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


#682

I think thats partly due to not having Joe or Smack as the main target for most of the year.


#683

Low expectated would indeed mean we are taking lower percentage shots. Usually this means:

  • set shots from tough angles (I.e the pockets) rather than in the corridor
  • a high ratio of shots from general play (thankfully TIPPA is converting more than his fair share of these)

#684

Although what you say seems correct - bear in mind that the variation is minimal.
Try and imagine what the difference between an ‘expected conversion’ of 50% vs one of 48% may look like, given the only way that percentage was arrived at.


#685

https://cdn4.theroar.com.au/wp-content/uploads/2019/06/Chart-3-copy.jpg

Interesting to see the shift is gameplans across the league. “Small ball” is on the way out, Richmond, Bulldogs, Melbourne the teams still invested in the territory+pressure system; only one of those making it actually work.

The trend in 2019 is possession and minimising scores from turnovers. We are actually on trend with our tactics for the first time in years, but we are only average at it. Hopefully improvements to our forward line will improve our turnovers and carry us to an improved back half.

Source https://www.theroar.com.au/2019/06/06/numbers-game-which-afl-teams-are-set-to-rise-and-fall-in-the-second-half-of-2019/


#686

Just read the article on the secret revolution of analytics in AFL and wasn’t terribly surprised to see that Essendon was one of the sides that doesn’t have either a full time or part time analyst on board.

God I hope that article is not accurate because analytics have proven to be successful across many sports throughout the world and the Hawks proved it can work in our sport when they had an analyst on board for 3 of their recent flags.


#687

link?

Not overly surprising given our shambolic off field organisation until very recently.

That said, I feel like we are more “up with” current trends than at any point since Hird got our press going. The current push towards possession football, even by sides that were run and gun 2 years ago (Adelaide is almost unrecognisable in their approach), jumps out as an analysts response to current data. If 2/3rds of opposition scores come from turnovers, minimising these would be the first thing an impartial analyst would recommend imo.


#688

I think along with injury management analytics is the area we can gain the greatest advantages.

Whether it’s through list management or game plans there is a lot of advantages.

I would love to know just how much influence Darren O’Shaughnessy has on the hawks flags.

Clearly Clarkson believes in it because he is very much a system based coach.


#689

Cheers.

Clarko was out of the box from day 1. In footballistics they detail how when he signed on he sacrificed part of his salary to bring in an academic who was an expert in kicking biomechanics (hope that’s the correct term). Together they developed the Hawks gameplan and recruited/trained players accordingly.


#690

Yeah, I think we have the basis of a good side and I’m not as down on the gameplan as others but I believe we aren’t playing the players in the positions to get the best value out of them.

We do very well at a lot of important metrics but the one thing that continues to let us down is kicking efficiency. Or more importantly kicking efficiency in the most important parts of the ground.

The trouble that Woosha finds himself in is that he’s playing each week for his job so he can’t experiment as much as I’m sure he would like too.

I’m one that really hoped we would figure highly this year but I think the best we can hope for now is to gain a little respect and hopefully build for next year.


#691

We would be WAY behind on this outside looking in

Clarkson’s decision to recruit kicking skills and develop kicking efficiency through biomechanics was clearly based on data.

I would suggest that Clarkson had a less skilled group of footballers than what Thompson had at Geelong but managed to get better results.


#692

We would have to do some analytics but to what extent I’m not sure.

With the amount of data that Champion Data produces sides would be crazy not to invest in that area of the game.


#693

Look at the impact serious technical tuition has on the Irish guys


#694

Liverpool have 5 PhD s on staff for analytics! Thats kinda awesome


#695

“Football Analytics: if you’re not doing it, you’re nowhere” says guy hired by club as a Football Analytics guy.

Listened to the episode of Trends with Binuk Kodituwakku. Interesting personal story, not much detail on the analytics stuff aside from telling us they use it.


#696

I am completely on board with analytics, but I’m also 100% sure that there’s going to be plenty of dank equivalent snake oil salesmen out there trying to sell the club’s something.

I hope we have the good ones.


#697

Actually I’m interested in the way people think that analytics can best be used to advantage.

Is it breaking apart the gameplay and tactics to help you tweak gameplay s for head to head battles? This seems to be the “sell” from the media etc, but is it the most valuable use of analytics currently?

Is a better use for the 1 or 2 analysts on staff to support the performance and fitness guys tracking players fitness markers and managing individual programs.

Or is it better having the analysts supporting the recruitment staff in deep diving data on players for recruiting purposes.

Or is it better deep diving individual players gameplay to identify strengths and weaknesses and tracking the development of them over time to get a better handle on the strengths and weaknesses of your development programs.

Or…and on and on and on…

There are so many potential uses of data within the program, and such limited time and resources to dedicate to developing these systems, where do you decide to spend your money? Where is the bang for your buck?

And how should a diverse team of analysts look to get the best outcomes? How much “champion data” background, vs how much non sport, or health or whatever background, and why?


#698

Harding’s from an analytics background. Worked at North then Adelaide in those roles before being poached across

I thought we had RMIT sports/ex science pHd students as well? Not that they’d really get involved in on-field stuff.


#699

You’d want people with AFL/sporting knowledge, because even “objective” models have highly subjective assumptions within them and require context knowledge to ensure validity.


#700

I think the value lay in the overlap between coaching, gameplan and recruitment and using key indicators to find the signal in the very noisey game that is Aussie rules.

If we take the Hawthorn example, some ways I’ve observed they’ve maximised performance:

  • Superclear gameplan and roles allows the “one soldier in, one out philosophy”
  • clear roles makes identifying value “moneyball” style trades- i.e. Jarman Impey is a mediocre player but a perfect fit for their gameplan and cost them little in the trade department
  • premium on footskills to suit their gameplan made their recruiting easier (finding the Matthew Suckling types)

I think a flow on is that sacking coaches is a more dangerous proposition than ever (or will necessitate minimising their role). Clubs need to establish what they stand for from top to bottom and align all areas of the organisation. The “master coach” belongs to a bygone era.


#701

The biggest thing is that it removes bias, which if you have heard any commentator or fan talk about a game is huge in sport. It is essentially just another thing to consider when making the decisions though, it does not make the decisions for you - that is the key way to use analytics well. I do analytics for mining companies, and we get the same suspicion.