Effect on Tyres,Horsepower, AND weight on A-Spec Points: a 'Scientific' Study Updated

  • Thread starter scraps
  • 40 comments
  • 4,157 views
S

scraps

11
I have wasted a bit of time in the last couple of days seeing what effect changing tyres and horsepower has on A-Spec points. It has been rumoured that it was somewhat based on the Power to weight ratio, so i set out to see if this is indeed the case.

So far i have only thoroughly tested one car, in one series. It is the Mitsubishi Mirage Cyborg 1600 RZ '97. I bought it used, and with old oil it has 158ps stock, and weighs in at 1060kg. When entered in race 1 of the one make mirage series, the opposition it comes up against is a mixed field consisting on varying numbers of A: '97 Cyborgs, and B: '78 Mirage 1400GLX, the latter being the 'tougher' car of the two.

If you were to enter the race with an otherwise stock car, on S2 tyres, if the field is made up of 4 Cyborgs, and 1 GLX, you would be awarded 146 A-Spec points. 3 Cyborgs and 2 GLX's, 161 points, 2 Cyborgs and 3 GLX's warrented 175points and finally, 1 Cyborg and 4 GLX's was granted a nice 190pts.

I have done my testing so far primarily with the 'Toughest' opponents available, for both HP and Tyres, but i have also graphed the effects of varying hp with the second hardest opposition. it is marked on my graph as "s2 diff opp".

I had bought every single piece of horsepower tuning equipment for this car, barring the non removable ones, so i have 44 different horsepower values from 158, stock, to 272, being the most i could extract without changing the oil and port and polish etc.

As you can see from the graph, A-spec points decrease LINEARLY with horspower, and changing the tyres. seems to be just a modifer worth the same amount of points in each case.

I have also included a Weight to Power Ratio Graph, for comparison. this is from the same data, and most importantly, you can see that the A-Spec points DO NOT follow the same trend as the WPR graph.

And now, without further ado, i present you my findings so far.

I will plot the effects of weight, by adding ballast in the next couple of days or maybe over the weekend.

Cheers.



 
Great post! It is very informative... I was waiting for this kind of graphics...
Thank you, great job!
I'm waiting for your next graphs...
 
Yeah, nice test. Now we finally get some insight in the A-spec point system. Good graphs. Didn't know that the points don't change when you change from N1 to N3 tyres or so. Thought that the points always changed by changing tyres.

Nice test, keep it going.:)
 
Amazing write up, very informative, undisputablely well tested and graphed, with a concluded statement non can object. Thanks for the information, and puttin the time into endin the discussion once and for all. 👍
If this was a school assignment, you'd get 100%, and a gold star. Unfortuantly, I can't give you a gold star, so heres a smiley sticker instead. :)
 
Fascinating. Great job. It explains a lot.

Some day, somebody out there is going to combine info like this with the effect on points of each car on each track with every possible set of AI competitors, and come up with a long, long equation for the resultant A-Spec points.

(And, of course, some wise-ass will post a response saying something about getting a life...)
 
Zardoz
Fascinating. Great job. It explains a lot.

Some day, somebody out there is going to combine info like this with the effect on points of each car on each track with every possible set of AI competitors, and come up with a long, long equation for the resultant A-Spec points.

(And, of course, some wise-ass will post a response saying something about getting a life...)
Click to expand...

lol,
700 cars X 522 races X 100 tracks X 10 sets of tyres X 4 levels of weight X several thousand combinations of power mods, ballast and downforce settings X lots of AI combinations + any number of weird glitches (Dodge RAM?) = about a million years of maths.

Good luck to anyone who wants to try that, but in this case I'd say that the wise-ass is probably going to be right! :crazy: :crazy: :)
 
Yeah, my original thoughts on replying to this post was something like, "Wow, I finally met someone with more time on his / her hands than I!" :sly: :dopey:

Good job! Even though that is for one car, you could still probably get a rough estimate from that alone by taking a car's starting A-spec points and finding the delta using the graph you posted.

