Does the pp system work?

[Polsixe]

Good graph, although it's not necessarily a power relationship. There's only a few points at the end that make it look that way, if those are outliers then it could just as easily be linear. Or it could be linear within each tyre type, and the different slopes make it look like a power curve.

Stunning work though, @Polsixe and @eran0004.

If I was Polyphony, I'd have an AI program to run something like this for every car, on every tyre type, for every combination of upgrades. We can see above that the correlation is pretty good, but there are definitely some cars that are significantly slower than their PP should seem to indicate. Being able to pick those out and hand tweak the PP would be very handy if you were Polyphony, and on a graph of every car in the game the outliers should stick out like a sore thumb.

Then there's the issue of cornering, which is trickier, but I definitely think that standing mile testing could be automated productively.

Oh it is a curve, has to be. A car with 0 PP will have an infinite time , 100 PP may be five minutes. The other side even the most powerful PP cars will have a converging minimum time.

As long as they match a mathematical equation that is considered "linear".

 

[Polsixe]

The scale and size of the dots is deceiving. I can see gaps of what looks like 4-5 seconds over 30-40 seconds. That's as much as 15% difference in times or translated into lap times, 9 seconds in a 1 minute lap. Even a gap of 1 second over 30 seconds, translates into 2 seconds in a one minute lap and 4 seconds in a two minute lap.

Look, I didn't invent PP and don't have to defend it to the n-th degree. I suggest anyone can see that it is not random and does provide a good reasonable single measure of a car's performance. Do some reading on "regression analysis" and curve fitting. Life just doesn't give the exact relationships you seem to be expecting from a video game. In fact by building in the uncertainty that upsets you, PD has made a more realistic game.

 

[Imari]

Oh it is a curve, has to be. A car with 0 PP will have an infinite time , 100 PP may be five minutes. The other side even the most powerful PP cars will have a converging minimum time.

As long as they match a mathematical equation that is considered "linear".

Firstly, a quick point on terminology. Either the equation is linear, or it describes a curve. Not both.

Secondly, it doesn't have to be a curve. It would make sense for it to be, but I can think of at least one way in which it wouldn't need to be. It's bad science to simply assume that what makes sense to you is the way things are. Gather your data, and when you have enough the answer will be self-explanatory. Bad things happen if you make assumptions too early, it leads to you potentially overlooking things that don't fit your picture of how it "should" go together.

Look, I didn't invent PP and don't have to defend it to the n-th degree. I suggest anyone can see that it is not random and does provide a good reasonable single measure of a car's
performance.

I'd suggest that you don't defend it at all, and simply let the data speak for itself.

However, while anyone with eyes in their face can see that there's definitely a relationship of some kind going on there, it's entirely subjective whether it's a good single measure of a car's performance.

I'd suggest a good starting point for figuring that out would be to compare it to the most basic single value performance stat, power to weight. If you want to post up the data and cars you used to generate your graph, I'll make a power-to-weight vs. PP graph and we can compare to see whether PP is a better fit to the actual measured performance.

In fact by building in the uncertainty that upsets you, PD has made a more realistic game.

Meh. They didn't build uncertainty into the equation, if there's uncertainty then it's there because there are variables that affect performance that aren't being taken into account.

Given that the whole purpose of PP is to performance match, and that there's no such thing as real life PP, then realism doesn't really come into it. The whole measure of whether PP works or not is simply whether it provides a decent way of equalising performance.

In real life, performance equalisation tends to either be restricting cars/parts to use identical or near identical components, or it's an iterative process where overperforming cars are given some sort of handicap, or both. F1 restricts components. Super GT adds weight to cars that win. And so on.

It doesn't matter how the system works, as long as it produces close racing. PP arguably doesn't in some cases. Which is sort of to be expected with 1200+ cars, but I also sort of expected them to take the player data that they undoubtedly have from two games and make some manual corrections on a per car basis. At least to the major outliers. 9 seconds gap per minute is too much.

 

[Sick Cylinder]

Although the PP system could definitely be improved, when combined with weight and power limits it is a useful tool for Hosts. We don't have many options open to us, but it can work well. It would be nice if it could be improved and also if Hosts could be given more options - a top speed limiter for example would be a helpful additional tool.

 

[UnkaD]

I just happened across this thread, and it's an interesting discussion. I don't really like the PP system as I do see huge variations. A few nights ago I was in a PP lobby on High Speed Ring. Half of the lobby were running Shelby Series Ones and were all 3-4 seconds faster a lap than I could manage. The rest of the field, on the other hand, couldn't keep up with me. So I was running five-lap races all by myself. Not fun at all. This is ALWAYS what I experience in PP lobbies.

