40 replies to this topic

[Primo]

In the light of the USGP at COTA, I've asked myself what kind of software the F1 teams are using. 
We had (at least) three strange decisions:
 

1. Putting Ricciardo on a one stopper
   Now, I can imagine that they wanted to make his first race a bit more "leisure", something that automatically becomes a the result of having to save tires. With that strategy they could also hope to get lucky with an SC, but it was still strange.
    
2. Ferrari putting their pole car on a one stopper.
   This makes absolutely no sense at all. The "maybe we get an SC" should only be part of plan D-Z if you start at the front. Leclerc was the one with the easiest of Plan A's since was already in the lead of the race when the lights went out. Not only am I  confounded by Ferrari's choice there, I am also surprised, and a bit disappointed that Charles agreed to it. There was absolutely nothing pointing towards that a one stopper could give a competitive advantage.
    
3. Mercedes not knowing what strategy they was on. 
   This was strange. Seems that they, like Ferrari, had a one stopper in mind but they had not stamped it as "Plan A". At the grid they talked about an "offset" of 5 laps, that probably meant "go on for 5 laps after the leaders pit". A 1.5 stopper...?

This is the situation Mercedes from when Max drives into the pitlane:
 

• LAP 17 VER PITS
  
  NOR - 1.8
  VER 4.6 lap 17 (last lap VER+HAM  1.42.6
• LAP 18 NOR PITS
  Distances to HAM: NOR + 18,5 and VER + 22.6
  RADIO: Pitwall ask about "+5 laps" and HAM answer, "I'm not sure, pretty tough"
• LAP 20
  RADIO: "Verstappen inside our pit window" (VER laptime on lap 19 2.5s faster than HAM)
  HAM: "I'm struggling out here"
• LAP 21 HAM PITS
  
  Comes out 5 seconds behind VER, 6 Behind NOR
  End of lap, the distance is 6,5 and 9s to Max and Norris

In four laps he has lost 11 seconds to Verstappen and Norris!

I mean, they have sector times! How is it possible that they needed 4 laps to realize they had made a mistake?

So, going back a bit: Let's forget Ricciardo's one stopper since it is possible they took precautions due to his hand, but Ferrari choosing it as Plan A and Mercedes wanting to keep it as a possibility. One stop less means they win 20 seconds over the race, but that is only 0.35s per lap! That is not much considering a new tire initially give you 2 seconds and then it keeps on giving. They also had almost zero long runs in practice and the track is very bumpy which increase the risk of lock ups and excessive wear which means that any indications that a one stopper might work must have been very vague.

We now know that a one stopper was bad, but the question is - why didn't Ferrari, and Leclerc, know? We also know that those 4 extra laps that HAM did might have cost him the win (it feels better to be DQ'ed after a win than after 2nd), but why did not Mercedes see that after just a couple of Max's new tire sectors?

I'm a software engineer and I am baffled by these mistakes, and in Ferrari's case the amount of mistakes they do. You do not need a particularly sophisticated software system to catch the stuff they fumble with.

So does anyone know anything about what kind of software they use at pitwall? What do they have at their disposal when they make the decisions?

Oh, and let's not forget Ferrari's question to Leclerc about a possible Plan Þ when there was 10 laps left. Amazing. Charles must learn from Carlos and speak up early when they ask him to do something stupid.
 

 

[OO7]

Hamilton gained time due to being on fresher tyres, but how much...?

Edited by OO7, 25 October 2023 - 17:38.

[noikeee]

I'm a software engineer and I am baffled by these mistakes, and in Ferrari's case the amount of mistakes they do. You do not need a particularly sophisticated software system to catch the stuff they fumble with.

Same. I know sometimes a deep dive into data brings out patterns we can't see, but the amount of times I'm just watching a race on the TV without live timing and mid race I figure out they're on a hopeless strategy and there's still time to switch but they don't, is absolutely baffling.

I think they're using a shitty Excel spreadsheet.

By the way this has been going on all the way back since Kimi Raikkonen returned to Ferrari. Back then you could kinda see why sacrifice Kimi for the faster driver, and try hail Mary weird strategies with the slow guy, but now they do the stupid **** with the faster driver. It's spectacularly self-defeating.

