|
|
|
|
|
|
|
Jens 'n' Frens
Idle thoughts of a relatively libertarian Republican in Cambridge, MA, and whomever he invites. Mostly political.
"A strong conviction that something must be done is the parent of many bad measures."
-- Daniel Webster
|
|
|
|
|
|
Monday, July 09, 2018 :::
I probably enjoy sports differently from many Americans, but would posit that one of the reason that many Americans like our football better than soccer is well-captured by the kurtosis of changes in win-probability graphs. Look at these games from fivethirtyeight.com :
This is, in fact, a pair of games that is somewhat generous (in the relevant sense) to soccer, in that the former in particular is unusually high-scoring. What you see in the graphs is ten jumps with regions of gentle drift in between. Compare this to the chart from advancednflstats.com :
This is, in fact, a game that is somewhat ungenerous (in the relevant sense) to football, in that it is a particularly close game. It is still a much more continuous graph; there are occasional "big plays", but the eleventh-biggest moment of the game has much more impact on the final result than in soccer.
One can advance arguments about information sets or the like (and perhaps a graph based on an active live-betting market would better make the point), but there's no plausible way the most expert of soccer observers could enrich the practical state-space of soccer enough to appreciably smooth down the soccer graph; possession and field position count for something, but not very much. Soccer advocates occasionally snark that the problem that soccer is low-scoring could be solved by multiplying the goals by 7 as football (more or less) does, but this confuses the issue of "scoring" with the more fundamental issue: even in a football game with 18 points between the two teams (7 times the average total score of world cup matches this year) will have meaningful progress toward points that is far less ephemeral than exists in soccer.Labels: sports
::: posted by dWj at 1:29 PM
|
|
|
|
