Analysis Interview with Binuk Kodiuwakku - AFC's player analytics expert

[baldandbroke]

The science of recruiting: The numbers v What you see.

RECRUITING is no longer just an art – it's now also a science.

Binuk Kodituwakku, who works in player analytics at Adelaide, joins Marc McGowan on this week's episode of Trends to discuss how the Crows leave no stone unturned to find the right player.

https://www.afl.com.au/news/2019-06-05/listen-how-the-crows-are-turning-recruiting-into-science

 

[lunacy]

I heard that too. A very good listen and it's one area of the club where we're leading the way.

 

[Tex200]

I know the analytics person who worked with Port on the 2017 draft.

Glad we’re catching up.

 

[Drugs Are Bad Mackay?]

SOS mixed acids and bases

 

[dude316]

SOS mixed acids and bases

 

[Truck25]

SOS mixed acids and bases

and made salty tears..

 

[dogs105]

and made salty tears..

Liked for the correct chemistry reference

 

[MRB37]

Pretty interesting, and i'd be honest in having no real idea he was working with the club, but for those of you who don't/can't listen to the podcast, here's the cliff notes.

- Position is Player analytics - List Management - Reports to Justin Reid and Hamish Ogilvie

- Builds models and evaluation techniques across all levels - U18, state league and AFL covering both recruiting from drafting and trading

- Describes his job as bringing an objectivity (numbers) to their subjective analysis (watching football). Works under Stephen McCrystal (Eastern Recruiting Manager) and Scott Degabriele (Western Recruiting Manager) who work under Hamish and are responsible the "complete recruiting portfolio

- Describes his job to challenge them on biases that could occur, as well as bring new players that they may have not have liked/looked at, but look good based on the models he's built.

- Covers scenario analysis and trade evaluations as well

- Also covers VFL scouting himself, as well as "pro-scouting"

- Wanted to include scouting in his work as part of upskilling, thinks needs both skills, objective and subjective, to be good at his job. When he was interviewing with clubs he wanted the live scouting as part of his job, which Reid and Ogilvie also wanted him to do.

- Describes Reid and Ogilvie as open to his input and valuing what he brings in terms, understanding it's an area they are not experts in, but recognising the value it can bring to recruitment, even if it disagrees with their own opinions.

- Says there has been a lot of buy-in from the Club, from Andrew Fagan down to the coaches as well as Brett Burton.

- Still defers to Ogilvie for final decision even if the numbers say different to his opinion.

- Said simplest way to to describe it is covering all bases.

- Says you can't rely on just one for success (scouting, no analytics or no scouting and all analytics) and you need to merge both to have the best chance of success.

- Studied law at University, but wanted to work in sports

- Interned with player agencies during Uni to get his foot in the door on the sports side of things.

- Researched what was happening in the US as they are always at the head of sports development and noticed the increase in sports analytics in the NBA.

- Then recognised it as a "gap in the market" in Australia and then focused his efforts on learning statistical models and programming.

- Moved to the US when he finished law school, worked for an NFL agent, then at a startup basketball analytics company (Vantage Sports). Wrote analysis that they'd then sell to NBA teams. Says almost every NBA team now uses their software.

- Then got a job with the Washington Wizards in the NBA on the business side as an Account Executive. Says the Wizards were behind the curve on analytics, so it showed him where they could fit in with a team.

- Moved back to Australia, but found reluctance from teams to employ analytics in full-time roles, so worked for SportsBet and then PointsBet.

- Contacted Brad Lloyd who is the List Manager for Fremantle who employed him as a consultant.

-Worked with Nick Maxwell on his own time, who assisted him in developing his models to work with football and how to communicate them in football terms.

- Describes "modelling", using R programming, describes it as "Excel on steroids".

- Describes a week in his job

- Monday starts with recruiting meeting, covers the weekends scouting across the different leagues

- He then builds/updates his models for the week to cover how players are progressing.

- Sometimes goes with Reid to meetings with player managers, works with contract evaluations at times. Says this can be harder in the AFL compared to say the NBA because salaries are not made public.

- Is given a lot of free space to learn "best practices" across other leagues in learning new methods to apply to different aspects of analytics.