Edited by noikeee, 24 October 2023 - 19:20.

[1player]

I'm a software engineer and I am baffled by these mistakes, and in Ferrari's case the amount of mistakes they do. You do not need a particularly sophisticated software system to catch the stuff they fumble with.

I'm one as well, and I don't see how this problem is anything but enormously complex.

I don't have insight into the world of F1, but I'm pretty sure the entire strategy system is a massive machine learning algo that takes millions of variables as input and spews probabilities of outcomes on the other side. These algos don't tell you this is better than that, but that plan A has a 51.2% of success and plan B has a 49.8% of success.

Given the impossibility of having perfect realtime input variables, the strategists need to interpret this data, and sometimes the interpretation is incorrect. Garbage data in, garbage data out.

Described like this, F1 strategy is basically ultra-high tech reading of tea leaves. You're bound to get it wrong many times, but not as many times as throwing a coin flip.

Edited by 1player, 24 October 2023 - 19:21.

[Primo]

Hamilton have gained time due to being on fresher tyres, but how much...?

He lost 11 seconds and got half of it back. 

[noikeee]

I'm one as well, and I don't see how this problem is anything but enormously complex.

I don't have insight into the world of F1, but I'm pretty sure the entire strategy system is a massive machine learning algo that takes millions of variables as input and spews probabilities of outcomes on the other side. These algos don't tell you this is better than that, but that plan A has a 51.2% of success and plan B has a 49.8% of success.

Given the impossibility of having perfect realtime input variables, the strategists need to interpret this data, and sometimes the interpretation is incorrect. Garbage data in, garbage data out.

Described like this, F1 strategy is basically ultra-high tech reading of tea leaves. You're bound to get it wrong many times, but not as many times as throwing a coin flip.

Ok but their algorithm seems to be overvaluing all the million tiny variables instead of the large variables right in their face screaming at them - if your driver is losing 2+ secs a lap to everyone else in fresh tyres and there's the entire race ahead of you, you probably should pit.

[Primo]

I'm one as well, and I don't see how this problem is anything but enormously complex.

I don't have insight into the world of F1, but I'm pretty sure the entire strategy system is a massive machine learning algo that takes millions of variables as input and spews probabilities of outcomes on the other side. These algos don't tell you this is better than that, but that plan A has a 51.2% of success and plan B has a 49.8% of success.

Given the impossibility of having perfect realtime input variables, the strategists need to interpret this data, and sometimes the interpretation is incorrect. Garbage data in, garbage data out.

Described like this, F1 strategy is basically ultra-high tech reading of tea leaves. You're bound to get it wrong many times, but not as many times as throwing a coin flip.

If the above is true then, regardless of its complexity, it is useless if
A) It cannot handle real time data and
B) It spits out results that, in spite of millions variables, is crap. 

Those kind of systems can certainly be complex, but complexity has only a value if the results become more accurate.

[1player]

Ok but their algorithm seems to be overvaluing all the million tiny variables instead of the large variables right in their face screaming at them - if your driver is losing 2+ secs a lap to everyone else in fresh tyres and there's the entire race ahead of you, you probably should pit.

The hard part is interpreting those results that have large margins of uncertainty, of course.

The Machine can tell you with 99.9% certainty that to win the race you need to qualify. But to choose tyre A instead of B, the certainty drops to close-to-coin-flip chances. Therein lies the complexity.

EDIT: and chances are constantly updated, in real time. By the time you call your driver to box, the Machine might tell you that boxing now is no longer a good idea. It was, 200 milliseconds ago, but now not anymore.

Of course after the fact, in hindsight, bad choice seem obviously bad. None of these machines have the benefit of hindsight.

Edited by 1player, 24 October 2023 - 19:34.

[noikeee]

The hard part is interpreting those results that have large margins of uncertainty, of course.

The Machine can tell you with 99.9% certainty that to win the race you need to qualify. But to choose tyre A instead of B, the certainty drops to close-to-coin-flip chances. Therein lies the complexity.

Of course after the fact, in hindsight, bad choice seem obviously bad. None of these machines have the benefit of hindsight.

