Wednesday, 18 November 2026

Formula 1 2026 - Sao Paulo Grand Prix

Qualifying: 
Obviously a couple of people weren't trying hard because they had ginormous engine penalties (a 75-place penalty being the worst, but there were also a couple of 30 place penalties), but it's noticeable who was higher than usual like Hulk, Bearman and Albon. 

I'm not saying that when the differences in cars are reduced by circumstances the good drivers prosper but I have a Hulkenberg bias for a reason. 

The race itself: 
Hopefully Cadillac's reliability will improve next year. 

The McLaren starting issue rears its head again. But those were excellent starts from Verstappen, Hamilton and Lawson. 

Of course, the wet weather tyres start to go off while the track's drying but there's a forecast for more rain and the risk of losing time either way increases exponentially - oh that's how the strategy teams earn their money! (As with all unpredictable systems, it turned out the weather forecast was inaccurate. By about 20 laps, which isn't that long in the grand scheme of things, but it's a long time in an F1 race.) 

I doubt Russell's spin is the first spin in a pit lane but it's certainly the only one I remember. 
Also a spinny Kimi! 

I am trying to decide if Verstappen's car radio is engaging in a mid-season strike to reduce the amount of profanity it's exposed to. 

Any swearing from Hadjar is justified. Let him keep positions sometimes, Red Bull! 

Congratulations to Mercedes on winning the constructor's title too. I can only hope Ferrari will be a stronger challenger next year.

Sunday, 15 November 2026

Rugby League World Cup 2026 - Data visualisation project update after England game 3

You know the advice to visualise your data, that will help you understand. It doesn't help with this project. 

I'm okay with data that makes no sense because of external factors, for instance the equivalent diagrams for the England Rugby Union team at the 2023 World Cup, where Borthwick had to play George Ford for the first 3 games because Owen Farrell was banned, and then started to play Farrell the minute he could, which meant that the visualisations looked really messy in games 4, 5 and 6. 

But there's nothing like this. 

Now partly, it might be that there will only ever be 3 games worth of data for the England men's rugby league team at the 2026 World Cup and less data is always more confusing, because the signal to noise is lower. 

But I can't shake the feeling it's this messy because Brian McDermott didn't know who his ideal matchday 17 should be. 

After 3 games, when do England score? Bar chart of when England score.  Three minutes have 3 point scoring moments, minutes 39, 63 and 68. 

No major patterns. 

The "when do England concede" diagram is even less revealing because the opposition in one of the games (France) didn't even score. Bar chart of when England concede.  No minute has more than one point-conceding moment.  It has been coloured as well.  There are 11 turquoise bars for Tonga scoring against England, and 4 peach bars for Papua New Guinea scoring against England.  The interesting thing is the pattern of the bars.  All 4 peach bars are between minutes 44 and 51, none of the turquoise bars are then. There is no overlap of point-conceding minutes. It just looks intriguing. 

Who scores for England? Bar chart of who scores for England.  Savala, the kicker, has the most with 19 point-scoring moments.  He is followed by Dom Young with 5. 

12 different players scored for England. 

Savala, who was the kicker for all but the final 15 minutes of the Papua New Guinea game, is understandably way ahead of everyone else on 19 point-scoring moments. Next highest is Dom Young on 5. 

Which players are on the pitch when England score? Bar chart of who is on the pitch when England score.  4 players were on the pitch for every England point-scoring moment, Pearce-Paul, Smith, Young and Brimson.  They are followed by Farnworth, present for all but two moments.  There is a slow decline in number of point-scoring moments present for, then a drop to Walters at 31, followed by King and Williams on 30.  It drops again to Luke Thompson on 25, then Radley on 21, Lewis on 15.  Kallum Watkins is at the bottom on 3. 
Which players are on the pitch when England concede? Bar chart of who is on the pitch when England concede.  7 players were on the pitch for all 15 England point-conceding moment, Pearce-Paul, Savala, Farnworth, Smith, Williams, Young and Brimson.  Next comes Walters and Nsemba at 12.  Ben Currie is bottom on 2. Comparing the position on the "present when England score chart" vs the "present when England concede chart", is screwed up by there being 21 players on the "score" chart and 19 players on the "concede" chart, but the major differences are: 
Williams is 10 places higher on the concede chart. 