Or maybe you could just muck around with various combinations more quickly. :dopey:



Actually, when I decide to finish my power and torque graph project, you'll all see that I have plenty of time on my hands as well. :crazy: :sly: Maybe that gives me something to do this evening........
 
scraps
I have done my testing so far primarily with the 'Toughest' opponents available, for both HP and Tyres, but i have also graphed the effects of varying hp with the second hardest opposition. it is marked on my graph as "s2 diff opp".

I had bought every single piece of horsepower tuning equipment for this car, barring the non removable ones, so i have 44 different horsepower values from 158, stock, to 272, being the most i could extract without changing the oil and port and polish etc.


I will plot the effects of weight, by adding ballast in the next couple of days or maybe over the weekend.
Click to expand...




I think you've what you've done is fantastic and it's much appreciated. :)

But... and please don't take this the wrong way... a lot it's unnecessary. Don't want you to burn out!


I've thought about the system a bit and have a gathered a great deal of data... and, well, if I might make some suggestions:

Don't think like a scientist, think like a programmer. A programmer will sacrifice complexity for a higher performing approximation any chance they get. I don't know if you've ever programmed before. I'm assuming you probably have? But... think about this as if you were debugging a program. IMO, the keys to debugging are pattern recognition followed by partitioning and isolation. Basically, you want to slowly drill down to the problem isolating as much code as possible to reduce the number of variables you have to deal with at one time. Isolate as many variables as you can individually and then look for simple correlations. Then, and only then do you start to rebuild the puzzle into a larger picture.

Another issue with programming for efficiency, is that approximations will inevitably create slight discontituities. I've seen this a lot. Usually, it's small... but then there's the RAM which is just a bug. Basically, the data you're operating on isn't 100% correct. So be wary of anecdotal evidence that may *seem* to throw valid theories off course...


- Tires, you can determine by observation of the point differences, are a simple addition/subtraction operation. That's why all your lines have the same slope and are equidistant from each other. The number of points by which they differ seems dependent on the car. But the same car will have the same +/- values regardless of the opposing lineup.

One less variable, one less dimension to explore :sly:

- Points saturate at 10p and I'm told at 190p. That's why you get a tail (or change in slope) at either end. Where you don't see the tail is a place where you have a sample that DID NOT cross these thresholds so it "looks" as if it's a straight line, when in fact it's not and is really a sampling error.

- I would propose NOT looking at the overall ASpec points, but rather the point *differences* to determine a correlation. This doesn't mean much when looking at one car, but may mean more with further experiments.

I'm currently operating under the assumption that each car has a *base* ASpec value, something that was pre-calculated and stored on disk. I'm assuming this for two reasons - 1) it makes sense for a programmer 2) it would explain some inconsistencies. If power/weight was calculated on the fly, the inconsistencies are much harder to explain. So... Occam's Razor and all...

With that said, by looking at the point *differences* you may have the ability to isolate the contribution of each mod. For tires it's easy. For DF, it should be easy. For HP, you'd have to figure out raw and pct value changes to HP rather than just the mod itself. Is there consistency? Ballast may be tricky as well (do you have to recalculate some kind of average against the field or is it already isolated?)


If you're interested, I had a few ideas for focused experiments, intended to isolate particular variables, written up on Pg 11 of the 200p thread. I have a couple more written in my notes. The one you have is decent but you already have an variable of two different opponent cars with two different power/weight ratios. That could be eliminated by looking at a "one car" race like the S7 or A3.

Good luck! Look forward to hearing more 👍
 
there are always people out there who take it all that little bit further. respect to Scraps and JDW (JDW i have much respect for already - if you're as committed to driving as you are to working stuff like this out then my laptimes aren't going to trouble you for long!).