To be honest, I get better races in my lobbies with almost no restrictions in that the people who come in just naturally try to match performance with others. A simple power limit of about 1000 or 1100 hp and weight around a minimum of 750 kg seems to keep all of the crazy "ME!ME!ME!" guys out. I drive whatever I want and tell others to do the same. Interestingly enough, I haven't seen anyone pull out a car close to the limits I set.

 

[Polsixe]

Firstly, a quick point on terminology. Either the equation is linear, or it describes a curve. Not both.

Secondly, it doesn't have to be a curve. It would make sense for it to be, but I can think of at least one way in which it wouldn't need to be. It's bad science to simply assume that what makes sense to you is the way things are. Gather your data, and when you have enough the answer will be self-explanatory. Bad things happen if you make assumptions too early, it leads to you potentially overlooking things that don't fit your picture of how it "should" go together.



I'd suggest that you don't defend it at all, and simply let the data speak for itself.

However, while anyone with eyes in their face can see that there's definitely a relationship of some kind going on there, it's entirely subjective whether it's a good single measure of a car's performance.

I'd suggest a good starting point for figuring that out would be to compare it to the most basic single value performance stat, power to weight. If you want to post up the data and cars you used to generate your graph, I'll make a power-to-weight vs. PP graph and we can compare to see whether PP is a better fit to the actual measured performance.



Meh. They didn't build uncertainty into the equation, if there's uncertainty then it's there because there are variables that affect performance that aren't being taken into account.

Given that the whole purpose of PP is to performance match, and that there's no such thing as real life PP, then realism doesn't really come into it. The whole measure of whether PP works or not is simply whether it provides a decent way of equalising performance.

In real life, performance equalisation tends to either be restricting cars/parts to use identical or near identical components, or it's an iterative process where overperforming cars are given some sort of handicap, or both. F1 restricts components. Super GT adds weight to cars that win. And so on.

It doesn't matter how the system works, as long as it produces close racing. PP arguably doesn't in some cases. Which is sort of to be expected with 1200+ cars, but I also sort of expected them to take the player data that they undoubtedly have from two games and make some manual corrections on a per car basis. At least to the major outliers. 9 seconds gap per minute is too much.

Since you asked, see plot of all cars HP/kg versus PP. You're right, let the data speak for itself. PP is a HP/kg function, with some other variables thrown in. If it was me I would put some height and wheelbase factors into it, hello PD? I am puzzled why people expect a strictly defined goof proof x-y relationship to help them decide why a 500 PP GSX may be better than a 500 PP Tank Car. By the way the 4 second difference in the standing mile results posted above are a 483 PP Fisker Karma and a 484 PP '69 Corvette L46. Which one do you think is 32.5 secs and which is 36.5 sec?

Interestingly, only tested a little over 100 vehicles but you "need" 519 PP to break 30 seconds. 580 PP to break 26 seconds. 440 PP to break 35 seconds. The charts on the Suzuka and Midfield threads show the same kind of curves and performance bands.

 

[Imari]

I am puzzled why people expect a strictly defined goof proof x-y relationship to help them decide why a 500 PP GSX may be better than a 500 PP Tank Car.

The point of a performance metric is to be able to have cars with comparable performance without having to exhaustively test them yourself. If a 500PP GSX isn't comparable with a 500PP Tank Car, then the PP rating is no use if you want to race them together fairly.

People expect it because that's what PP is supposed to be able to do. Mostly it's kind of OK, but we all know that there are some cars that are well above the curve and some that are well below.

By the way the 4 second difference in the standing mile results posted above are a 483 PP Fisker Karma and a 484 PP '69 Corvette L46. Which one do you think is 32.5 secs and which is 36.5 sec?

I think it doesn't matter and you're just trying to score points off me. What if I just say congratulations, you're right?

It doesn't matter which is which, two cars with essentially the same PP and a 4 second difference is a problem.

Depending on how they designed the PP system, it may be more or less of a problem. As mostly a road racing game, PP likely gives some weight to cornering characteristics that won't show up much (or at all) in a pure drag test. But still, it's not great. This is not a difference that is surmountable by skill.

Interestingly, only tested a little over 100 vehicles but you "need" 519 PP to break 30 seconds. 580 PP to break 26 seconds. 440 PP to break 35 seconds. The charts on the Suzuka and Midfield threads show the same kind of curves and performance bands.

