The Forest from the Trees: A post for the stats nerds

Nocturnal 2017-07-26 00:29:05 UTC #41

Yeah no doubt the media are a long way behind. Some of the amateur data scientist would be worth them following.

ivan 2017-07-26 02:25:02 UTC #42

Which is the blog with the mid, fwd, def ratings with colour coding which showed it midfield was ■■■■ but other areas were tracking well

ivan 2017-07-26 02:50:09 UTC #43

Don’t worry found it.

barnz 2017-07-26 02:52:07 UTC #44

share it?

ivan 2017-07-26 02:59:45 UTC #45

It was linked in the op

Interesting to note improvement in our mid ranking over the year

barnz 2017-07-26 03:35:20 UTC #46

Hmm, that i50 stat is kinda useless though to calculate the teams midfield worth.

Ours a few years ago was considerably above other teams but it was just get ball from centre bounce, smash it 40 metres forward, other team rebounds.

BringBackBoris 2017-07-26 03:54:37 UTC #47

Which would have greatly reduced our fwd line score. It’s not a perfect measure (what is?), as gameplans will always vary but it’s interesting to note we’ve been in improving an area that was a weakness.

It’s certainly felt like we’ve improved too. Someone on here likened us to an Italian soccer team earlier in the season, sitting back and trying to sting on the counter. The last 4-6 weeks I feel like games have been much more on ‘our’ terms, but we are still getting used to maintaining an effective forward press (Hartley coming on in leaps and bounds here for mine)

ivan 2017-07-26 04:23:17 UTC #48

it would be interesting to create a measure of midfield output, if you had the necessary data, using the following:

disputed balls wins
1a retention of disputed balls win (for/against)
1b. uncontested possessions generated from disputed balls (for/against)
1c contested possession generated from disputed balls (for/against)
1d outnumber situations at disputed balls (for/against)
clearances
2a. retention of clearance (for/against)
2b. uncontested possessions generated from clearance (for/against)
2c contested possession generated from clearance(for/against)
some sort of pressure index including tackles and pressure acts
number of uncontested inside 50 entries
number of uncontested inside 50 entries against
number of contested inside 50 entries
number of contested inside 50 entries against
unforced skills errors (for and against)
forced skills errors (for and against)
turnovers (for and against)
Number of meters per possession
ball retention from inside 50 (for and against)
13.distance of final possession inside 50 (for/against)
14.depth of inside 50 (for/against)
degrees to goal of inside 50 (for/against)

some of these would obviously be influenced by how good/bad the forward/defense is but it is trying to measure how well the midfield shields the defense and how clean the opportunities presented to the forwards are.

TrevorBix 2017-07-26 04:40:32 UTC #49

You could definitely calculate that from champion data’s advanced package, however at the cost of 100 dollars per game it would cost a bit.

ivan 2017-07-26 04:44:04 UTC #50

Is that even an option for the public

TrevorBix 2017-07-27 03:48:46 UTC #51

yep, I think you get 500 variables at 100 bucks a game (not live).

ivan 2017-07-27 03:52:53 UTC #52

We should do syndicate and buy a few games. Anzac Day, port, Sydney, St Kilda would provide an interesting data set

ivan 2017-07-27 03:57:35 UTC #53

Actually we should probably Chuck in a stinker like the second lions game

TrevorBix 2017-07-27 04:17:27 UTC #54

If we can scrape together 100k a year we can get 500 variables with live 30 second updates!

If I had 100k though I would buy 5 years of data and update AutoTippa to be the greatest predictor of all time. Calling all seed investors.

barnz 2017-07-27 04:31:13 UTC #55

no brisbane? I need some stats to prove how that calamity happened.

Killer_Mike 2017-07-27 04:51:51 UTC #56

frosty 2017-07-27 07:15:09 UTC #57

If those aren’t flag numbers I don’t know what are.

BringBackBoris 2017-07-27 07:39:14 UTC #58

Strongly mirroring the metrics from HurlingPeopleNow, as others have noted our midfield is improving dramatically (granted, from a base where we were 17th in the league) to complement our elite bookends.

Sunday’s game will be a huge test to our system, as the dogs whole philosophy is to strangle teams in the middle by throwing extra players at the contest, effectively hitting us at our weakest point. If we can score a convincing win the lid will be the furthest off it has been in my mind since 2003

andrewb 2017-07-27 07:56:23 UTC #59

We should be sending those stats straight to Devon Smith and Josh Kelly’s managers. One gun mid away from a flag.

Killer_Mike 2017-07-27 09:56:28 UTC #60

Based on the top four defence and attack metric we are in premiership territory. We obviously need some luck from here due to the slow start but even if we don’t make it this is a clear indication that Woosha has us going in the right direction.

The core of our group being so young means with more experience we are only going to get stronger. Good times are coming.