just_COYB
Player Valuation: £8m
Based on this work: http://pena.lt/y/2012/12/03/applying-the-pythagorean-expectation-to-football-part-two/
I know it's early, but I was curious to see who has more points than they would be expected to by their goals scored and goals allowed.
I just quickly took the league table, calculated the expected points (using the formula from the above article by Martin Eastwood), and saw how the predictions differed from reality.
I've coded some columns (GF, GA, PTS, Expected PTS and Difference) to make the scale easier to see. The higher the difference, the more points the team has than would be predicted by the historical relationship between GF, GA and PTS.
We can clearly see that while Chelsea lead the league by quite a margin, Southampton's 8-0 thumping gave them such a huge differential that it suggests that they should be top. If we just cut the score in half to 4-0, we get this:
This makes more sense, with Chelsea and Southampton joint top, and City just behind. I'm going to work from here, throwing out the 8-0 victory as a bit of surrender by Sunderland. Also, two own goals is rare (Liverpool were even luckier with their pair of OGs), and five goals were scored after 2/3 of the game was over.
Looking a little lower, we see that the teams fighting relegation in our virtual points world are Villa, Burnley and QPR. Villa and Burnley are similar, scoring only a goal per two, the same as Jelavic(!). QPR are the third worst offense, and have passed us as the worst defense in the Premier league by allowing a worrying 2.25 goals per game. These three seem right to me, but of course Villa have had tough opponents, too. Still, scoring 1/3 as many as your opponents seems a sure way to join the Championship soon.
So whose points tally doesn't quite matchup with their expected points? Chelsea and Villa have nearly 5 points more than they would be expected to have. Chelsea have been great this year on breaking the game open and then taking control defensively. Still, it seems that they have been winning more games than they might be expected to just based on GF and GA. Villa have won small (1-0 and 2-1) and lost big (3,3,2,2 without scoring in any match), that's easy to figure out, really. Liverpool also have 2 more points than expected.
On the other side, Everton come in as the team which has under-performed most, being nearly 2 points lower than would be expected. In the expected points table, they're 9th. Crystal Palace and Southampton both are about a point less than they would expected to be, but everyone else is pretty close to their expected value, even after only 8 games (the expected points explains approximately 81% of the variation in Actual Points):
I will try and remember to update this post periodically, but I think this type of analysis gives evidence to our thoughts like "How are Villa above us?", "Liverpool are pretty average, how are they 5th?" and "Chelsea aren't that good, really." Despite how early it is to ask these questions, stats can help show us why we have these feelings about teams, and give them a little support.
I hope you guys enjoyed this. COYB!
I know it's early, but I was curious to see who has more points than they would be expected to by their goals scored and goals allowed.
I just quickly took the league table, calculated the expected points (using the formula from the above article by Martin Eastwood), and saw how the predictions differed from reality.
I've coded some columns (GF, GA, PTS, Expected PTS and Difference) to make the scale easier to see. The higher the difference, the more points the team has than would be predicted by the historical relationship between GF, GA and PTS.
We can clearly see that while Chelsea lead the league by quite a margin, Southampton's 8-0 thumping gave them such a huge differential that it suggests that they should be top. If we just cut the score in half to 4-0, we get this:
This makes more sense, with Chelsea and Southampton joint top, and City just behind. I'm going to work from here, throwing out the 8-0 victory as a bit of surrender by Sunderland. Also, two own goals is rare (Liverpool were even luckier with their pair of OGs), and five goals were scored after 2/3 of the game was over.
Looking a little lower, we see that the teams fighting relegation in our virtual points world are Villa, Burnley and QPR. Villa and Burnley are similar, scoring only a goal per two, the same as Jelavic(!). QPR are the third worst offense, and have passed us as the worst defense in the Premier league by allowing a worrying 2.25 goals per game. These three seem right to me, but of course Villa have had tough opponents, too. Still, scoring 1/3 as many as your opponents seems a sure way to join the Championship soon.
So whose points tally doesn't quite matchup with their expected points? Chelsea and Villa have nearly 5 points more than they would be expected to have. Chelsea have been great this year on breaking the game open and then taking control defensively. Still, it seems that they have been winning more games than they might be expected to just based on GF and GA. Villa have won small (1-0 and 2-1) and lost big (3,3,2,2 without scoring in any match), that's easy to figure out, really. Liverpool also have 2 more points than expected.
On the other side, Everton come in as the team which has under-performed most, being nearly 2 points lower than would be expected. In the expected points table, they're 9th. Crystal Palace and Southampton both are about a point less than they would expected to be, but everyone else is pretty close to their expected value, even after only 8 games (the expected points explains approximately 81% of the variation in Actual Points):
I will try and remember to update this post periodically, but I think this type of analysis gives evidence to our thoughts like "How are Villa above us?", "Liverpool are pretty average, how are they 5th?" and "Chelsea aren't that good, really." Despite how early it is to ask these questions, stats can help show us why we have these feelings about teams, and give them a little support.
I hope you guys enjoyed this. COYB!