For my own future reference, and everyone else's:

https://www.gtplanet.net/forum/threads/suzuka-lap-times-for-640-cars.314470/page-4#post-9960509

https://www.gtplanet.net/forum/thre...mes-for-714-cars.325386/page-12#post-10703303

And of course there is. At any given PP, there's going to be a slowest car, and so that is the PP you need to break a certain time. That's just how curves work. PP is the same on all tracks, so it should work similarly everywhere.

I think you're reading into things that aren't there. This doesn't tell you anything other than "there is a relationship", which we'd already established.

What I find interesting is how the relationship pretty much totally breaks down below about 300PP. The variance just becomes enormous. It's not helped by the fact that there's not very many data points, but whatever the system is it clearly doesn't work well at low levels.

Edit: Which is highly indicative of a polynomial relationship now that I come to think of it, and exactly why you don't want to use that. Because unless you jiggle it so that you're using a relatively flat section of the polynomial, you get parts where the curve is so steep that it just does wacky things.

 

[Johnnypenso]

I am puzzled why people expect a strictly defined goof proof x-y relationship to help them decide why a 500 PP GSX may be better than a 500 PP Tank Car.

Strawman argument. What I believe most people want is the ability to race cars that are similar against each other, competitively. A 91' NSX vs. a 91' RX7 vs. a 90's Corvette vs. a 90's TVR. All similar power/weight stock or easily made so, same era, same tires etc. If the system works well, those cars should be fairly close to each other on most average sized tracks. They aren't. The GSX is an obvious case of an outlier that was never addressed. It blows everything away in it's PP class. PD has access to the TT results from GT5, so why would they not tweak this and other "outlier" cars, if for no other reason than to add variety to PP governed TT's and Quick Races?

 

[Polsixe]

Strawman argument. What I believe most people want is the ability to race cars that are similar against each other, competitively. A 91' NSX vs. a 91' RX7 vs. a 90's Corvette vs. a 90's TVR. All similar power/weight stock or easily made so, same era, same tires etc. If the system works well, those cars should be fairly close to each other on most average sized tracks. They aren't. The GSX is an obvious case of an outlier that was never addressed. It blows everything away in it's PP class. PD has access to the TT results from GT5, so why would they not tweak this and other "outlier" cars, if for no other reason than to add variety to PP governed TT's and Quick Races?

I guess they wanted and chose an objective formula with no individual tweaks, and has to be a function of available in game parameters. The PP does exactly what you want, maybe not the four cars you have listed.

 

[Polsixe]

The point of a performance metric is to be able to have cars with comparable performance without having to exhaustively test them yourself. If a 500PP GSX isn't comparable with a 500PP Tank Car, then the PP rating is no use if you want to race them together fairly.

People expect it because that's what PP is supposed to be able to do. Mostly it's kind of OK, but we all know that there are some cars that are well above the curve and some that are well below.



I think it doesn't matter and you're just trying to score points off me. What if I just say congratulations, you're right?

It doesn't matter which is which, two cars with essentially the same PP and a 4 second difference is a problem.

Depending on how they designed the PP system, it may be more or less of a problem. As mostly a road racing game, PP likely gives some weight to cornering characteristics that won't show up much (or at all) in a pure drag test. But still, it's not great. This is not a difference that is surmountable by skill.



For my own future reference, and everyone else's:

https://www.gtplanet.net/forum/threads/suzuka-lap-times-for-640-cars.314470/page-4#post-9960509

https://www.gtplanet.net/forum/thre...mes-for-714-cars.325386/page-12#post-10703303

And of course there is. At any given PP, there's going to be a slowest car, and so that is the PP you need to break a certain time. That's just how curves work. PP is the same on all tracks, so it should work similarly everywhere.

I think you're reading into things that aren't there. This doesn't tell you anything other than "there is a relationship", which we'd already established.

What I find interesting is how the relationship pretty much totally breaks down below about 300PP. The variance just becomes enormous. It's not helped by the fact that there's not very many data points, but whatever the system is it clearly doesn't work well at low levels.

Edit: Which is highly indicative of a polynomial relationship now that I come to think of it, and exactly why you don't want to use that. Because unless you jiggle it so that you're using a relatively flat section of the polynomial, you get parts where the curve is so steep that it just does wacky things.

It may be a 2nd or 3rd order polynomial if you do a full mean square differences but the power curve looked the best. So we are all agreed there is a relationship it's just that some of us expect a thinner line?