Seriously though, if you need some help doing any testing then PM me with what you want done and I'll run them for you, I'll need some stuff to do when I finish my A-spec point maxxing in a day or two.
 
scraps nice job!!! 👍
but I think you’re approaching the problem with the wrong variables.
I think that the a-spec points are based in the maximum grip and acceleration of the car. This are variables that are automatic calculated for every car in the game and it’s a more precise way to differentiate cars.

And there is no way to test my theory because there is no way to calculate the maximum grip of a car in GT4, at least I can’t :dopey:.
 
G25RJ
scraps nice job!!! 👍
but I think you’re approaching the problem with the wrong variables.
I think that the a-spec points are based in the maximum grip and acceleration of the car. This are variables that are automatic calculated for every car in the game and it’s a more precise way to differentiate cars.

And there is no way to test my theory because there is no way to calculate the maximum grip of a car in GT4, at least I can’t :dopey:.
Click to expand...

I don't think that this is really the correct approach, JDW has already shown that A-spec pts:tyre type is a linear relationship and that the points saturate at <10 and >190. Also, it is unlikely that a programmer would spend time calculating additional (invisible) grip values when they already have useable values.

Additionally, accleration is clearly not taken into account as otherwise transmission settings (which can dramatically change the acceleration of a car) would affect A-spec points, and they don't (trust me, i've tested this thoroughly).
 
What a fun way to spend a friday night!

I have now got data for values of Ballast from 200kgs down to -120kg (using weight reduction) in increments of 20kg, for the same race series, with the constant tyre choice of S2 tyres and the added variable of horsepower, to again guage the effect of power to weight on A-Spec points.

I have added another series on the bottom of the graph, it is the total Value of the 320kgs of added ballast, for a particular horsepower.

Now the trend of this series is surprising. One would expect that after having increased the Horsepower significantly, the effect of 320kg of ballast would be less than its effect on a lower horsepower car, ie, a 320kg change in a 158ps car should be rather more valuable, A-spec wise, than a 320kg change on a 272ps car.

I have some sort of theory about the way this all works together, but ill save that for another post.

enjoy.



PS. I have included the A-spec values for the series, +200kg, 0kg (stock) and -120kg, (stage 3 weight reduction(128kg loss) + 8kgs)
 
Just before I call it a night, i think i should also add a further graph.

it shows the Changing Value, in A-spec points, of Removing 20kgs of Ballast. from 200kgs back to 180kgs at horsepower values from 158 to 272 as usual. its not the best, as it only has small values, but it does show that the lower the weight overall, the higher the value of additions, in A-spec points. and should indicate that this calculation to A-spec points, may have some bearing on the weight Percentage of total. or something like that. im rambling now. hopefully someone else can come up with a theory.

 
scraps
What a fun way to spend a friday night!

I have now got data for values of Ballast from 200kgs down to -120kg (using weight reduction) in increments of 20kg, for the same race series, with the constant tyre choice of S2 tyres and the added variable of horsepower, to again guage the effect of power to weight on A-Spec points.

[snip]

Now the trend of this series is surprising. One would expect that after having increased the Horsepower significantly, the effect of 320kg of ballast would be less than its effect on a lower horsepower car, ie, a 320kg change in a 158ps car should be rather more valuable, A-spec wise, than a 320kg change on a 272ps car.

[snip]
Click to expand...

Awesome scraps! 👍

That explains what I saw lastnight. My 200 A-spec AMG Merc Evo II is: max ballast + R1/R2 tyres and a little less downforce. Went to the last race of DTM: Fuji 2005 and added the Stage 3 turbo. My points went from 200 to 90+! I would have had more points by taking off the ballast and putting on sticker tyres. :banghead:
 
in my first post i put up a weight to pwr graph that was curved..

but upon looking again, if i plot a spec points on the y axis, and the pwr ratio on the x, then it follows that linearly and i am wrong to say that it doesnt.

here is a graph showing that for 2 diferrent opposition fields.

