Rugby League World Cup 2021 - Semifinal Experience and Wrap Up of the England team data viz project

I would like to say this has not been delayed because I have been sent into the pit of despair by England, again. 

I would like to say it wasn't that, and the delay is partly because real life got real in a variety of ways, but England have done it to me again. 

I was in the crowd at Wembley when New Zealand beat England in the last 30 seconds of the semifinal in 2013. 

I have now been in the crowd as they lost a semifinal in golden point extra time. 

Because a Samoan who had never scored a drop goal before scored one. 

I start to feel I might be a jinx. 

The only saving grace is that England did it to better people than me too.  

Photo of Jamie Peacock and Willie Poching looking nervous courtesy of L - I was too busy worrying. 

So much of what England did right in the group stages and the quarterfinal, they just didn't do in the semifinal. Part of that was their own mistakes (particularly trying to force passes that just weren't on) but partly it was Samoa dominating the pitch. 

Field control is as much a thing in rugby league as ring and octagon control is in fighting, and Samoa controlled the pitch magnificently. Compare where on the pitch England had to kick on the 5th versus where Samoa were doing it. 

It was clear in the stands. 

Also clear in the stands was the gap in the right side of England's defence between Dominic Young and Kallum Watkins - and if I could see it, Matt Parish could too, and that's where Samoa kept attacking. 

My pre-match guess, based on the data, for the starting 13 was "Williams, Welsby, Young, Makinson, Burgess, Hill, Bateman, Farnworth, Radley, Whitehead, Watkins, Tomkins and McIlorum. 13 out of 13 guessed correctly, which is a nice confirmation that the methodology is showing something real. 

England were involved in 87 point-scoring moments.

Number of point-scoring moments per player:    

18 of the 24 players in the squad scored in one game or another.

When do England score?    

Mostly minute 20-29, then 45-50 then 72-78.

England point-scoring moments by time and player:    

Number of point-scoring moments that England players were on the pitch for: 

How low Luke Thompson is on this list is fascinating. Not sure if he was just unlucky to be used against fresher oppositions, or with a less creative back row, or what, but it might suggest Shaun Wane shouldn't have swapped Thompson in for Lees or Batchelor. On the other hand, the data may be skewed by the number of points scored versus Greece (about which, more in the future work section), and I can see why you'd want a more experience player in the semifinal. 

The "when are players on the pitch" view didn't quite work out, because Wane sometimes started players and sometimes used them as subs. The one player where anything did show up was Morgan Knowles, whose use as an impact player is really visible.

The matrix now looks like this:

The network diagram looks like this:

 

England conceded in 20 point-scoring moments. Unfortunately mostly in the second match versus Samoa. There being a second match against Samoa caused a small amount of difficult in this set of  visualisations - I have gone with calling the team in the semifinal "Samoa2".

Points-scoring moments conceded by England:

  Samoa's improvement, possibly an advert for warm-up games.

When do England concede points?    

That 83rd minute point score is Stephen Crichton, breaking English hearts. I have no idea how he didn't win man of the match even before that, but it definitely showed why awarding MOTM before full time is a nonsense. (Also, he's only 22. If Samoa can keep most of that team together!!! Next time!!!)

Point-scoring moments against England by time and team:

   

Number of opposition point-scoring moments England players were on the pitch for:

 

Not normalised for number of games played, or minutes played, but Chris Hill is remarkably low in the list. He's probably my surprise of the tournament - I'm used to him being on Warrington teams that didn't quite win things or Leeds teams that didn't quite win things, but in this World Cup he's been a remarkable solidifying point. It's noticeable that him and Radley were both on when England picked up in the last 20 minutes of the semifinal. This is also seen in the equivalent matrix.  The relatively low points scored against means that, I think, the network diagram is unfair. Discussion/Future Work: This did what it was intended to - from the data, I was able to predict Wane's starting 13 (okay, data and knowing that there's no way Wane wouldn't pick McIlorum). I think the rolling subs do make it easier than the "once you're subbed off, you're off" rules of rugby union. Because I included tries and conversions (because I had to include penalties for the rugby union) I probably overweighted the games where there were lots of points (e.g. 94-4 vs Greece) so in future, I might just cover tries and drop goals.)