 

[Patruck GT2]

Strawman argument. What I believe most people want is the ability to race cars that are similar against each other, competitively. A 91' NSX vs. a 91' RX7 vs. a 90's Corvette vs. a 90's TVR. All similar power/weight stock or easily made so, same era, same tires etc. If the system works well, those cars should be fairly close to each other on most average sized tracks. They aren't. The GSX is an obvious case of an outlier that was never addressed. It blows everything away in it's PP class. PD has access to the TT results from GT5, so why would they not tweak this and other "outlier" cars, if for no other reason than to add variety to PP governed TT's and Quick Races?

Yes. That's exactly the point. These days I tried the "seasonal" events after months not turning the PS3 on. It works terribly. I mean, it's frustrating to tune up a car up to a given pp value, expecting it to be competitive for a certain event, when suddenly you realize your car is the "wrong" one and that for some reason its pp is overrated, and you have to choose another one.
My finely LSD-TorqueConverter tuned 550pp Mitsubishi GTO is slower, by any means, than my not-so-finely tuned Alfa 8C, which has got less than 500pp. Not to talk about my TVR Speed 6, whose 500pp are enough to destroy the Alfa. And note that the GTO is really good. I mean, I spent half an hour tweaking the torque distribution between rear and front, and the differentials accordingly. It doesn't understeer, doesn't oversteer. The 8C does understeer horribly, but it's still faster. With 50 pp less.

 

[Polsixe]

Yes. That's exactly the point. These days I tried the "seasonal" events after months not turning the PS3 on. It works terribly. I mean, it's frustrating to tune up a car up to a given pp value, expecting it to be competitive for a certain event, when suddenly you realize your car is the "wrong" one and that for some reason its pp is overrated, and you have to choose another one.
My finely LSD-TorqueConverter tuned 550pp Mitsubishi GTO is slower, by any means, than my not-so-finely tuned Alfa 8C, which has got less than 500pp. Not to talk about my TVR Speed 6, whose 500pp are enough to destroy the Alfa. And note that the GTO is really good. I mean, I spent half an hour tweaking the torque distribution between rear and front, and the differentials accordingly. It doesn't understeer, doesn't oversteer. The 8C does understeer horribly, but it's still faster. With 50 pp less.

Ok, I will take up your challenge to see if it can be reproduced. If I read it correctly you have taken a Mitsubishi GTO that normally has 450 PP or less and even tuned to 550PP is slower than a 500PP Alfa 8C. And maybe @Johnnypenso or anyone else can do the same.

 

[Polsixe]

here we go, 500PP test of 8C and GTO

baseline Alfa Romeo 8C Competizione
423 HP
1475 kg
50:50 weight
3.48 kg/hp
1894*1341*4381
500 PP
CS tires

31.144 seconds Standing Mile
44.038 Motegi Oval
1:32.449 Willow Springs
-------------------------------------------------
Stock Mitsubishi GTO Twin a Turbo '96
313 HP
1710 kg
59:41 weight
5.46 kg/HP
1840*1285*4590
450 PP
CS tires

35.116 seconds Standing Mile
48.009 Motegi Oval
1:38.426 Willow Springs
------------------------------------------------
Tuned GTO
447 HP
1710 kg
3.82 kg/hp
500 PP
CS tires, racing suspension, full custom gearbox, LSD, Centre Diff, clutch 2.

30.912 seconds standing mile, 169 mph
45.25 Motegi Oval
1:33.863 Willow Springs
----------------------------------------------------

So that is heavier, front heavy, slower than 8C on a track, but not too far away.

Added Stage 1 lightning, ballast at -50 to bring back to 500 PP and 53:47 weight.
1:32.498 seconds Willow Springs, now only .049 secs slower than 8C

Willow Springs cross-check. Fastest Laps.com has a BMW M3 at 425 HP/1520 kg running a 1:32.5. This is a 503 PP car in GT6. QED, this a good representative time for 500PP cars.
-------------------------------------------------------
So I don't even have to try a 550 PP GTO to see that it can match the 8C with a bit of tuning at the 500 PP level.
That stock 59:41 weight seems to hurt. I didn't even own a GT6 GTO as even in GT5 the GTO/3000 twins were dogs. I don't know why you would have problems competing with a 550 PP GTO.

 

[DolHaus]

here we go, 500PP test of 8C and GTO

baseline Alfa Romeo 8C Competizione
423 HP
1475 kg
50:50 weight
3.48 kg/hp
1894*1341*4381
500 PP
CS tires

31.144 seconds Standing Mile
44.038 Motegi Oval
1:32.449 Willow Springs
-------------------------------------------------
Stock Mitsubishi GTO Twin a Turbo '96
313 HP
1710 kg
59:41 weight
5.46 kg/HP
1840*1285*4590
450 PP
CS tires

35.116 seconds Standing Mile
48.009 Motegi Oval
1:38.426 Willow Springs
------------------------------------------------
Tuned GTO
447 HP
1710 kg
3.82 kg/hp
500 PP
CS tires, racing suspension, full custom gearbox, LSD, Centre Diff, clutch 2.