Eckersley is 9 places higher (but I think that's because he played the game vs Tonga but not the other two games) 

Havard and O'Neill are 6 places higher 

Ashton is 10 places lower on the concede chart, but I think that's because he didn't play the Tonga game. 

Thompson is 5 places lower on the concede chart. 

So is Litten, and that's interesting because he played in all three games. I told you he was good. 

The mix and matchiness of England's team and combinations, and the outsize effect of the France game on both the points-scored and points-conceded moments, mean the matrix and network graphs are a little unusual. 

I'm going to start with the points-scored diagrams, because the lack of point-conceding moments in the France game has had some interesting effects. 

The point-scoring moments matrix diagram and network diagram: Matrix chart.  Darker areas = more often together when England score.  Paler = less often.  The really interesting thing is that there is a very dark often together section in the middle (containing Pearce-Paul, Smith, Young, Brimson, Farnworth and Savala), followed by a paler section bottom right.  Moving up and left we have the expected paler section but at the very top and leftermost is another darker section, with Radley and Thompson, possibly suggesting that they are together often when England score but they're not always with that most-often-together cluster. 

The odd Radley and Thompson dark zone at the top left is possibly an artefact of Radley not playing against Papua New Guinea. There are 18 players on this network graph, which is good because a match day squad is 18.  The central section is Nsemba, Thompson, Savala, Farnworth, Young, Litten, Williams, Radley, Walters, Smith, Brimson, Pearce-Paul, Ashton and King.  Sticking out top right is Welsby, bottom right is Lewis, and Knowles is bottom left. This makes sense because Welsby and Lewis didn't play vs Tonga and Knowles only played against Papua New Guinea. 

I love Morgan Knowles an entirely unreasonable amount, but I don't think players who aren't fit enough to be available for every game. (I mean, he was awesome in the game he did play in and if England had had him against Tonga how might that have gone etc etc.) 

The "point-conceding" graphs intrigue me because, due to France not scoring, we can directly compare the vs Tonga and vs Papua New Guinea squads. In an "ideal" world, there would be one very dark area of "play together most often", followed by increasingly paler areas. That does almost happen.

  <3 a="" against="" almost="" amount="" an="" and="" area="" areas.="" aren="" available="" awesome="" be="" because="" but="" by="" can="" compare="" dark="" did="" directly="" does="" don="" due="" england="" enough="" entirely="" etc.="" etc="" every="" fit="" followed="" for="" france="" game.="" game="" gone="" graphs="" guinea="" had="" happen.="" have="" he="" him="" how="" href="https://postimg.cc/2LDLXxkt" i="" ideal="" if="" in="" increasingly="" intrigue="" knowles="" me="" mean="" might="" morgan="" most="" new="" not="" of="" often="" one="" paler="" papua="" play="" players="" point-conceding="" scoring="" squads.="" t="" target="_blank" that="" the="" there="" think="" to="" together="" tonga="" unreasonable="" very="" vs="" was="" we="" who="" world="" would="">Matrix of England players who play together when England concede.  There is a dark area in the bottom right, and it gets steadily paler as you move up and leftwards.  Fuller details below. The part where it doesn't happen is interesting. Jez Litten's row is darker than Max King who lies between him and the darker areas. I think that's because Litten played more vs Papua New Guinea where England scored more, while King played in both games, even if he was subbed on/off more. The network graph also makes sense. 13 players in the network graph.  The core is Farnworth, Smith, Pearce-Paul, Brimson, Young and Williams.  Around them, clockwise starting at 12, are Walters, Nsemba, ONeill, Radley, Eckersley, Savala and Havard. Three things stand out: 

1 - There are only 13 people in the diagram. A matchday squad is 17 + 1 emergency replacement. I don't think that McDermott knew what the 17 were. 

2 - Eckersley is on it despite him only playing one game. I fear he was used as the scapegoat. 

3 - Savala isn't right in the centre despite scoring most of England's points. He only missed 3 of England's point scoring moments but that was enough to shove him slightly out. 

I'd also like to thank nrl.com for all the data. Their website is significantly more useful than the official one.

