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

At the start of the season, every second race featured a random check of a top 10 finisher. This pattern stopped about 6 races in, and I wondered if that was because of less than 15 drivers finishing. 

The British Grand Prix gave an opportunity to test this. Barcelona had technically provided one as well, but several of the “finishers” there had actually stopped on track and were only classified because they had completed more than 90% of the race distance. 

There was no check at the British Grand Prix even though there were 19 finishers. That suggests that this inspection is not a mandatory check, but something the Technical Committee can choose to carry out occasionally. 

Scrutineering at the British Grand Prix: While Car 5 (Bortoleto) looks like he's been checked far more the expected tests on first glance, compared to a statistically significant point of 2 standardised residuals, it really isn't. This is supported by the results having an overall χ² of 1.2216, when it would need to be >32.671 to be significant.

Scrutineering up to the British Grand Prix:  

The χ² over the season so far is 3.1607, which is not significant.  

The order is starting to make sense, with Hamilton and Russell who are doing well being the most over-scrutinised, while Stroll, Albon, Alonso, Hulkenberg and Bottas, whose cars keep breaking on them, being the most under-scrutinised. Antonelli has only been tested as many times as expected even though he is doing very well, but the number of events is small enough that that is not implausible.

Formula 1 2026 - Hungarian Grand Prix

Overall, the race wasn't that interesting - ignoring Verstappen and Hamilton's efforts to provide interest and entertainment, with help from Liam Lawson, who I'm sure wasn't influenced by the team he drives for. I'm not sure if it's because the track is that narrow that even the new regulations can't produce overtakes (plus the tyres going off in unexpected ways).

I'm not saying Aston Martin's season has been pants, but both cars finishing is being treated as a great success. At the same time, with one set of upgrades, they're boosted into the midfield.

Piastri appears to have inherited the terrible Antipodean luck of his manager. Possibly spending too much time in Australia explains why Bottas's brakes catching fire. Seriously, Cadillac, sort that out.

On the other hand, George Russell's luck isn't much better.

Withdrawals from the 2026 Tour de France

Unlike 2025, this year, riders were cut due to being over the time limit.  

The two unfortunate riders were Kellen O'Brien and Anthon Charmig. 

It's interesting to see how the pattern of withdrawals changes over the 3 "weeks" (weeks being used loosely here as always).  

In week 1, mid-stage abandonments are the most common, with 50% of all week 1 withdrawals being abandonments. 

In week 2, this switches to being a majority of "did not start" type withdrawals, possibly as the race caught up with people. 

Then, in week 3, it swings back to being mostly mid-stage abandonments, possibly because people have reached the "I'll be damned if I don't get to Paris after all this" mind-set. 

This pattern is repeated if you look at it as "withdrawal type by week". Which stages did the withdrawals happen on? 

This year, while a couple of the stages did have a few more withdrawals they're reasonably evenly spaced. Stage 6, the one with the most withdrawals, unsurprisingly, was the Tourmalet. Stage 15 was the one with Vingegaard's crash, while stage 13 was the one with Gaviria's horrible crash.  

You are looking at someone smug that she had figured out how to recolour charts in Excel. Because the "still ins" were a drab grey. I prefer a thematic yellow. 

Lotto-Intermarche, Caja Rural-Seguros RGA and Soudal Quick-Step were all down to 5 riders. Lidl-Trek, Bahrain Victorious, Cofidis, Decathlon, EF Education Easy Post, Ineos, Total Energies and Astana all retained all their riders. 

There's no real patterns, to this or in the Kaplan-Meier survival curve split by team.  

If the team names and colours stick for more than 5 minutes, I may recolour this to match team colours, but 23 teams is a lot of colours to keep track of. 

For the Tour as a whole, the Kaplan-Meier looks like this.  

My takeaway is that this time there was a long, slow decline, with no sudden drops, not even on the Tourmalet. 

I have been doing this for far too long and so can compare this year's Kaplan-Meier to those of the preceding 6 years. 2026 fits into the same pattern as the others, with only 2021 standing out.

Formula 1 2026 - Belgian Grand Prix

 I had a moment of revelation at the start of this grand prix - around about when Leclerc and Hamilton moved past Verstappen and Russell - this Ferrari car isn't bad. It's not as good as the Mercedes, but the Mercedes is a good car, even if it's not as good as *that* rocket ship Mercedes or that Red Bull. See also sort of keeping up with the Mercedes cars after they overtake.

I'm not quite sure what to do with the information.

Mostly because having a good car isn't enough, you need good drivers (we've got that) and a solid strategy team ... which Ferrari haven't got. This time it was Hamilton who was affected by strategy gremlins. Now I am guessing they had to do some quick thinking following the 5 second time penalty but they didn't seem to have communicated the revised plan to their driver which is a really quick way of losing the driver's trust.