 
Now, the problem i am going to have to explain, is that even tho the trend of the graph follows PWR, changing the power and the weight, both have a linear effect on the A-spec point value, but not in the same direction vs PWR

The 'horizontal' lines are when i changed the horsepower, and the 'vertical' ones are when i changed the weight.

And you can see, for any value of PWR, you can have many different A-spec values.

 
a simple linear PWR vs PS graph.



My Theory:

I think each car has a base 'performance' value which is correlated against the opposition field, and from there you get the stock 190points, on top of which is added and subtracted the modifications, as they each seem to be worth the same irrespective of the value of other mods.

eg tyres are worth 8pts from s2 to s1, and hp is worth -1point per hp, and weight is worth ~13points per 100kg

These values may (almost certainly) change from car to car

downforce is another modifier, but i havnt any data on that yet.

what do you think?
 
scraps
a simple linear PWR vs PS graph.



My Theory:

I think each car has a base 'performance' value which is correlated against the opposition field, and from there you get the stock 190points, on top of which is added and subtracted the modifications, as they each seem to be worth the same irrespective of the value of other mods.

eg tyres are worth 8pts from s2 to s1, and hp is worth -1point per hp, and weight is worth ~13points per 100kg

These values may (almost certainly) change from car to car

downforce is another modifier, but i havnt any data on that yet.

what do you think?
Click to expand...



Yeah, that's... pretty much what I said. 👍

I have some suggestions of what to do with this data if you don't mind. I say this because I believe the ultimate goal would be to say "X Mod or X amount HP/kg" will add "X % points". It may not work, but considering all the data seems to be laying itself out linerarly (just +/-) it might.

Downforce changes are likely to be linear. Tires I knew about. But... does it correlate to anything? Are the point additions/subtractions to a base value? If so, they may just layout as static percentage changes.

Wasn't sure about power-to-weight. Thank you for confirming that! ;)

Ideas on what to do with the data... Like I was asking before, please seperate the HP/kg contribution of the mods from the base HP/kg. With this I'm hoping that it will translate from car to car. Starting from a base value for each car will create discontinuities due to sampling error. Without value seperation, these discontinuties will be absorbed into the next round of data and could throw off the findings.

Next, write down the % difference in HP and kg and/or hp/kg of each mod and their point changes (NOTE: throw out changes that cross the 10p and 190p thresholds!).

I'm hoping this will correlate across cars. [cross fingers]


It gets more bizarre...

Although it seems natural to assume that your cars "base" value is compared against your opposing lineups "average" value, I'm not sure that that's the case.

I'm beginning to suspect that you car's "base" value may be given for a particular race and THEN changed according to the opposition... linearly! (that is, good old addition and subtraction and again)

Why?

Consider the following example from the Muscle Car comp where I'll compare 2 cars against the exact SAME three lineups:


'69 Vette - 1490kg/356hp - 4.19 kg/hp vs #1 - 40p, vs #2 - 44p, vs #3 107p
'63 Vette - 1370kg/253hp - 5.42 kg/hp vs #1 - 99p, vs #2 - 103p,vs #3 166p


Notice that, despite the disparity in power-to-weight the difference between #1 and #2 is 4p and #2 to #3 is 63p! :odd:


So, it may be that each race itself has a base value??? Still working that one out... Ahh... getting confused...

There's a simple workable theory here and I'm missing it because my brain's mush.

Help? :boggled:


I have tons more data that I just haven't sifted through? Want it? :dopey: :P
 
do i want more data? sure, but im not quite sure what data i want.

my next avenue will investigating some one make series, where one can be sure that there will an identically stock field, where one can compare another stock car and hopefully, if all 5 AI cars are identical to the one you are driving, then in theory the A-spec value of such a race would be identical to that of ANOTHER race, where all 6 cars were identical.

thoughts?

