on the case.
A simple way of assessing the impact of a red card is to see whether it changes the destination of the points. Most sendings-off take place late in the game, so looking only at where the points go on average may underestimate their impact on performance.
However, the damage is still considerable. This is where the Fink Tank computer model comes in. We are able to establish at any given point in any game how many goals we would expect a team to score or concede in the time left. We can therefore look at 958 games where one red card was shown and see what impact it had on the scoring and conceding rate of the ten-man team.
On average the team who had a player sent off were expected to obtain 1.25 points, making them the underdogs before they began. By the time they got a red card they were falling behind expectations and had only 1.02 (if the game had been stopped at that moment). By the end, on average, they had only 0.82 points. In other words, a red card produced a 20 per cent fall in the number of points the team could expect.
We then examined four hypotheses. The first was a simple one. When a team are winning and go down to ten men, their scoring rate will decrease. If an attacking player is sent off, they will not replace him.
But if it is a defender, they will take off a striker and bring on a defensive substitute. True? Yes. The average expected number of goals to be scored after a red card drops significantly, from 0.46 to the observed rate of 0.26.
Similarly, the hypothesis that if a team are behind when the red card is shown their conceding rate will increase, is true. The team are chasing the game, so they will ensure that they keep a strike-force on the park and sacrifice defending. The model shows a rise in conceding, again a significant one, from an expected 0.45 to an observed 0.72.
Fascinatingly, however, the remaining two hypotheses turned out to be wrong. We thought that a winning team would protect their goal and a losing team would keep scoring. But it seems that they do not. Winning teams concede significantly more than we would expect and losing teams score significantly fewer.
This leads us to question the strongest part of the red-card myth. The idea is that reduced player numbers are offset to some extent by a boost to the team’s psyche.
The numbers suggests that this might not be true. Teams are doing worse in parts of the field where they have the same number of players as their opponents.
www.timesonline.co.uk/tol/sport/football/fink_tank/article5532755.ece