Rugby League World Cup Data Visualisation Project - England point-scoring moments after 4 games

In the previous post, I made a data-driven prediction of who England's quarterfinal team would be. I went with Ackers, Bateman, Burgess, Hall, Radley, Knowles, Pearce-Paul, Sneyd, Tomkins, Welsby, Whitehead, Williams and Young. 

I got 8/13 right - I'd forgotten how fond Shaun Wane is of Michael McIlorum, hadn't expected him to choose Tommy Makinson over Sneyd and Hall and I completely forgot about Herbie Farnsworth - my apologies to Farnsworth. 

I will update my prediction, following an extra match's data, later.

When do England score?

    

It's more when don't they score? 

There's a concentration of point-scoring moments 21-27 minutes, 48-49 minutes and then 75-79 minutes.

Who scores for England?

   

No new players are on that list, but that's still 17 out of a squad of 24 scoring.

Below is the England point-scoring moments by player and time   

How many point-scoring moments were different players on the pitch for? 

The network diagram now looks like this:  And this is the matrix now: There's been a major shift, and I think it's because a lot of the team that scored so many against Greece weren't in the quarterfinal match. 

From how the figure looks now, there's still that central trio of Williams, Welsby and Young, then a second grouping of Makinson, Burgess, Hill, Bateman, Farnworth, Radley, Whitehead, Watkins and Tomkins. 

Which is 12 players. 

The 13th player is the hooker and it's clear that Wane will pick McIlorum, even though the data suggests to go with Ackers. 

The other interesting point, from watching the game, and it's interesting how watching and analysing at the same time changes the way you watch the game, but from that rather than anything that shows up in the data visualisations - England are so much better when Radley and Hill are on. 

Which wasn't a sentence I was expecting to write. It's not just scoring that dropped off without them, there was also an increase in handling errors. 

Looking at when England concede: 

 

I am still not sure how to cope with France being the opponent that scored the most points against England when they've played Samoa and Papua New Guinea. 

It does seem to be the second half of the halves when opposition scores.

Number of opposition point-scoring moments players were on the pitch for

 

This is really skewed by the low number of points scored versus number of games, as is the matrix view of who is on the pitch at the same time when the opposition score.  Tom Burgess and Michael McIlorum are probably lighter on the figure than expected, and it's possibly that defensive ability that is the reason why McIlorum is being picked over Ackers. 

The equivalent post for the semifinals will be delayed because yours truly will be at the Emirates watching it.  

The only positive to Tonga losing is that it means I can now wholeheartedly cheer for England (I find it difficult to cheer against a team with Kristian Woolf and Konrad Hurrell).

L will be acting as crowd control. 

Rugby League World Cup Data Visualisation Project - England point-scoring moments and point-conceding moments after 3 games

Once again, I'd like to that nrl.com for having the data in a really easy-to-read format. For this game, the link is here - https://www.nrl.com/draw/rugby-league-world-cup/2022/round-3/england-v-greece/

There were 64 scoring moments for England.

When did England score?    

There are a couple of minutes where there are a higher concentration of points. They are at 5, 12, 23-26, 39, 49, 62, 66-67, 74, 76 and 79 minutes.

Who scored for England?

   

17 different players have scored for England. The only England players who haven't scored are Michael McIlorum, Mikolaj Oledzki, John Bateman (!), Chris Hill, Mike Cooper, Morgan Knowles and Sam Tomkins of all people. What's going on there then?

England point scoring moments by time and player:

 

Number of England point-scoring moments players were on the pitch for:

Dominic Young and George Williams were on the pitch for every England scoring moment (all 64). Oledzki was on for the least of any English players (6). 