I have done a fair bit of testing with several different fields in the mirage cup, and so far, all the mods have been worth EXACTLY the same amount of points.

i do beleive that the points available for any race is based on ALL 5 of the AI cars, and their 'performance index' which is what i will call this theory until something better comes along.
 
just read your above post jdw, and i agree with you all the way there, i gotta check the differences betw mods, such as tyres in combination with steadily increasing power etc.

my primary objective was to confirm/dispel the PWR theory, and that has been dealt with i think, so i will now take another direction now.

i have, on the ballast graph, the changing value of the 320kgs of ballast as Hp increases. i will knock up some other ones a little later.
 
scraps
do i want more data? sure, but im not quite sure what data i want.
Click to expand...

Sorry, I'm half kidding there... because it's on all on paper. ;) I've been taking notes throughout the game because I've been preparing an A-Spec FAQ which is mostly done. As I've been actively searching for 200p races, I'll typically compare a 1/2 dozen cars against the same 1-3 lineups. Sometimes, notes on mods. Anyway...


scraps
my next avenue will investigating some one make series, where one can be sure that there will an identically stock field, where one can compare another stock car and hopefully, if all 5 AI cars are identical to the one you are driving, then in theory the A-spec value of such a race would be identical to that of ANOTHER race, where all 6 cars were identical.

thoughts?
Click to expand...

Well... Yeah... I mentioned that. It's in my first post.

It's also here where I referred you to other experiments I had in mind:
https://www.gtplanet.net/forum/showthread.php?t=61761&page=11&pp=20

But, to summarize
1) One make races - the S7 and A3 would be good, because it's important to determine if there is consistency despite differences in the power and PWR of the respective cars.
2) Static lineup races - the Vette and M3 races are good for this. Start with one of each of the stock cars you'll encounter in the lineup. Now play with a couple mods and see if there is consistency. I'll bet there is.
3) Now back to regular races only with ONE lineup. Find maybe 3 cars which offer a range of different base A-Spec point values. (NOTE: you can repeat the same set of lineups provided you reset the console without saving) Play with mods again.

If there's consistency, I think a formula will be easy to create... Well, with the one unknown variable of what the program stored as the initial value for each car.



scraps
I have done a fair bit of testing with several different fields in the mirage cup, and so far, all the mods have been worth EXACTLY the same amount of points.
Click to expand...

I know the tires are... but that's holding true for all the power mods as well? Is it regardless of ballast? Somehow I'm not surprised. That would also fall into the same category as the observation of two cars with distinctly different PWRs against the same lineups providing the same point differences.


scraps
i do beleive that the points available for any race is based on ALL 5 of the AI cars, and their 'performance index' which is what i will call this theory until something better comes along.
Click to expand...


Yes, the game calculates points based on the entire field.


Did you look at my example?

Doesn't it strike you as fishy that, despire having different power-to-weight ratios, the difference in A-Spec points each car experience between lineups was... EXACTLY THE SAME? :odd:

AFAICT, this is *always* the case.
 
This is just an assumption, but would it not be fair to assume that the "base" value is 60pts as this is what you get in a stock car against an identical field (Cobra vs Cobra cup, S7 vs S7 cup, A3 vs A3 cup)?
 
route_66
This is just an assumption, but would it not be fair to assume that the "base" value is 60pts as this is what you get in a stock car against an identical field (Cobra vs Cobra cup, S7 vs S7 cup, A3 vs A3 cup)?
Click to expand...


I believe the A-Spec value for, what the game considers a completely even field (tthe EXACT same cars) is 60p. That's not quite the base value I had in mind. When I mean by base value, is that, each car in the game *may* have some pre-stored value. This value is used to compare against the other car's values to come up with your points.
 