I get that, I understand uncertainty, I even understand that if the car is particularly bad in the race compared to qualifying there is no strategy that can save it (and there have been occasions where this has definitely happened to Ferrari), but it seems to happen way too often to one particular team and one particular driver. And mid race we always get the feeling that they are wrong and that they can still switch and limit the damage. Mid race we don't have hindsight yet.

[dissident]

I think the problem is they don't watch the race itself. At least I can't think of another explanation.

 

Without having read any debriefs or anything, my feeling is both Mercedes and McLaren didn't think they could beat Max on the same strategy and hence the funky calls.

 

McLaren even had a second chance once Norris managed to stick with Max after the overtake, yet they failed to react. 

[1player]

If the above is true then, regardless of its complexity, it is useless if
A) It cannot handle real time data and
B) It spits out results that, in spite of millions variables, is crap.

Those kind of systems can certainly be complex, but complexity has only a value if the results become more accurate.

If there is only one team using machine learning on the grid, it'll always win and beat the others. But ALL teams have it, so it's basically an infinite game of chess with enormously powerful machines. The margins of strategic victory are very small, so to actually win as a team, you need more than just strategy.

[PlatenGlass]

If there is only one team using machine learning on the grid, it'll always win and beat the others. But ALL teams have it, so it's basically an infinite game of chess with enormously powerful machines. The margins of strategic victory are very small, so to actually win as a team, you need more than just strategy.

It will only win if it's good. Machine learning isn't automatically good.

[Grippy]

Don't know if any names here can guide you, list appears old

https://www.f1techni.../links.php?c=39

 

Processia were with Renault.

SmithBayes were McLaren

 

Tenuous 2006 article

https://www.f1technical.net/news/4102

 

With your software knowledge there might be some buzz-words you can glean to dig deeper. 'decision engine'

 

I learnt that the teams computers have to have their internals coated due to the carbon particles flying around and causing shorts.

[PlatenGlass]

 

Without having read any debriefs or anything, my feeling is both Mercedes and McLaren didn't think they could beat Max on the same strategy and hence the funky calls.

 

 

 

I think this was the problem at COTA. Normally the team behind might try something different to get in front. But the teams are so intimidated by Max / Red Bull that even when they're in front they can't bring themselves to just stick with a normal strategy and put the onus on Red Bull to pass them.

[Primo]

If there is only one team using machine learning on the grid, it'll always win and beat the others. But ALL teams have it, so it's basically an infinite game of chess with enormously powerful machines. The margins of strategic victory are very small, so to actually win as a team, you need more than just strategy.

What we have seen so far, on our TV screens, is that the AWS predictions regarding tire life etcetera, are usually far from accurate so I doubt the teams put much confidence in whatever their system gives them during the race. Maybe Ferrari and Mercedes did rely on them, but that would only be proof their systems are bad.

I would also argue that Mercedes did a huge strategic blunder in Qatar when they put Hamilton on softs and Russel, who started ahead, on mediums. With one step softer tires you can realistically only hope to jump one car and in the case of Mercedes, they could hope that Hamilton jumped Russel and having their cars 2 & 3 instead of... 2 & 3. Yes, certainly it would have been possible for HAM to also challenge VER, but from a mathematical point of view their strategy was more likely to impede Russel than to threaten Verstappen.   

[Primo]

I think this was the problem at COTA. Normally the team behind might try something different to get in front. But the teams are so intimidated by Max / Red Bull that even when they're in front they can't bring themselves to just stick with a normal strategy and put the onus on Red Bull to pass them.

This might very well be the case and that is partly why I started this thread because I cannot see how a computer system would rank those Merc and Ferrari strategies above the ones of McLaren and Red Bull (which was also the strategies that Pirelli predicted to be the fastest).  
 

[flyboym3]

Great thread BTW.
The models are only are only as good as the assumptions the humans have modelled.

In the end you can't beat a big brain and a banana. Or Hannah.

There are people in this forum that have developed way better competency than those in race strategy sheerly because of the experience they have gained from watching lots of races unfold.