- Also watches a lot of footage to suggest/cross-off potential recruits

- Says the environment has been very collaborative, people feel free to challenge, but are accepting of different opinions and feedback.

- Still talks with NBA teams in the US regarding analytics, went with Justin Reid in March to Boston to an analytics conference, met with different teams and then went to Philly and NYC to meet professional teams, some ahead of the curve, others behind. Says Philly 76ers in the NBA were one of the leading teams in that space.

- Describes some teams in the US as potentially being a decade behind others in the analytics area. Says if you don't "innovate" then you can be left behind very quickly. Says one team blatantly told him they don't use, or value analytics. Still says they were valuable to talk to as you can use their drafting/recruiting to backup how valuable analytics can be. Won't name the team, but says they have a very poor record in both drafting and recruiting.

- Says teams that don't use analytics usually have people running the team who come from "old school backgrounds" and feel threatened by it, but don't understand it's a part, but not the entire process, that goes into recruiting/trading. May have used it briefly, but didn't bring instant results, so discarded the entire concept entirely.

- Says 76ers are light-years ahead of any AFL team, but are also very ahead of most NBA teams in terms of analytics.

- Says NFL have great combination of analytics and coaching, Bill Belichek from the Patriots is big on it, but tried to hide it for years. Patriots hired a team from MIT to run their department.

- European Soccer also have big differences. Liverpool have 6 PHD's on staff running their department. Says Manchester United don't use analytics at all, says since 2013 they've been a mess. Says United have spent the same amount of money on players that Manchester City, but results are hugely different between the two sides.

- Talked with "21 Club" which is a soccer analytics business in the UK who trad to sell their product to Manchester United and others, who didn't want to talk to them at all, believing that their past success means they don't need to look at analytics.

- Says there is still growth in analytics in football, but stresses the need still for the subjective football people to be involved.

- Describes the live trading on Draft night, said they did a lot of prep and scenario analysis before the draft on live trading. Was in charge of covering picks, in terms of who had what picks to trade, scenario analysis so every pick was covered. Reid had one phone, Binuk had another to talk to teams if Reid was on the other, he'd relay information back to the team.

- Talked about how to rate if a pick is "successful". Says they have a draft value index metric, which compares picks compared against previous years. Eg Pick 8 from 2018 vs Pick 8 in previous years. Then compare how they pick to picks after, 9,10,11,12 etc. Says you don't use earlier picks to compare since you didn't have access to them. Also talks about how sometimes even if you had pick 8, it's hard to compare them to later picks since they wouldn't have been available for various reasons (i'm assuming he's talking about the "I don't want to leave Victoria" type kids).

- Says Richmond had done an excellent job with their drafting, using Alex Rance as an example. Says the majority of their Premiership team was built across 2005-2011, showing how you can't expect instant results.

 

[GreyCrow]

Bookmarked

 

[Footypie32]

Fascinating.

 

[Doc_]

Thanks for that.

 

[skittlebrew]

Appreciate the write up OP

 

[1990crow]

I don’t understand what value he is adding though. Like what decisions would get changed from his input.

 

[tinman]

This is really interesting and is an area where we can get well ahead of other clubs. All drafts aren't equal, and also even in a "weak" draft there can be standout players. Anything that assists us in becoming more professional and assisting our decision making has got to be of benefit.

What's interesting would be to look back and see how players we took were rated at the time, and how they've gone since, to see if it supports the decisions or we would have selected differently. I'm thinking players like Doedee and even Dangerfield.

I'm also thinking that you don't have to have much expertise in data analytics to see that the live pick swaps we did last draft we smashed it out of the park. If the swaps were run by Binuk before we agreed, I'll bet he was laughing his head off.

The club's preparation for live pick trading was A1, at least judging by the results. And this can now set us up for years.

I just have one question for Binuk though.

Rowell or Anderson?

 

[GreyCrow]

I don’t understand what value he is adding though. Like what decisions would get changed from his input.

As an example, and I have no way of knowing, if Don asks for pace then he may go through all his stats and suggest Gooch is quicker than Berry

 

[MRB37]

I don’t understand what value he is adding though. Like what decisions would get changed from his input.