Saturday, 14 November 2026

Rugby League World Cup 2026 - Men's and Women's final network diagrams

I'm posting them together mostly for timing reasons as I'm away for part of the weekend. 

First, a word on the Wheelchair Rugby League World Cup. I really wanted to make similar diagrams for the Wheelchair World Cup and it's so nearly there. 

This time, full squad information was available but club team information wasn't, not even when I checked on nrl.com, whose pages are more useful that the official website. This is a move forward, last World Cup, full squad information wasn't available. Three World Cups ago (so 2017), that was the position the women's event was in, and now the information is easily available. Hopefully in 2030 I can make the diagrams for the wheelchair event as well. 

The Women's World Cup final: 

Australia vs New Zealand but at least they were in different groups in the group stage. Unlabelled network diagram of the connections between the teams in the final.  Larger and redder circles are teams with more degrees of connection, darker and smaller circles have fewer.  There are two large red circles, some smaller mid-grey ones and many dark grey ones. Same diagram but labelled.  The two large red circles are in the two diagonals.  New Zealand are in the top left, Australia bottom right. Brisbane Broncos are the club team closest to the centre, but they're not the team with the most representatives. That would be the Sydney Roosters, who have 9 representatives, all playing for Australia. 

The Broncos are one of four clubs with five representatives each. They are the Broncos, Wests Tigers, Cronulla-Sutherland Sharks and NZ Warriors. Interestingly, all except for NZ Warriors have players representing both national teams. 

The Men's Final: 

One thing that interests me is that, given it's the same national teams, and the same squad sizes, the shape of the diagrams is different. Where the women's diagram is on a diagonal line, the men's is on a horizontal line, but again New Zealand on the left and Australia on the right. Unlabelled network diagram of the connections between the teams in the final.  Larger and darker circles are teams with more degrees of connection, redder and smaller circles have fewer.  There are two large black circles, some smaller mid-grey ones and many red ones. Same diagram but labelled.  The two large black circles are almost in a horizontal line.  New Zealand on the left and Australia on the right You'll notice I deliberately inverted the colours. I think the black/dark grey for small and red for bigger colourway used for the women's works slightly better. 

Penrith Panthers are the central club team, and are linked to both national teams, although they have slightly more links to New Zealand. 

Penrith are also the club team with the most representatives left, with 9. They are shared between both national teams. The club team with the next highest number of representatives left are New Zealand Warriors with 7, all playing for New Zealand, which again reflects the more diffuse club representation in the Australian squad. 

We started with New Zealand vs Australia, and we're going to end with it, so at least there is symmetry.  Hopefully New Zealand will keep Kalyn Ponga for the full match. 

If it's even half as good as that opening match, it's going to be a cracker.

Wednesday, 11 November 2026

Formula 1 2026 - Mexico City Grand Prix

Yes, I know I ought to be posting something more relevant, like the Rugby League World Cup final diagrams, but they are in the works. The data visualisation project is slightly stuck. I'm currently at the "looking at diagrams, screaming that nothing makes sense" stage. 

Then again, that also covers most of my Mexico City Grand Prix commentary. 

I'm used to Ferrari snatching defeat from the jaws of victory, but crashing into each other and then having a strategy oops is a bit much, even for us. My theory is that the drivers know the car isn't as good as the Mercedes, and because they know the strategy team can't get them out of mistakes, they put too much pressure on themselves and it leads to more mistakes. 

The circuit is still not racing-friendly, with most of the interest coming from whether the hybrid engines produced by this set of regulations could cope with the heat and the altitude. 

Very glad to see Bottas was alright. On the radio, that crash sounded nasty. 

But no matter the other issues with this race, this is its (and possibly the season's) defining image: Max Verstappen on the left and Lewis Hamilton on the right lifting Kimi Antonelli onto their shoulders after he wins the Formula 1 Drivers World Championship at the Mexico City Grand Prix.  It is adorable. 

Is that not just the loveliest thing?

Sunday, 8 November 2026

Formula 1 2026 - United States Grand Prix

Delayed, because I had notes, but forgot to post them. 

