4 ms·

That was a really interesting read, thanks for sharing it. > Impect disregards all passes that go backwards or don’t beat any defender. It’s a correct assessme

That was a really interesting read, thanks for sharing it. > Impect disregards all passes that go backwards or don’t beat any defender. It’s a correct assessment, a pass that doesn’t travel forwards, doesn’t help you score. Can’t argue with that. This is a really odd statement. Plenty of passes don't go forward but absolutely do increase the likelihood of scoring. For example, the final pass here [0] is backwards, might even allow an additional defender time to track back (-1 packing?) but opens up the space required to take a shot and score. [0] https://www.youtube.com/watch?v=w0oihU_fLas https://www.youtube.com/watch?v=w0oihU_fLas

hinkley 8y ago
Isn’t a pass going backward about avoiding a defender they can’t get past? Useful backward passes fall into at least two categories. The ones that eventually result in an impect, and the ones that run down the clock at the end of the game.
- harlanlewis 8y ago
  And a third category, whatever Spain was doing against Russia.
- Harvey-Specter 8y ago
  > Isn’t a pass going backward about avoiding a defender they can’t get past? Yes, that's basically what I'm saying is bad about the statement that backwards passes don't help you score goals. > Useful backward passes fall into at least two categories. The ones that eventually result in an impect, and the ones that run down the clock at the end of the game. I think this is almost correct, but I think there is a third category: Backwards passes that immediately result in a shot resulting in a goal. In the example video I posted above, the final pass is backwards and is immediately shot into the net. This doesn't increase impect (ie: Messi dribbles past 4 players, then shoots and scores past 3 more, is only impect +4, according to the article).
mnarayan01 8y ago
The article looks like it's just a generic sports' journalist reporting on a novel statistical technique, with cross language translation added into the mix for extra confusion. I wouldn't take it too literally. I'd guess that either "Impect disregards all passes that go backwards or don’t beat any defender" would be better phrased as "Impect disregards passes unless they go forward or beat defenders" or the statisticians would defend the original solely in terms of it being a heuristic that makes the problem more tractable. Additionally, your linked example, with #25 picking up the ball in a danger position after a deflection, is the kind of "penalty box slop" that, naively at least, seems extremely difficult to handle analytically.

hinkley 8y ago

Isn’t a pass going backward about avoiding a defender they can’t get past? Useful backward passes fall into at least two categories. The ones that eventually result in an impect, and the ones that run down the clock at the end of the game.

harlanlewis 8y ago
And a third category, whatever Spain was doing against Russia.
Harvey-Specter 8y ago
> Isn’t a pass going backward about avoiding a defender they can’t get past? Yes, that's basically what I'm saying is bad about the statement that backwards passes don't help you score goals. > Useful backward passes fall into at least two categories. The ones that eventually result in an impect, and the ones that run down the clock at the end of the game. I think this is almost correct, but I think there is a third category: Backwards passes that immediately result in a shot resulting in a goal. In the example video I posted above, the final pass is backwards and is immediately shot into the net. This doesn't increase impect (ie: Messi dribbles past 4 players, then shoots and scores past 3 more, is only impect +4, according to the article).

mnarayan01 8y ago

The article looks like it's just a generic sports' journalist reporting on a novel statistical technique, with cross language translation added into the mix for extra confusion. I wouldn't take it too literally. I'd guess that either "Impect disregards all passes that go backwards or don’t beat any defender" would be better phrased as "Impect disregards passes unless they go forward or beat defenders" or the statisticians would defend the original solely in terms of it being a heuristic that makes the problem more tractable. Additionally, your linked example, with #25 picking up the ball in a danger position after a deflection, is the kind of "penalty box slop" that, naively at least, seems extremely difficult to handle analytically.