30.912 seconds standing mile, 169 mph
45.25 Motegi Oval
1:33.863 Willow Springs
----------------------------------------------------

So that is heavier, front heavy, slower than 8C on a track, but not too far away.

Added Stage 1 lightning, ballast at -50 to bring back to 500 PP and 53:47 weight.
1:32.498 seconds Willow Springs, now only .049 secs slower than 8C

Willow Springs cross-check. Fastest Laps.com has a BMW M3 at 425 HP/1520 kg running a 1:32.5. This is a 503 PP car in GT6. QED, this a good representative time for 500PP cars.
-------------------------------------------------------
So I don't even have to try a 550 PP GTO to see that it can match the 8C with a bit of tuning at the 500 PP level.
That stock 59:41 weight seems to hurt. I didn't even own a GT6 GTO as even in GT5 the GTO/3000 twins were dogs. I don't know why you would have problems competing with a 550 PP GTO.

When I tune a car I expect to drop around 2 seconds over a 1:30ish lap, you have tuned the GTO and gotten the expected improvement but your still behind a stock car of the same rating. If the 8C was tuned to a similar degree you'd be 2 seconds behind again. Therefore the rating system is not functioning correctly

Go look through the FITT archives and check the results, in events where we multiple cars competing you will see huge differences between cars that are otherwise similar in terms of stats/design. Our data is sourced from the average best lap times of multiple drivers and still the gap is as clear as day which rules out driver bias. The system just doesn't work, cars rated as equal aren't even close

 

[Patruck GT2]

Today I had some tragic attempts at the 600pp seasonals. These further convinced me the system doesn't make sense. How can a 308Hp Lotus Elise, with 5 gears and no LSD, compete with a bunch of GT1 cars? As a sporadic player I would have supposed the best I could get for 600pp was something like an Audi R8 racing-tuned. I bought it only to discover the competition would be really though, while relatively easy with a same-rated LM car by Nissan (don't remember the name).
I mean, they can't seriously give the same rating to a car designed from scratch to be raced and to one developed from an ordinary production one.

Back in GT2, you had a power limit for each event. As simple as that. Did it work better than PP? No. Did it work much worse? Not either. Was it simpler to understand? God, it was.
I am not saying that developing a "scale" for vehicle general performance is easy, indeed I think it's extremely difficult. But since it is so difficult, I would have understood PD if they didn't even try. The PP rating, on the other side, seems to me a completely non competent attempt at this difficult task. And the fact I have to withstand it to enjoy the game makes me mad.

 

[Polsixe]

Results spreadsheet @ Google Drive

Super Muscle Lap Time Winners

• @Motor City Hami - 1'33.765"
• @MrGrado - 1'34.530"
• @iainoflo85 - 1'34.903"

Standard Classics Lap Time Winners

• @NEWDRIVER2 - 1'39.643"
• @Motor City Hami - 1'39.650"
• @Otaliema - 1'39.675"

Super Muscle Driver's Choice Winners

• @Motor City Hami - 10
• @ACSR421 - 9.5
• @MrGrado - 9.25

Standard Classics Driver's Choice Winners

• @NEWDRIVER2 - 8.625
• @Motor City Hami - 7.875
• @PKS 80 - 7.5

So the FITT group sets some parameters and attempts to see who can get the best out of cars. In the example above the winners are pretty close results, I reckon if they swapped cars the placings would change, "he can take his'n and beat your'n and take your'n and best his'n". I would expect the FITT results to converge for any PP , with a number of like-minded individuals competing independently.

Tuning is not well represented in PP as we know only a few variables seem to affect PP - power, weight, weight distribution, oil age. The suspension, transmission, aero mods don't affect PP and that seems to be main complaint.

Out of 10 cars at same PP two will be rabbits, two will be tortoises and six will be a pack of hounds. Pick your rabbits wisely.

 

[Johnnypenso]

Today I had some tragic attempts at the 600pp seasonals. These further convinced me the system doesn't make sense. How can a 308Hp Lotus Elise, with 5 gears and no LSD, compete with a bunch of GT1 cars? As a sporadic player I would have supposed the best I could get for 600pp was something like an Audi R8 racing-tuned. I bought it only to discover the competition would be really though, while relatively easy with a same-rated LM car by Nissan (don't remember the name).
I mean, they can't seriously give the same rating to a car designed from scratch to be raced and to one developed from an ordinary production one.