If we look at the players who were on the same time with each other when points were scored, looking at both the matrix and the network diagrams, Oledzki has played himself out. He is the player with the least interactions in the matrix diagram and the only England player not present on the network diagram. From the matrix, you can see a clear core (Young, Williams, Welsby), with two groups of possible teammates. 

The lower group of teammates are slightly darker, suggesting more frequently co-occurring when England score. If Wane goes with that, it's a team of Young, Williams, Welsby, Sneyd, Tom Burgess, Tommy Makinson, Ryan Hall, Andy Ackers, Joe Batchelor, Kai Pearce-Paul, Morgan Knowles, Chris Hill and John Bateman as your starting 13. Which is ... not a terrible idea. 

It's obviously skewed by the sheer number of points England scored against Greece, a game Sam Tomkins missed, but I suspect that 13, with Tomkins instead of Makinson, Radley instead of Hill (providing he doesn't have an elbow injury) and Elliot Whitehead instead of Joe Batchelor might be the starting 13 for the quarterfinal. 

If we look at the points England conceded:  

 I still cannot reconcile myself to the French game having been tighter than the Samoa game for England. It makes no sense. 

The Greek game put paid to my theory that England concede more in the 30-40th minute.  

As to who is on the pitch when England concede:  

It's slightly unfair because Williams and Young were on for more than anyone else because they played 80 minutes in all 3 games. 

The interesting and unexpected is that Chris Hill wasn't on the pitch for any concession of points, the only player where that is the case. Partly, I suspect it's because he's only played limited minutes, but it suggests there's life in the old dog yet.  

I think England only conceding 9 point-scoring moments is skewing this, just because most were in the France match so the players from that match are strongly highlighted.

Rugby League World Cup Data Visualisation Project - England point-scoring moments after 2 games

After the second game, when do England score their points?

 

The points are reasonably spread through time - the most interesting thing is probably that this is two matches in a row where England have scored in the fifth minute, which suggests that Wane's philosophy may well be hit 'em hard, early. 

If we look at who scored the points, Tommy Makinson still tops the scoring-moments chart, even though he didn't play the second game. As a rugby league fan, it feels very odd saying an England game vs France was harder, with fewer scoring opportunities, than a game against Samoa. 

If we look at who scored when, Yes, that is Ryan Hall in the top 6 points scores despite only playing one game. Because if England give him the ball he will score. If only Wayne Bennett had remembered that during the last World Cup.

Number of point-scoring moments England players have been on the pitch for: Four players have been on the pitch for all of England's points - Herbie Farnworth, George Williams, Elliott Whitehead and Dominic Young. 

Other than enjoying pulling L's tail about George Williams coming up good, one of those four players is not like the others. 

Elliott Whitehead is a second rower who has now played two 80 minute matches back to back. Given how much of the noise before the tournament was about his age, I think large meals of humble pie are required for many people. 

The network diagram and matrix figures have changed, but I think the major differences reflect the personnel changes between the two matches (Ackers, Oledzki, Bateman, Sneyd and Hall in, Makinson, McMeeken, Knowles, Lees and Cooper out) rather than either set of players being better or worse at creating chances. 

  I've also now got some more info on when the opposition have scored. It's not as well formed as the scoring information so I'm only going to share one picture at the moment. The main thing I take away from this is that England concede in the last 10 minutes of the first half. 

Now, due to Samoa's implosion, that wasn't a problem against them, and again England had a decent head start against France, but later on in the tournament, that could really come back to bite them in the bum. The teams England are most likely to face in the quarter and semifinals are both momentum teams and a score then will really give them heart and might well give them the extra oomph to score another before halftime. 

What Australia might do in that situation ... it's too horrible to contemplate. 

The match against Greece is unlikely to provide final proof either way on this one, but if Greece do score in those last 10 minutes of the first half, then I think that'll prove the pattern and demonstrate a really weakness that can be exploited by England's opponents.