It's not necessarily about changing opinions or decisions, but about producing data that can be used to make better decisions.

The data and statistics that football clubs have available to them is massive. Only a fraction of them get made available to the public, I've previously had access to a wider array of the Champion Data statistics (not all, but more than get put up on the internet), it's crazy, they track everything you can think of.

 

[latinoheat]

Great summation MRB37

 

[70lethal]

Things like if we bid on Thomas how does that change the rest of the domino's.
If we don't bid on Briggs what's it worth to GWS.
Where are Carlton likely to finish in 2019 to evaluate the pick swap potential.

 

[tinman]

I don’t understand what value he is adding though. Like what decisions would get changed from his input.

It might stop a club over rating a player based on a 5 minute video taken on a very windy day, where they get excited and rate the player as the 6th best player in the draft, despite no-one else seeing it. That club then swaps out their number one draft selection the next year just to grab the player they appear to have over-rated.

And yeah, we aren't Carlton, but let's stay ahead of them. Binuk has stressed that subjective football people to be involved. Any additional information that helps support the decision making and stop "howlers" helps.

Even if it is just helping to identify players that weren't on the club's radar, that adds value.

There are also so many variables because draftees come from so many different places. How does SANFL reserves compare with the Vic U18 comp vs National Champs vs WAFL U18? What if a player is injured & out of the national champs, or injury impacts their performance? How are they rated?

The football people still make the decisions, it's about being as well informed as possible.

Most club recruiters place a massive importance on the U18 national champs as that's where these players go head to head so they can be compared directly. But the better recruiting clubs have been watching players for years and it doesn't all come down to these few games. If the analytics helps identify talent and allow for a comparison which takes the many variables into account, it's a yes from me.

 

[tinman]

As an example, and I have no way of knowing, if Don asks for pace then he may go through all his stats and suggest Gooch is quicker than Berry

And this raises a very interesting question. How much weight is placed on AFL Draft Combine results?

 

[MRB37]

It might stop a club over rating a player based on a 5 minute video taken on a very windy day, where they get excited and rate the player as the 6th best player in the draft, despite no-one else seeing it. That club then swaps out their number one draft selection the next year just to grab the player they appear to have over-rated.

And that's where he talks about trying to remove the "biases" from football watching. We all have our favourite players, or guys we don't rate as highly, or not at all, so even though we all probably watch a lot of football, we have biases that can influence our thinking, either positively or negatively. So part of his job is to provide hard evidence on that player, whoever that may be, so we can try to remove that bias and make an objective decision. Head rather than heart stuff I guess.

 

[Pea Nut]

I'm gonna guess that he is the superstar from the US that was spoken about on the Foxtel telecast as working for us on draft night. It's good to put a name to him, and to also know that he is actually working for us in a full time role. To now know the full details of his role is fantastic. Have to say it's exciting that we are so progressive as a club in the recruiting and drafting space. Such big gains to be made. Makes me have a bit more faith in Fagan's quote that he wants to put the best people available in to the most important seats at the club. Maybe it's just going to take a lot longer than you would might expect. Still plenty of room for improvement in relation to some of the staff at the club.

Oh, and I'm going to go out on a limb and guess that Carlton are one of those clubs that are old fashioned in their drafting and recruiting strategies.

 

[Grenville]

and made salty tears..

1 like just wasnt enough..

 

[70lethal]

Calculating what sort of player they would get for McGovern would have been very interesting.
He would have been running the numbers of pick 8, 13 and 16 for pick 1 before the draft.
Jones, McHenry and ?? vs Rankine or Lukosious.
I wonder how and if he's evaluating which players to trade and their value.

 

[Pea Nut]

Calculating what sort of player they would get for McGovern would have been very interesting.
He would have been running the numbers of pick 8, 13 and 16 for pick 1 before the draft.
Jones, McHenry and ?? vs Rankine or Lukosious.
I wonder how and if he's evaluating which players to trade and their value.

I think absolutely they would be. The analytics would extend right across the recruiting, drafting and trading. And the trading component of it would be easier as the analytics is working with AFL stats, rather than underage or second tier league stats.