Back in GT2, you had a power limit for each event. As simple as that. Did it work better than PP? No. Did it work much worse? Not either. Was it simpler to understand? God, it was.
I am not saying that developing a "scale" for vehicle general performance is easy, indeed I think it's extremely difficult. But since it is so difficult, I would have understood PD if they didn't even try. The PP rating, on the other side, seems to me a completely non competent attempt at this difficult task. And the fact I have to withstand it to enjoy the game makes me mad.

That's an issue that's more related to the AI, rubberbanding and the generally weak offline racing in GT6.

So the FITT group sets some parameters and attempts to see who can get the best out of cars. In the example above the winners are pretty close results, I reckon if they swapped cars the placings would change, "he can take his'n and beat your'n and take your'n and best his'n". I would expect the FITT results to converge for any PP , with a number of like-minded individuals competing independently.

Tuning is not well represented in PP as we know only a few variables seem to affect PP - power, weight, weight distribution, oil age. The suspension, transmission, aero mods don't affect PP and that seems to be main complaint.

Out of 10 cars at same PP two will be rabbits, two will be tortoises and six will be a pack of hounds. Pick your rabbits wisely.

Cherrypicking results will do that. When you don't cherrypick things look completely different. No one is saying you can't cherrypick a few cars and get them close together, but the PP system is supposed to work on a wide range of cars. This is your cherrypicked result:

Super Muscle Lap Time Winners

• @Motor City Hami - 1'33.765"
• @MrGrado - 1'34.530"
• @iainoflo85 - 1'34.903"

Here is the full results:

As anyone can clearly see, once you get past the cherrypicked results from the top of the standings, there is a massive difference in lap times, as much as 4.5 seconds. The biggest fail IMO, is in the part you didn't cherry pick, the bottom 4 cars on the list excluding the GT40. All are similar era muscle cars, relatively similar specs, same engine layout and front heavy design, relatively primitive tire and suspension designs. If the system is working, those 4 cars should be very close on the track but their lap times are as follows:

1:35.474
1:36.722
1:37.390
1:38.207

I did extensive testing in GT5 and for me, 1pp = .05 seconds on the track. In other words 50pp = 2.5 seconds. This was on Deep Forest, a relatively short track. Look at it another way, 1 second = 20pp. So if you take those rough numbers and use 1:35.474 as your base PP number the other times line up as follows:

Base PP
2nd car - 25pp
3rd car - 39pp
4th car - 55pp

That's an average error of 40 pp among the three slower cars. That's huge!! I don't see how anyone can objectively look at the evidence and think the system is even close to accurate other than when you cherrypick data to support your cause.

 

[DolHaus]

So the FITT group sets some parameters and attempts to see who can get the best out of cars. In the example above the winners are pretty close results, I reckon if they swapped cars the placings would change, "he can take his'n and beat your'n and take your'n and best his'n". I would expect the FITT results to converge for any PP , with a number of like-minded individuals competing independently.

Tuning is not well represented in PP as we know only a few variables seem to affect PP - power, weight, weight distribution, oil age. The suspension, transmission, aero mods don't affect PP and that seems to be main complaint.

Out of 10 cars at same PP two will be rabbits, two will be tortoises and six will be a pack of hounds. Pick your rabbits wisely.

We've been doing this a long time and the difference is never truly overcome by tuning ability, many top draw tuners have tried to pull an underdog into contention and while they might pick off a few poorly tuned cars they'll never get close to equal performance. Every contest we try to put in multiple car choices but we know that there will always be one car that is just unbeatable and if you want to be in contention for a lap time victory then you've got to be in that car. We put in a lot of time trying to find cars that fit together in terms of theme and performance but it is impossible, the system just doesn't work

 

[Polsixe]

That's an issue that's more related to the AI, rubberbanding and the generally weak offline racing in GT6.