First, a complaint. I have no idea who US TV have doing the interviews, but I think Mercedes nailing the new regulations has more to do with Antonelli's results than any mindset change from Kimi from last year to this. 

I would also like to have a word with whichever arranger stuffed up the poor lady singing the anthem. There is no key change there, sir. 

On to the race itself: 

Ferrari, how dare you have the worst engine. There are traditions! Ours is strong engines. Occasionally exploding engines, but strong, powerful engines. 

Ferrari tried clever things with strategy. This worries me more than when Red Bull do it. 

If Russell didn't have bad luck, he'd have no luck. 

Did Red Bull not tell Hadjar that he is race driver number 2? (I believe this is the usual time for the traditional song from Muppets: Most Wanted - I'm Number One and You're Number 2) 

I would like to thank Leclerc for bringing some racing! Normally COTA brings much more entertainment than this race did. (Also, yes BBC radio, the Ferrari fans are numerous enough to be heard :) )

Saturday, 7 November 2026

Are the scrutineers picking on my driver? - Update after the Azerbaijan Grand Prix

The major update, in terms of this project, was Hadjar coming back after his wrist injury. The actual race scrutineering chart is very straightforward with all the drivers being very near the expected value of 38 tests. Bar chart showing how often cars were scrutinized compared to the expected number at the Azerbaijan Grand Prix.  The values are displayed as a standardised residual.  Cars 3 (Verstappen), 12 (Antonelli) and 63 (Russell) were the most over-scrutinized compared to expected, with a value of 0.32.  Cars 1 (Norris), 14 (Alonso), 18 (Stroll), 43 (Colapinto) and 77 (Bottas) were the least scrutinised compared to expected with a value of -0.32.  7 cars were tested exactly as much as expected. This can be seen by how far away those values are from the dotted lines that indicate significance. 

It makes sense that Norris, Alonso, Stroll, Colapinto and Bottas were tested fewer times than expected because they were amongst the drivers whose cars didn't finish the race (race report here). 

Interesting things are happening in the season-long charts. 

For the season to date, the χ² has gone up to 4.9509, still not significant but increasing. Bar chart showing how often cars were scrutinized compared to the expected number up to the Azerbaijan Grand Prix.  The values are displayed as a standardised residual.  Car 63 (Russell) was the most over-scrutinized compared to expected, with a value of 0.63, he was followed by car 44 (Hamilton) on 0.58 and car 3 (Verstappen).  Car 18 (Stroll) is the least scrutinised compared to expected with a value of -1.03, followed by 77 (Bottas) on -0.85 and 11 (Perez) on -0.71.  No cars were tested exactly as much as expected. Russell is still the most over-scrutinized, followed by Hamilton and Verstappen. Antonelli is the 5th most scrutinized (although I know that's just luck of the draw). 

The bottom section is all the unreliable cars (Aston Martin and Cadillac) but Hadjar is remarkably low, coming just above those four. And that's not a mistake in the coding, I've checked. The code has considered the three races where Hadjar was replaced by Tsunoda. 

I'll see if this pattern continues in the next few races.

Friday, 6 November 2026

Rugby League World Cup 2026 - Women's Semifinal Network Diagrams

Sorry for the delay on posting the equivalent diagram for the Women's Rugby League World Cup 2026. Unlabelled network diagram of the players and countries in the semifinal of the women's rugby league world cup.  The large red circles are the countries, the smaller black circles are the teams and players.  There are three teams in a sort of triangle-ish, with the fourth team sticking out making a larger triangle. Labelled network diagram of the players and countries in the semifinal of the women's rugby league world cup.  The large red circles are the countries, the smaller black circles are the teams and players.  There are three teams in a sort of triangle-ish, with the fourth team sticking out making a larger triangle.  The smaller triangle is has Samoa and Australia in the bottom corners left and right, then New Zealand at the apex.  Fiji are the sticking out team.  The larger triangle has New Zealand in the middle. 

The interesting thing is how much less symmetrical the women's semi-final diagram is. There's a definite New Zealand, Samoa and Australia triangle with Fiji sticking out. This does not bode well for the Fijian women's team, but on this blog, hope springs eternal. 

New Zealand are the national team closest to the centre, with Brisbane Broncos being the club team closest. 