Hamilton did seem to be having a day of it; between the penalty and the pit-stop incident and the "not doing what he wanted doing to his front wing" and ... Thankfully the pit crew member appears to be fine.

I am of the camp that think Hamilton got walloped with a penalty where Leclerc didn't because Russell ended up out while Piastri managed to get the car to the finish.

As to other non-Ferrari matters (other than Bottas holding up Leclerc to allow Antonelli to catch him and my tin foil hat insisting it was deliberate), Hadjar and Verstappen appear to be the pairing Red Bull have been looking for for years. Not sure if that's because the way the cars are set up under the current regulations doesn't suit Verstappen as well as previous cars or Hadjar is that good (I tend towards "Hadjar is that good" but I acknowledge my biases). Racing Bull and Alpine seem to be coming along nicely, as do Audi. For Aston Martin, the only way is up. I'm intrigued about their update coming at Hungary.

World Cup 2026 - Final Network Visualisation

The unlabelled network graph looks like this: Even without the labels, you can probably guess the big circles are Spain and Argentina ;) 

The labelled graph looks like this:  

It's the battle of the Atlético Madrid players! 

Atlético Madrid have 9 players left, 3 playing for Spain, 6 for Argentina. 

Barcelona have 8, but they all play for Spain. 

Arsenal and Athletic Bilbao are next, with 3 players each. All 6 of those players play for Spain. 

Atlético Madrid, Chelsea, Tottenham Hotspur and Bayer Leverkusen have players on both teams. 

Chelsea, somehow, are the club team closest to the centre of the diagram. 

No prediction at this stage, the two teams are understandably equidistant.

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

Some interesting points from the underlying data: The Austrian GP featured the first test where the check said "all classified cars" rather than all cars, or giving the number of the car tested. It was the steering wheel test. I have been assuming that "all cars" meant all cars entered. I think the wording of this check supports that but I want to give that caveat on all this data. 

The Austrian GP race scrutineering itself:  

No, I don't know what Liam Lawson did to the scrutineers. 

Obviously, it's not that extreme, because his standard residual of 1.06 doesn't hit the value of 2 that suggests significance but that's a lot of tests on one car. 

The difference from expected is not statistically significant with an χ² of 2.88. 

Season to date:  

The difference from expected is not statistically significant with an χ² of 3.06. 

The top 3 most over-scrutinized (Russell, Hamilton and Norris) make some sort of fits with their performance, because they are 3 of the top 5 at this point of the season. 

Similarly, two of the bottom 3 make sense, because the Aston Martins of Stroll and Alonso have not been going long distances so might miss some of the post-race tests, ditto for Bottas and Albon who have retired from half the races and a quarter of the races respectively. Hadjar being so low down still makes no sense, he's only not completed 2 races but all but one of the rest of the races he's been in the top 10. It's getting to the point where I want to triple check all my numbers, he's that low. 

For the British Grand Prix results, I'm going to be looking for that random top 10 finisher check, providing there are at least 15 finishers (which *spoiler* there were). If it doesn't happen, I may have to conclude there is no pattern.

World Cup 2026 - Semifinal Network Visualisation

I have listened to L's editorial guidance and am posting the post with a deadline. Austrian and British Grand Prix scrutineering posts are in the works, as is the first of the weekly Tour de France posts. I am making no promises on timelines because a work project is stealing my time. 

We're down to four teams and the network graph looks like this.  

The teams are close together, but one sticks out.  

Sort of understandably, being as Argentina is on a different continent to the other finalists, they are the team slightly further away from the others. 

The thing I find intriguing is that each national team has one club team close to it, because that club team provides a lot of national team representatives. For Argentina, it is Atlético Madrid (my heart belongs to Diego Simeone), England have Manchester City, France have Paris Saint Germain and Spain have Barcelona. 

It shows the financial strength of European football that the club team with the most players playing for Argentina is Atlético Madrid, not a team closer to home. 

The club teams with the most representatives left are Barcelona with 10, Atlético Madrid with 9 and Arsenal with 8. 

England are just about the team closest to the centre, with Chelsea the club team closest. 

There's something interesting going on in the community view. Although there are only 4 teams left, there are 8 communities. Labelled, it's clear that four of the communities are the nations, and the others are those teams that link more than one country (with some exceptions).  

The thing that intrigues me is that some teams aren't separate communities even though they link multiple teams. 

I think that happens when most of the players play for one national team, for instance of the five Aston Villa players left, three play for England. 

Chelsea will have a player in the final no matter what combination of teams wins the semifinal matches. 

Making predictions at this point is almost impossible, but the diagram says England and unclear because of how equidistant Spain and France are from the centre.