Cherrypicking results will do that. When you don't cherrypick things look completely different. No one is saying you can't cherrypick a few cars and get them close together, but the PP system is supposed to work on a wide range of cars. This is your cherrypicked result:



Here is the full results:

View attachment 491661

As anyone can clearly see, once you get past the cherrypicked results from the top of the standings, there is a massive difference in lap times, as much as 4.5 seconds. The biggest fail IMO, is in the part you didn't cherry pick, the bottom 4 cars on the list excluding the GT40. All are similar era muscle cars, relatively similar specs, same engine layout and front heavy design, relatively primitive tire and suspension designs. If the system is working, those 4 cars should be very close on the track but their lap times are as follows:

1:35.474
1:36.722
1:37.390
1:38.207

I did extensive testing in GT5 and for me, 1pp = .05 seconds on the track. In other words 50pp = 2.5 seconds. This was on Deep Forest, a relatively short track. Look at it another way, 1 second = 20pp. So if you take those rough numbers and use 1:35.474 as your base PP number the other times line up as follows:

Base PP
2nd car - 25pp
3rd car - 39pp
4th car - 55pp

That's an average error of 40 pp among the three slower cars. That's huge!! I don't see how anyone can objectively look at the evidence and think the system is even close to accurate other than when you cherrypick data to support your cause.

I agree discount GT40 as you don't need a PP system to tell you that is better than a Camaro. No cherry picking involved, the FITT results archive were openly presented by @DolHaus

Dig a little deeper in your FITT data, as your post numbers are averages. Granted there were only two testers for the super muscle cars and four for the classic.

Your best car Camaro had times of 94.935 and 96.013.

A Ford Mustang '71 had 96.147 and 97.297. The low end of one almost overlaps the high end of the other, it all comes down to the driver. Sure three seconds is huge but .134 seconds is a slightly missed apex difference. You can even argue an average of two guys is not a large enough sample.

In the Standard Classic series with four testers you'll see the same overlaps of high and low times. Discount the Superbird '70 as an obvious tortoise, the one slow tester, and the rest of the cars are very closely bunched at 98.5 to 99.5 seconds. If you can't see why the Superbird is not competitive with a smaller car at Laguna Seca than no PP system will help. Roger Penske and Parnelli Jones never tried to run Superbirds in Trans-Am. But, I have a Charger 440 tune that will do low 1:38s in case anyone's interested.

FITT, thanks you have proven PP works!​

 

[Polsixe]

We've been doing this a long time and the difference is never truly overcome by tuning ability, many top draw tuners have tried to pull an underdog into contention and while they might pick off a few poorly tuned cars they'll never get close to equal performance. Every contest we try to put in multiple car choices but we know that there will always be one car that is just unbeatable and if you want to be in contention for a lap time victory then you've got to be in that car. We put in a lot of time trying to find cars that fit together in terms of theme and performance but it is impossible, the system just doesn't work

I suspect there will be always be obvious "other reason" why you can't make a silk purse out of a particular sow's ear and vice versa why one or two cars will out perform others. Visual clues will help, MR versus FR and FF, all that is open to anyone picking their horses for courses.

 

[DolHaus]

I suspect there will be always be obvious "other reason" why you can't make a silk purse out of a particular sow's ear and vice versa why one or two cars will out perform others. Visual clues will help, MR versus FR and FF, all that is open to anyone picking their horses for courses.

You know, I never thought to consider those things. Man, don't I feel foolish, its all so obvious now I think about it...

 

[Johnnypenso]

I agree discount GT40 as you don't need a PP system to tell you that is better than a Camaro. No cherry picking involved, the FITT results archive were openly presented by @DolHaus

Dig a little deeper in your FITT data, as your post numbers are averages. Granted there were only two testers for the super muscle cars and four for the classic.

Your best car Camaro had times of 94.935 and 96.013.

A Ford Mustang '71 had 96.147 and 97.297. The low end of one almost overlaps the high end of the other, it all comes down to the driver. Sure three seconds is huge but .134 seconds is a slightly missed apex difference. You can even argue an average of two guys is not a large enough sample.

In the Standard Classic series with four testers you'll see the same overlaps of high and low times. Discount the Superbird '70 as an obvious tortoise, the one slow tester, and the rest of the cars are very closely bunched at 98.5 to 99.5 seconds. If you can't see why the Superbird is not competitive with a smaller car at Laguna Seca than no PP system will help. Roger Penske and Parnelli Jones never tried to run Superbirds in Trans-Am. But, I have a Charger 440 tune that will do low 1:38s in case anyone's interested.

FITT, thanks you have proven PP works!​

I have a feeling I could post up any data at all, showing cars 5-10-15 seconds apart and your answer would be the same every time. Excuse, discount this car, excuse, discount that car, excuse, oh there is this exception, excuse...etc...wilfuly ignoring the facts and then concluding that everything is fine and works as it should.