Rugby League World Cup Data Visualisation Project - What Combinations Score For England

This is a follow up to the Six Nations 2021 project.  In the summary for that I said, "It would be interesting to see how things like this look in sports with rolling subs (odd that there's a Rugby League World Cup this year that might fill that gap ;) )". 

It was a good plan, it was just that the Rugby League World Cup was delayed for a year. 

But it's here now.  

Since what I am interested in is whether having rolling substitutes affects the shape of the patterns, I will focus on a single team. I have, for reasons of obvious bias, chosen England. 

England's first match was versus Samoa, and I was expecting it to be close and tight. There was always the chance that it was going to be a blow out victory for England against a Samoan team who hadn't had any warm-up games. But I don't think even the most optimistic England fan (or pessimistic Samoan fan) was expecting a score of 60-6 for England (match report here - https://www.bbc.co.uk/sport/rugby-league/63271368). 

Okay, so that should have given me plenty of data to work with. Shame that the official RLWC site didn't have any stats, nor did the BBC. 

Luckily for me, NRL.com did have the info. 

For the first game, all the information came from here - https://www.nrl.com/draw/rugby-league-world-cup/2022/round-1/england-v-toa-samoa/

When do England score?   

 

The points come in waves, basically at 20 minutes, 45 and then 60-80 when Samoa ran out of energy, players because three of them had either broken bones, dislocated hips or injured ligaments I've never even heard of and were a man down due to Anthony Milford being sin-binned for a late tackle.

Who scores for England?    

Tommy Makinson, so England can join Saints fans in the traditional pastime of hoping he doesn't get injured.

Who was on the pitch when England scored?   

 

And then same but as percentages: 

 

I'm going to watch how this changes, and, at the end of tournament correct for percentage of games played. At the moment it mostly shows which forwards were being rotated out (about which, more next).

Network diagram and matrix diagram of which players were on the pitch together when England scored     

After 1 game, the network diagram doesn't tell us much, but the matrix diagram interests me. 

Okay, so forwards play together less than the backs, that's expected - forward are the players you sub on and off to keep them fresh. 

What I didn't expect was the pattern: 

Morgan Knowles (LF, SR) and Matty Lees (Prop) played together infrequently, which is odd since they play different positions. 

Matty Lees (Prop) and Chris Hill (Prop) not playing together makes more sense because they're both props. 

Michael McIlorum (hooker) and Chris Hill (Prop) and Tom Burgess (Prop) not playing together is odd, because, again, not the same position 

Mike Cooper (Prop, LF) and Chris Hill (Prop) and Thomas Burgess (Prop) and Chris Hill (Prop) and Tom Burgess (Prop) not playing together, again, makes sense because they'd be prop for prop replacements. 

The two "proper" forwards that played the full 80 minutes were Mike McMeeken (second row) and Elliot Whitehead (also second row). It's both interesting, given the pre-tournament chat about Whitehead's age, and not surprising because McMeeken is the second most underrated player in Super League. 

The thing that strikes me is that there was a period of the game, when McIlorum was off the pitch, when there was no recognised hooker on the pitch. In fact, had anything happened to McIlorum, there was no other recognised hooker in the matchday squad - Shaun Wane has only picked two of them, McIlorum and Andy Ackers. 

Now I'm sure that someone, probably Victor Radley, could have deputised but it's a risky tack to take. 

Obviously, I'll keep updating these as the tournament carries on. I'm also going to look at who is on the pitch when England concede. However, at the moment, it's an uninformative set of information. 

The R project is here https://github.com/fulltimesportsfan/RLWC2021/blob/a49975981f6054b70a48b7b265a86b5a6d599d61/Rugby%20League%20World%20Cup%202021%20Summary.R 

I've fixed all the random movies and casts. I think it's given me a better idea of how the code works. 

Json file if you want to do cooler things with it is here - https://github.com/fulltimesportsfan/RLWC2021/blob/35effbc084c65f3383e420f02ccfaf81524ac1b1/England1stgame.json