DC summarised it well. Its a high degradation track. Ferrari suffered the most with the wear from front runners in sprint race. Yet they modelled a 1 stop. Incompetent assumptions leads to incompetent outputs.

Some of these strategy folk are oxford graduates with degrees in maths but imo that kind of iq is not important.

[Primo]

@flyboym3

Yeah... I am trying to remember strategies that did not seem to makes sense but eventually turned out great, but only thing that springs to mind is Schumi's 4-stopper in France. I'm sure there are others examples, but the Plan A strategy is very far from rocket science. Give the available data to a class of fifth graders armed with only pen and paper and 90% of them would deliver the same strategy as the teams own strategists.

The only time you really need a computer to help you is when something happens and you need to make a quick decision and there are a lot of things in motion. A computer should have told the Merc pitwall already at lap 18, with a myriad of red alerts accompanied by the theme from Jaws,  that Max would be inside their pit window already at lap 19.  

 

[Grippy]

I found a course in the uk

https://www.schoolof.../race-strategy/

They have a 'Build your own Race Strategy gapper tool'

https://www.schoolof...gy-gapper-tool/

 

RaceWatch is used by 7 of the F1 teams; How Formula 1 Teams Optimize Race Strategy and Analysis – A Comprehensive Review

https://www.catapult...rategy-analysis

 

 

 

https://sbgsportssof...watch-for-teams

[Grippy]

 i don't know if this is posted anywhere else

Pierre Gasly’s hilarious response to finding out Ferrari’s US GP strategy plan

https://www.planetf1...eclerc-one-stop

[F1 Mike]

These problems are similar across every industry in the world - if input data is incorrect or not fully up to date in computer, computer then tells lies.

In engineering, it's common for something to be changed on a technical drawing, and the draughtsperson then expects the computer to shuffle everything else around correctly. This is something the computer does amazingly well. However, one single incorrect parameter on input and it can cause a heavy amount of chaos to the whole project.

This is the scariest thing about AI as well - if left to it's own devices without the human element of common sense, everything can unravel very quickly

[GrumpyYoungMan]

I think they rely far too much on software rather than “human/driver” instinct…

As for the software they use, some kind of clever AI software, but is it bespoke to the teams, it makes you wonder… there is so much white labelling of products now…
 

These problems are similar across every industry in the world - if input data is incorrect or not fully up to date in computer, computer then tells lies.

In engineering, it's common for something to be changed on a technical drawing, and the draughtsperson then expects the computer to shuffle everything else around correctly. This is something the computer does amazingly well. However, one single incorrect parameter on input and it can cause a heavy amount of chaos to the whole project.

This is the scariest thing about AI as well - if left to it's own devices without the human element of common sense, everything can unravel very quickly

AI needs to be put back in the box, before it’s late, which it is already…

AI will kill off more jobs than automation every has…

Edited by GrumpyYoungMan, 25 October 2023 - 07:36.

[Rediscoveryx]

I think there is another aspect to consider.

I don’t think that it’s difficult att all to theoretically calculate the quickest strategy, or to factor in safety car probabilities (and even potential rain etc). But the problem is that it’s not difficult to do for the other teams either.

So if your car is not the quickest one out there, you could imagine that if you chose the fastest (seemingly most logical) strategy then you’ll be doomed to finish second at best assuming that the quickest team also goes down the same strategic route.

Example; perhaps Ferrari could see that a two stop strategy would be theoretically superior to a one-stop for Leclerc. However, assuming that other teams also opted for a two-stop, their win probability could be pretty much zero regardless. While an offsetting strategy (while theoretically slower) could still yield a higher win probability if there are scenarios that unfold differently through the race (digferent wear, safety car timing etc).

Then it becomes a question about how to prioritize. A safe third place? Or take a gamble with a lower expected finishing position, but with a higher potential win probability?

[Jops14]

Ok but their algorithm seems to be overvaluing all the million tiny variables instead of the large variables right in their face screaming at them - if your driver is losing 2+ secs a lap to everyone else in fresh tyres and there's the entire race ahead of you, you probably should pit.

Ultimately i think this is a problem and one Ferrari particularly seem afflicted by. I think it comes from poor leadership and people essentially being paralysed by indecision and not wanting to be wrong.