 

[synon]

As anyone can clearly see, once you get past the cherrypicked results from the top of the standings, there is a massive difference in lap times, as much as 4.5 seconds. The biggest fail IMO, is in the part you didn't cherry pick, the bottom 4 cars on the list excluding the GT40. All are similar era muscle cars, relatively similar specs, same engine layout and front heavy design, relatively primitive tire and suspension designs. If the system is working, those 4 cars should be very close on the track but their lap times are as follows:

1:35.474
1:36.722
1:37.390
1:38.207

I did extensive testing in GT5 and for me, 1pp = .05 seconds on the track. In other words 50pp = 2.5 seconds. This was on Deep Forest, a relatively short track. Look at it another way, 1 second = 20pp. So if you take those rough numbers and use 1:35.474 as your base PP number the other times line up as follows:

Base PP
2nd car - 25pp
3rd car - 39pp
4th car - 55pp

That's an average error of 40 pp among the three slower cars. That's huge!! I don't see how anyone can objectively look at the evidence and think the system is even close to accurate other than when you cherrypick data to support your cause.

If you look at the individual times eg the Corvette Convertible and the Firebird Trans Am, one tester's times are 98.293 & 98.511 and the other's are 98.121 & 96.268, which is a difference of over 2 seconds, so you can't really draw that conclusion.

Back in GT2, you had a power limit for each event. As simple as that. Did it work better than PP? No. Did it work much worse? Not either. Was it simpler to understand? God, it was.
I am not saying that developing a "scale" for vehicle general performance is easy, indeed I think it's extremely difficult. But since it is so difficult, I would have understood PD if they didn't even try. The PP rating, on the other side, seems to me a completely non competent attempt at this difficult task. And the fact I have to withstand it to enjoy the game makes me mad.

PP is going to be a more accurate gauge of performance though.

 

[Polsixe]

I have a feeling I could post up any data at all, showing cars 5-10-15 seconds apart and your answer would be the same every time. Excuse, discount this car, excuse, discount that car, excuse, oh there is this exception, excuse...etc...wilfuly ignoring the facts and then concluding that everything is fine and works as it should.

You discounted the GT40 times yourself, why can't I then discount the Superbird? And yes, the driver's times had more to with the spread than the car's PP. This is the FITT results, in a small group of cars, one will be very fast, one will be very slow, the rest even. QED.

 

[Johnnypenso]

You discounted the GT40 times yourself, why can't I then discount the Superbird? And yes, the driver's times had more to with the spread than the car's PP. This is the FITT results, in a small group of cars, one will be very fast, one will be very slow, the rest even. QED.

If you are going to post up data for us to look at, you don't just cherry pick to remove all the anomolies that disprove your case and after removing all the data you don't like, conclude that all is well. That's not how it works.

 

[DolHaus]

@Polsixe if you were to put together a means of proving that the system works how would you do it? Lets skip past the semantics and just put it to the test

 

[MrGrado]

PP system not broken enough to effect the grid selection or race outcome during recent online racing.

In fact I had extremely close racing with a vastly "mismatched" car but equal for lap speed thanks to PP system.

 

[Johnnypenso]

PP system not broken enough to effect the grid selection or race outcome during recent online racing.

In fact I had extremely close racing with a vastly "mismatched" car but equal for lap speed thanks to PP system.

/thread

 

[Polsixe]

If you are going to post up data for us to look at, you don't just cherry pick to remove all the anomolies that disprove your case and after removing all the data you don't like, conclude that all is well. That's not how it works.

Oh c'mon now. Number 1, as you know it's not my data it is a FITT data link presented as-is, Number 2 removing extreme outlier data from a set is perfectly acceptable, as you yourself excluded the GT40, taking the outlier from the other (slow) side is not cherry picking. Now in the Standard Classic series observing driver's times another outlier is noticed.

Number 3 you seem to discount driver's skill as an independent variable upon which the laptimes are dependent. This random sample of FITT data shows the driver skill has more to do with the time difference than PP.

I get it, you established a position and now refuse to look at a contrary view despite the independent evidence. There's another rhetorical term for that which you probably know.

 

[Polsixe]

@Polsixe if you were to put together a means of proving that the system works how would you do it? Lets skip past the semantics and just put it to the test

Well you FITT guys were on the right track even though demonstrating PP was not your aim, your group presented test data which illustrates a number of factors. Removing or limiting the driver variability or being careful with averaging is useful. The exhaustive solo tests by people such as SuzukaStar and Midfield Maven have already shown the good PP correlations, using constant test methodology and standard conditions. Just open your eyes a bit more and enough with the knee jerk denials. I hope you're not so stubborn in your work-a-day life.