Inside Sport

what passing statistics predict a league success this year


https://forum.insidesport.com.au/Topic1838177.aspx

By grazorblade - 26 Nov 2013 2:17 AM

procrastinating at work :-"

I decided to look at a few statistics from the a league so far and compare teams passing game. The statistics I compare with points per game (rather than points since wp and ccm have played 1 less) are

Pass attempts, pass completions, pass success rate, attempted passing speed (number of attempted passes divided by time in possession), successful passing speed (number of successful passes divided by time in possession), attempted passes in the attacking third, succesful passes in the attacking third, passing accuracy in the attacking third, attempted passing speed in the attacking third and successful passing speed in the attacking third.

There is a lot of noise which you would expect since we are only looking at one aspect of the game however, here are a few surprises.

1. Passing speed (attempted or successful) is by far the biggest predictor of success. The strongest correlation is for attempted passing speed in the attacking third. The correlation is incredibly clean, only a couple of teams deviate from the norm

2. there is no correlation between possession and points but perhaps that is because it is early in the season

In order the best predictor to weakest predictor is

1. Passing speed in attacking 3rd (attempted)
2. Successful passing speed
3. Passing speed in attacking 3rd (successful)
4. Attempted passing speed
5. pass attempts in the attacking 3rd
6. pass accuracy in the attacking third
7. pass completions in the attacking 3rd
8. pass completions
9. pass attempts
10. passing accuracy
11. possession


I suspect that teams with the best attacking 3rd succesful passing speed will the best when we have more rounds however what is significant is that all 4 measures of passing speed appear in the top four!

Here is the ladder for passing speed in the attacking third

1. Brisbane
2. Wellington
3. victory
4. Sfc
5. WSW
6. MH
7. Newcastle
8. CCM
9. Perth glory
10. adelaide

The two outliers are melbourne heart and wellington pheonix. Perhaps this indicates they might come good, perhaps it means other parts of their game are particularly week. WSW and SFC look like outliers but they are actually very close to the trend line which is a very tight trendline

It seems that passing speed (succesful or attempted), particularly in the attacking 3rd may be an under appreciated predictor of success
By Neanderthal - 26 Nov 2013 3:45 PM

grazorblade wrote:
Neanderthal wrote:
Fascinating stuff GRazor. Very interesting.

grazorblade wrote:
does anyone know if crosses are counted as passes?

If you're talking about OPTA stats, then I'm pretty sure crosses are NOT counted as passes.
At least they definitely weren't last season for the A-League.
Edit: just checked and it's the same this season.

grazorblade wrote:
2. there is no correlation between possession and points but perhaps that is because it is early in the season

Aside from the already mentioned Adelaide dilemma... Another factor that may skew the correlation between possession and success is that an A-League team who is leading a game will often sit back and concede more possession to the other team while in front and just defend their lead.

That means if one team scores a couple in the first half, they will often sit back and let the other side accrue a high possession stat while just defending their lead. That makes losers more likely to have a higher possession stat.

To get rid of this and see the true correlation between possession and getting ahead in games then I guess you would have to only use data during times that the game is still up for grabs.



My team Sydney FC are a classic example of a team who try to win possession while the game is up for grabs but then sit back as soon as they're two goals up or 1 goal up with less than ~half an hour to go.

possibly....I suppose if that were the case you might still expect a correlation between goals scored and possesion? I might check this?

Not really. You would need to section out periods of the game.
I have no idea how you could do it given the limitations of the current app.
For now we might just have to concede that it's a deceptive stat that gets skewed by certain tactics.