You end up essentially in a situation where, you can see its going horribly wrong but the model says X, and X mist be better because we talked about it

Then look at Red Bull, they regularly roll the dice and make proactive choices based on the race happening in front of them

[GrumpyYoungMan]

I think there is another aspect to consider.

I don’t think that it’s difficult att all to theoretically calculate the quickest strategy, or to factor in safety car probabilities (and even potential rain etc). But the problem is that it’s not difficult to do for the other teams either.

So if your car is not the quickest one out there, you could imagine that if you chose the fastest (seemingly most logical) strategy then you’ll be doomed to finish second at best assuming that the quickest team also goes down the same strategic route.

Example; perhaps Ferrari could see that a two stop strategy would be theoretically superior to a one-stop for Leclerc. However, assuming that other teams also opted for a two-stop, their win probability could be pretty much zero regardless. While an offsetting strategy (while theoretically slower) could still yield a higher win probability if there are scenarios that unfold differently through the race (digferent wear, safety car timing etc).

Then it becomes a question about how to prioritize. A safe third place? Or take a gamble with a lower expected finishing position, but with a higher potential win probability?

If you have the fastest car, you also have a far wider choice which you can make work to your advantage...

[Lowgrip]

Maybe Mercedes and Ferrari are too focused on what they can do and completely ignore what their rivals can(in terms of probalities) do too, so to be ready with countermeasures.

The fact they have been caught so many times with their pants down suggest to me this might be the case.

Or if they are already doing it, they will come up with better strategies by using their only eyes since their model is scrap or too slow.

Edited by Lowgrip, 25 October 2023 - 09:00.

[jcbc3]

I think there is another aspect to consider.

I don’t think that it’s difficult att all to theoretically calculate the quickest strategy, or to factor in safety car probabilities (and even potential rain etc). But the problem is that it’s not difficult to do for the other teams either.

So if your car is not the quickest one out there, you could imagine that if you chose the fastest (seemingly most logical) strategy then you’ll be doomed to finish second at best assuming that the quickest team also goes down the same strategic route.

Example; perhaps Ferrari could see that a two stop strategy would be theoretically superior to a one-stop for Leclerc. However, assuming that other teams also opted for a two-stop, their win probability could be pretty much zero regardless. While an offsetting strategy (while theoretically slower) could still yield a higher win probability if there are scenarios that unfold differently through the race (digferent wear, safety car timing etc).

Then it becomes a question about how to prioritize. A safe third place? Or take a gamble with a lower expected finishing position, but with a higher potential win probability?

 

Back in the mid-nineties, Ron Dennis was asked how they planned for one or two stoppers and when to do them tactically. He said that since 'anything' can happen, it would only make sense to plan your race for the fastest possible time totally. And after that it was a question of reading the race and respond accordingly.

 

Seems pretty basic to me.

[Primo]

I think there is another aspect to consider.

I don’t think that it’s difficult att all to theoretically calculate the quickest strategy, or to factor in safety car probabilities (and even potential rain etc). But the problem is that it’s not difficult to do for the other teams either.

So if your car is not the quickest one out there, you could imagine that if you chose the fastest (seemingly most logical) strategy then you’ll be doomed to finish second at best assuming that the quickest team also goes down the same strategic route.

Example; perhaps Ferrari could see that a two stop strategy would be theoretically superior to a one-stop for Leclerc. However, assuming that other teams also opted for a two-stop, their win probability could be pretty much zero regardless. While an offsetting strategy (while theoretically slower) could still yield a higher win probability if there are scenarios that unfold differently through the race (digferent wear, safety car timing etc).

Then it becomes a question about how to prioritize. A safe third place? Or take a gamble with a lower expected finishing position, but with a higher potential win probability?

That is certainly aspect to consider, but then you also must consider that Ferrari put Leclerc on the same starting tire as a the rest of the front runners for the start. That meant that he did not have tires that could bring him further than the opposition and instead had to rely on a disciplined right foot while the others rocketed away towards glory. So it still made no sense. If he had started 7th, then maybe it would have been a reasonable gamble, but not from pole.