There are 6 clusters in the community view, the 4 national teams, then one cluster that includes Cronulla Sutherland Sharks and West Tigers which seems to exist because they link multiple national teams, then the St George Illawarra cluster, because St George Illawarra players link Samoa and Fiji, but only Samoa and Fiji. Same diagram, labelled with community colours.  Orange community in the bottom left corner, blue in the right, middle community is pink.  There is a sub-community in a sort of grey around the middle, then a green community linking the orange community to the olive green one at the top. Same diagram, labelled with community colours.  Orange community in the bottom left corner (Samoa), blue (Australia) in the right, middle community is pink (New Zealand).  There is a sub-community in a sort of grey around the middle (Cronulla Sutherland Sharks and Wests Tigers), then a green community (St George Illawarra) linking the orange community to the olive green one at the top (Fiji).

Monday, 2 November 2026

Rugby League World Cup 2026 - Men's Semifinal Network Diagrams

I will bemoan and bewail England at some later date, look at how geometrically symmetrical and therefore suggestive of some seriously stonking semifinals this network diagram looks. Unlabelled network diagram of the players and countries in the semifinal of the mens rugby league world cup.  The large black circles are the countries, the smaller red circles are the teams and players.  It is round and symmetrical. Network diagram of the players and countries in the semifinal of the mens rugby league world cup.  The large black circles are the countries, the smaller red circles are the teams and players.  It is round and symmetrical.  The four teams are basically north, east, south and west.  Samoa is north, Australia east, Tonga south and New Zealand west. Penrith Panthers or West Tigers are the club team closest to the centre. 

The four national teams are equidistant, which suggests a fairly even spread of players and club connections between them, and that they may be of reasonably similar strength. 

Which also suggests that Australia's narrow win over New Zealand might not have been a one off, and only raises my hopes that someone might beat Australia. 

I may be looking forward to New Zealand vs Tonga a frankly unhealthy amount but both semifinals have the potential to be so good.

Sunday, 1 November 2026

Are the scrutineers picking on my driver? - Update after the Madrid Grand Prix

We'll ignore that I've somehow forgotten to write up my US Grand Prix notes when I swore I had, and that I'm that far behind in this series that I accidentally did the maths for the Azerbaijan Grand Prix before doing this one. I will catch up, eventually. 

The Madrid Grand Prix itself: 

Like the race itself, nothing spectacular in the scrutineering data. 

The χ² for the race is 1.32, so none of the differences are statistically significant. Bar chart showing how often cars were scrutinized compared to the expected number at the Madrid Grand Prix.  The values are displayed as a standardised residual.  Car 12 (Antonelli) was the most over-scrutinized compared to expected, with a value of 0.34, followed by 6 cars at 0.17.  Cars 11 (Perez), 18 (Stroll) and 55 (Sainz) were the most under-scrutinized compared to expected with a value of -0.51.  5 cars were tested exactly as much as expected. 

The bottom 3 cars (Perez, Stroll and Sainz) are 3 out of the 4 cars that retired, which makes sense. 

This is the first time that Antonelli (car 12) has been that high up. 

The season up until the Madrid Grand Prix: 

The continuing increase in the χ² for the season-long scrutineering can best be seen in the figure with the dotted lines for significance. χ² is now 4.58 (compared to a threshold for significance of 33.92). Bar chart showing how often cars were scrutinized compared to the expected number up to the end of the Madrid Grand Prix.  The values are displayed as a standardised residual.  Car 44 (Hamilton) is the most over-scrutinized compared to expected, with a value of 0.60, followed by cars 1 (Norris) and 63 (Russell) on 0.56, then 3 (Verstappen) on 0.46.  Car 18 (Stroll) was the most under-scrutinized compared to expected with a value of -0.98, followed by car 77 (Bottas) on -0.79 and car 11 (Perez) on -0.70.  5 cars were tested exactly as much as expected.  Only car 81 (Piastri) has been tested as many time as expected. 

It really is the cars at the bottom of the pile that are pulling the numbers up, because their absolute values are so much larger than the over-scrutinized.

Wednesday, 28 October 2026

Rugby League World Cup 2026 - Data visualisation project update after game 2

Game 2 itself: 
The game highlights one of the problems England face. They can beat the French, and beat them handily (Match report here), but against the stronger teams, they just don't seem to gel. 