This MIGHT be a really stupid question, but has anyone thought to try and get ahold of PD for an interview of some sort and have THEM explain how all this stuff works? Props to scraps for this amazing piece of research, but if they could just come out and say, "Add a sport chip and subtract 5 pts." so on and so forth..... off to PD in search of some direct answers.... :crazy:
 
jdw
I believe the A-Spec value for, what the game considers a completely even field (tthe EXACT same cars) is 60p. That's not quite the base value I had in mind. When I mean by base value, is that, each car in the game *may* have some pre-stored value. This value is used to compare against the other car's values to come up with your points.
Click to expand...

So your saying there's a base value per car in a database somewhere and when all the other cars are equal then it calculates it to 60.

i.e difference between base values of cars = 0, A-spec points=60
yes?

In that case do you think the value are always worked out like this? The changes do seem to pretty linear (change tires one grade, +/-20pts)
So A-Spec points = difference between base value of cars + difference in mods + 60?
 
route_66
So your saying there's a base value per car in a database somewhere and when all the other cars are equal then it calculates it to 60.

i.e difference between base values of cars = 0, A-spec points=60
yes?
Click to expand...


That is the theory...

Or it could be that the game is comparing your car to a particular race to get a value and then +/- the opposing field.

Car OP Race -/+ (Avg of opposing field) = 60

That's not very good...

route_66
In that case do you think the value are always worked out like this? The changes do seem to pretty linear (change tires one grade, +/-20pts)
So A-Spec points = difference between base value of cars + difference in mods + 60?
Click to expand...


Figuring out what the equation looks like is probably the tricky part. :boggled:

Because... it's tempting to just assume... well, if the program used an average of the base value of the competitor cars along with your car to determine ASpec points, you could start with the single car races as a base...

e.g.

6x = 60. x = 10.

But that would make just about... zero... sense

:grumpy:
 
Axolot2
This MIGHT be a really stupid question, but has anyone thought to try and get ahold of PD for an interview of some sort and have THEM explain how all this stuff works? Props to scraps for this amazing piece of research, but if they could just come out and say, "Add a sport chip and subtract 5 pts." so on and so forth..... off to PD in search of some direct answers.... :crazy:
Click to expand...


No, that would make too much sense :sly:

No, I haven't tried to get a hold of them. Guess I just naturally try to reverse engineer the software instead. If you do ask them and find out anything, please let us know!
 
good thread, very interesting. i spent about an hour trying to explain this very fact to my brother that tuning a car in for racing in the pro-hall to make it equate to 1 A-Spec point in a race doesnt do you any justice or real advantage as the competitors cars are just improved aswell to compensate.
 
WorksAccess
good thread, very interesting. i spent about an hour trying to explain this very fact to my brother that tuning a car in for racing in the pro-hall to make it equate to 1 A-Spec point in a race doesnt do you any justice or real advantage as the competitors cars are just improved aswell to compensate.
Click to expand...


Got me confused there a little bit.

The opposition lineup doesn't improve to compensate for changes that you make to your car. That would defeat the purpose of the point system. Now, it's VERY common to have lineups where the opposing cars aren't stock.

As a matter of fact, the cars aren't stock in about 1/2 of the races. (Check the PREVIEWs).
 

Similar threads

ErikaNPC
I finally did it. 10,000 B Spec Points
Replies
6
Views
3K
bruseccio
bruseccio
jco
Performance points disappearing on some cars after latest update
Replies
8
Views
8K
jco
jco
P
(Suspended) Shark's Muscle Mayhem Series Saturday Races Will run a lobby August 11 about 4pm central time
Replies
7
Views
2K
Cold Warrior
Cold Warrior
gti_sdn
At what PP would you consider close if trying to Balance a group of cars
Replies
4
Views
940
gti_sdn
gti_sdn
llNovall
  • Locked
SNAIL-Porsche GT3 Cup Challenge (using new 992 GT3), Season Champ: Sand_68, Congratulations on the Win and a Great Season!!!
23 24 25
Replies
724
Views
67K
Noob01_lmdead
Noob01_lmdead

Latest Posts

Back