[Rediscoveryx]

That is certainly aspect to consider, but then you also must consider that Ferrari put Leclerc on the same starting tire as a the rest of the front runners for the start. That meant that he did not have tires that could bring him further than the opposition and instead had to rely on a disciplined right foot while the others rocketed away towards glory. So it still made no sense. If he had started 7th, then maybe it would have been a reasonable gamble, but not from pole.

Right.

By the way; how certain are we that Ferrari were planning a one-stopper all along? Maybe that was their fall-back option should Leclerc fail to convert pole into an early lead?

[Grippy]

Right.

By the way; how certain are we that Ferrari were planning a one-stopper all along? Maybe that was their fall-back option should Leclerc fail to convert pole into an early lead?

Check the link in my post #20 above, LEC dicussed with GAS prior to race.

[Sterzo]

One day we will hear the following radio exchange:

LeClerc: "Are you sure this is the right strategy?"

Reply: "This is the Ferrari pitwall here. You may have called the wrong number."

[taran]

Back in the mid-nineties, Ron Dennis was asked how they planned for one or two stoppers and when to do them tactically. He said that since 'anything' can happen, it would only make sense to plan your race for the fastest possible time totally. And after that it was a question of reading the race and respond accordingly.

 

Seems pretty basic to me.

Then why were McLaren (and Williams, another old school team) made to look silly by Brawn and Schumacher in both the Benetton and subsequent Ferrari days?

 

Brawn had an ability to think outside the box both before and during a race and change strategies on the fly. Schumacher was then able to execute them (like run 40 qualifying laps in the race for example). Williams in the 1990s seemed particularly weak in race strategy decisions and McLaren wasn't that great either.

 

At the time, it felt like they relied too much on the speed of their car to win and never truly factored in how race strategies could help win the race for them. Benetton typically operated with a power disadvantage when Brawn was there (until they got Renault engines in 1995) and likely honed their strategy options to overcome the outright speed deficit.

 

And that would bring us neatly to Red Bull which also operated under a power disadvantage since 2014 and also required clever strategies to be truly competitive. Once you have that mindset and can keep it and you have a fast car too, you get dominance.

 

We often admire Red Bull's quick thinking during a race. When did Mercedes or Ferrari surprise anyone with a surprising strategy twist?

[jcbc3]

...And after that it was a question of reading the race and respond accordingly.
 
...

 
 

Then why were McLaren (and Williams, another old school team) made to look silly by Brawn and Schumacher in both the Benetton and subsequent Ferrari days?
 
...

My post was meant to comment on the thinking before the race started. I.e. do not over think what might or might not happen in the future. I will posit that Brawn and Ferrari also planned for a fastest race scenario. What Brawn and Schumacher did extremely well, was the adaption of the strategy during the race.

[DW46]

Then why were McLaren (and Williams, another old school team) made to look silly by Brawn and Schumacher in both the Benetton and subsequent Ferrari days?

Brawn had an ability to think outside the box both before and during a race and change strategies on the fly. Schumacher was then able to execute them (like run 40 qualifying laps in the race for example). Williams in the 1990s seemed particularly weak in race strategy decisions and McLaren wasn't that great either.

At the time, it felt like they relied too much on the speed of their car to win and never truly factored in how race strategies could help win the race for them. Benetton typically operated with a power disadvantage when Brawn was there (until they got Renault engines in 1995) and likely honed their strategy options to overcome the outright speed deficit.

And that would bring us neatly to Red Bull which also operated under a power disadvantage since 2014 and also required clever strategies to be truly competitive. Once you have that mindset and can keep it and you have a fast car too, you get dominance.

We often admire Red Bull's quick thinking during a race. When did Mercedes or Ferrari surprise anyone with a surprising strategy twist?

Mercedes - 2019
Ferrari - Every week but not in a good way.

[OO7]

He lost 11 seconds and got half of it back. 

But it's difficult to know how much of that time he gained back was down to him being of fresher tyres.  However I realise Mercedes should have simply undercut Norris.

[Jarninho]

I don't have insight into the world of F1, but I'm pretty sure the entire strategy system is a massive machine learning algo that takes millions of variables as input and spews probabilities of outcomes on the other side. These algos don't tell you this is better than that, but that plan A has a 51.2% of success and plan B has a 49.8% of success.