Like in 2023, yes, they beat a touring Tonga team (who were without several of their players), but it wasn't anything like convincing, and yes, that was under a different coach but I don't blame Wane for it. 

There needs to be more regular international rugby league to allow England to get any better (also Lebanon, France, Ireland, Scotland etc). 

McDermott appears to have come around on the topic of Jez Litten, but not quite sure why you'd suddenly swap Ashton for Eckersley after you'd trusted Eckersley against Tonga. 

Data visualisation: 

For the data visualisation, I've only done the cumulative charts. This is partly because half the game 2 charts would have been blank because France didn't score, and because the thrust of the project is the developments game to game. 

When do England score? Bar chart of when England score.  The important detail is that the first bar is at 22 minutes.  Most point-scoring moments are in the last 20 minutes.  Minute 63 has the most at the moment with 3 point-scoring moments.

Even against France, it took England 22 minutes to score. Obviously, they won the game handily, and you have to wear all oppositions down but it does suggest a certain lack of oomph. 

Who scores for England? Bar chart showing who scores for England.  Savala, the kicker, is way in the lead with 15 point-scoring moments.  He is followed by Dom Young on 4, then Farnworth on 3. 

Savala being the player who has the most point-scoring moments makes sense, because he's the kicker. Dom Young being in second place shows that giving him the ball is a good idea. Let Dom Young do the thing he's best at. 

The France game really skews the "which players are on the pitch when England score?" line chart over time, but you can see which players England use as impact subs. Line chart of which players are on the pitch when England score.  The number of tries England scored against France have moved this around.  Lewis and Welsby are being used as impact subs. 

Which players are on the pitch the most when England score? Bar chart of players present when England score.  Pearce-Paul, Savala, Smith, Young, Brimson are present for all 31 point-scoring moments.  Eckersley, ONeill and Havard are present for the least with 4. 

Pearce-Paul, Savala, Smith, Young and Brimson are obviously McDermott's chosen core, with Ashton having taken over from Eckersley on the wing. Not sure why he didn't play Havard against France, but England do have sufficient forwards that it's not that weird. 

The who plays together when England score matrix graph also shows pattern. Matrix chart of which players are on the pitch together when England score.  The central purple section is smaller than it was after the Tongan game, and now only includes Pearce-Paul, Savala, Smith, Young and Brimson.  The next darkest section has Ashton, Litten and Walters.  The rest is a bit of a muddle, described below, until you hit the palest section, those who played together least often, which is Eckersley, ONeill and Havard. Because there were a number of changes in the squad between the two games, and because there were far more point-scoring moments in the game against France, the matrix has developed an interesting shape. 

Eckersley, O'Neill and Havard are consigned to the pale, not-often-on-the-pitch-together-when-England-score section. 

There's a shade of darker yellows to mid orange, these are the players that played more of a role in one game or the other (skewed, as I said, by more point-scoring moments in the France game) - they are Welsby, Lewis, Radley, Williams, Thompson, King. 

Radley, although he's in the middle of this group is the darkest orange line, possibly because he played quite a bit in both games, while Welsby and Lewis didn't play in the Tonga game. 

Then we have the next darkest section, a tomato ketchup red, with Ashton, Litten and Walters. Ashton only played in the France game, Litten played in both but only 20 minutes of the Tonga game, but Walters played in both. 

The darkest section, a deep red wine sort of colour, is those players who played large parts in both matches. They are Pearce-Paul, Savala, Smith, Young and Brimson, who seem to be McDermott's core group, plus whichever other winger he picks and a hooker. 

That's also reflected in the network graph, which there are now enough data points for: Network graph of England players who play together often when England score.  Pearce-Paul, Savala, Smith, Young and Brimson are in the centre.  Farnworth and Nsemba almost join them.  Williams is further out, as are Thompson, King, Lewis, Welsby, Litten, Ashton, Walters and Radley. 

I didn't expect Williams to be that far out, given he's captain and played in both matches. 

England's hopes rest on the Tonga vs Samoa game which was always one of my picks for most interesting match of the group stage.