Formula 1 2026 - British Grand Prix

 Ferrari only win when I can't pay attention. Barcelona - I was being driven back from a fencing competition. Silverstone - being driven back from a fencing competition. Clearly, the solution to get Ferrari winning more is for me to do more competitions. [Editorial note, the joke is L's]

Ferrari mustn't have been expecting Leclerc to start so well given it meant they had to switch to strategy B by lap 8.

It was a race with several "things I hadn't seen before". An umbrella causing a (virtual) safety car and a brake aero duct failure causing a car to stop being able to steer.

(Technical details sort of here - https://www.formula1.com/en/latest/article/a-car-should-not-break-wolff-laments-antonelli-mechanical-issue-with-italian-denied-epic-battle-for-silverstone-win.5juL2jTds26Hl4foTZIWnk)

That penalty for Antonelli was kicking a man when he was down. This time he wasn't deliberately going beyond track limits.

So happy that Ferrari got a 250th F1 Victory. Harry Benjamin on the BBC commentary was not helping by describing Leclerc as "the man who bleeds scarlet". We Ferrari fans are already unreasonable on the subject of Leclerc.

World Cup 2026 - Quarterfinal Network Visualisation

(The Austrian Grand Prix scrutineering post, and the British Grand Prix summary and scrutineering post are coming, I just thought this was a bit more urgent) 

Okay, last's round's predictions were not as good as they could have been. I got Brazil vs Norway right, and was right to suspect England vs Mexico would be close, got France, Argentina and Morocco's results correct, but was way out on Belgium vs United States and got Switzerland vs Colombia wrong (although I said it would be close). I was right that Spain being pulled out of the centre of the diagram by all their Barcelona players made it hard to predict their result from the diagram. 

Also, I didn't predict how close Egypt got to beating Argentina. 

What do the quarterfinal diagrams look like? 

The unlabelled diagram looks like this: Labelled it looks like this: France are the national team closest to the centre, with Paris Saint-Germain the club team closest. 

The club teams with the most representatives left in are Atlético Madrid, Barcelona and Arsenal with 10. 

The average number of degrees of contact each circle has is 1.364. 

The number of communities is still one per team, so 8. 

The unlabelled community view looks like this: Labelled, it looks like this:  

What about my predictions? 

It's much harder at this point, with the teams other than France pretty much equidistant from the centre, and no "gravitational" clump. 

France vs Morocco - Diagram says France 

Spain vs Belgium - Diagram says Belgium, although I think that's because Spain are still pulled out of the centre by all those Barcelona players. On the other hand, anything can happen in knock out football. 

Norway vs England - Norway just (that sound you can hear is L cursing me) 

Argentina vs Switzerland - Diagram says Switzerland. I think that's the Atlético Madrid players pulling Argentina out. Also [Legal team have redacted a paragraph-long rant about exactly how many World Cups the referees are going to give this Argentina team.] Switzerland could do it ... (I feel I need to add a conflict of interest statement, I drew Switzerland in a sweepstake at work.)

World Cup 2026 - Last 16 Round Network Visualisation

Delayed by fencing, which I enjoyed immensely despite the sprained ankle. 

Second round results: I said Senegal vs Belgium would be a cracker. 

The whole centre of diagram gravity vs centre of diagram really did come into play in the second round - if you were a team on the left hand side of the diagram you went out. Except Egypt, and they were playing another team from the left hand side. 

Of the matches where there was a clear prediction, gravitational centre ended up being more important, which is useful learning. 

The diagram for the last 16 teams isn't as tightly packed any more, which means that should be less of a thing going forward.  

The labelled version of the graph looks like this:  

The list of club teams with the most representatives has changed - especially Bayern Munich given Germany are out. What we've got instead are a small clump of teams with 12 players left in (and yes, these are the teams that were causing the cluster in blue in the first round and in the top right of the diagram in the second). They are Paris Saint-Germain, Arsenal, Barcelona and Manchester City. Atlético Madrid is the next on 11. 

Norway are the national team closest to the centre, while Inter Milan are the club team closest. 

The average number of degrees each circle is linked to has gone down to 1.377. 

The number of communities in the communities view is now 16, so one per team. Labelled it looks like this:  

As to my predictions (which I don't mind making late because you can see the diagrams I'm basing them on): 

France vs Paraguay - France 

Canada vs Morocco - Morocco 

Portugal vs Spain - Diagram says Portugal, reason says Spain; all those Barcelona players pull Spain out of the centre, the same thing that happens to their women's team and always skews the women's diagrams. 

United States vs Belgium - Diagram is not clear, this may be closer than people expect. 

Brazil vs Norway - Diagram says Norway, I don't think that's impossible. 

Mexico vs England - Diagram is also not clear 

Argentina vs Egypt - Argentina 

Switzerland vs Colombia - Diagram says Colombia but could be close!