 

A couple of years ago I heard somewhere that Merc was just using an excel sheet. 

 

Mind you, for every time I am right in thinking 'god these guys are way off!', I am wrong multiple times, or at the very least I have no idea how it's going to pan out.

[Rediscoveryx]

A couple of years ago I heard somewhere that Merc was just using an excel sheet.

Mind you, for every time I am right in thinking 'god these guys are way off!', I am wrong multiple times, or at the very least I have no idea how it's going to pan out.

Sure, but you haven’t (I assume) got access to anywhere close to the same level of data as the teams have.

[taran]

 
 My post was meant to comment on the thinking before the race started. I.e. do not over think what might or might not happen in the future. I will posit that Brawn and Ferrari also planned for a fastest race scenario. What Brawn and Schumacher did extremely well, was the adaption of the strategy during the race.

It does seem like some teams simply look at how long certain tyres can last at what lap time and then calculate the fastest strategy, e.g. 1 stop, 2 stop etc. Effectively the basics or base line strategy.

 

Other teams seem to consider such variables as the undercut, overcut, ease of passing at specific tracks and how certain choices made by other teams can impact your own strategy and might require an immediate change.

 

I think we can all agree that Ferrari and Mercedes are poor at reacting to changing circumstances and have a tendency to adhere to a race strategy even after it seems obsolete. These are also the biggest and most corporate teams in F1 so I would suspect that there is some kind of correlation. Maybe data is being received slower because it's being filtered through too many layers? Maybe multiple strategy chiefs need to discuss changing race strategy before a decision is made? After all, Ferrari fired their race strategist after Alonso lost the 2010 title so an amount of "sharing the blame" amongst multiple people is safer....It is remarkable IMO that both Sainz and Russell seem more willing to think out of the box than their own teams when it comes to (on the fly) race strategy.

 

However you look at it, race strategy has become a major performance differential, just like aero, chassis and engine. And it feels like some teams have not yet reached the same sophistication in it as others and still run races the old-fashioned way by simply running as fast as they can until the tyres are shot and then boxing for new ones without regard for what other teams are doing.

[Marklar]

Same. I know sometimes a deep dive into data brings out patterns we can't see, but the amount of times I'm just watching a race on the TV without live timing and mid race I figure out they're on a hopeless strategy and there's still time to switch but they don't, is absolutely baffling.

I think they're using a shitty Excel spreadsheet.

By the way this has been going on all the way back since Kimi Raikkonen returned to Ferrari. Back then you could kinda see why sacrifice Kimi for the faster driver, and try hail Mary weird strategies with the slow guy, but now they do the stupid **** with the faster driver. It's spectacularly self-defeating.

close, it's excel VBA

[krea]

What we have seen so far, on our TV screens, is that the AWS predictions regarding tire life etcetera, are usually far from accurate so I doubt the teams put much confidence in whatever their system gives them during the race. Maybe Ferrari and Mercedes did rely on them, but that would only be proof their systems are bad.

I would also argue that Mercedes did a huge strategic blunder in Qatar when they put Hamilton on softs and Russel, who started ahead, on mediums. With one step softer tires you can realistically only hope to jump one car and in the case of Mercedes, they could hope that Hamilton jumped Russel and having their cars 2 & 3 instead of... 2 & 3. Yes, certainly it would have been possible for HAM to also challenge VER, but from a mathematical point of view their strategy was more likely to impede Russel than to threaten Verstappen.

Is the AWS thing actually a real calculated thing or just a branding of whatever an intern thinks look about right?

[Primo]

Is the AWS thing actually a real calculated thing or just a branding of whatever an intern thinks look about right?

It is real and they usually the Q cutoff times dead right so it seems like the machines has learned to predict pretty well when the situations are fairly static (If newSofts then...). That makes it even more baffling that  Mercedes did not see what they needed to do at the same time as McLaren did. I mean, with a predictive software, taking into account how much faster a new tire is and how much speed they lose with 5 laps of extra degradation (even with optimistic calculation), there was not really any question marks there.