Car bonuses

[alanrotoi]

According to the statistics, the popularity of the winner cars (since 2008 when the car bonus rule started) is the following:

16 times Ferrari GTO
15 times Acura NSX
15 times Porsche March INDY
13 times Corvette ZR1
11 times Lamborghini Countach
11 times Audi Quattro Sport
11 times Jaguar XJR9 IMSA
10 times Lancia Delta Integrale
10 times Lamborghini LM002
10 times Porsche 962 IMSA
09 times Porsche Carrera 4

The most winner cars are those with Power Gear. Should we put under a magnifying glass the car bonus rules to make them appear not so often and not so surprising?

[Daniel3D]

It is actually surprisingly balanced sure there is a difference, but it seems the bonus system works.
The problem is more in crazy PG laps than the bonus system i think.

[Duplode]

I do think we have a bit of an issue with power gear surprises. While split podiums are generally a nice thing (multiple competitive cars!), split podiums with power gear cars are not so nice (it often means, in effect, two completely different races on the same scoreboard). It is a tricky situation, as the bonus system makes it hard to have one thing without the other. If y'all have any ideas to tackle that, I'd love to hear them!

(In particular, in an earlier discussion, I had suggested having a "quarantine" period in which power gear cars wouldn't recover bonus points after a win/podium would make power gear races happen less often, but probably wouldn't make it less likely for any given power gear race to be a surprise or split podium one, so it doesn't look too attractive a solution.)

P.S.: Two years ago, I made a comparison between car choices before and after the 2019 rule changes. It might be useful to update that with the 2021 and 2022 results...

[Daniel3D]

Exclude a PG car from the rest of the competition after a PG win. That would make the amount of PG races equal or less than the amount of PG capable cars.

[Duplode]

Exclude a PG car from the rest of the competition after a PG win. That would make the amount of PG races equal or less than the amount of PG capable cars.

That's one possibility. However, I feel the bonus "quarantine" I mentioned above could have a similar effect with two potential advantages:

• It would mesh more easily with the rest of the bonus system -- in particular, there would be no discontinuities from one season to the next (e.g. it would make no difference if the PG race happened in February or December).
• It would be possible to tune how much of a penalty we want to give PG cars by making the quarantine longer or shorter.

(Also, it's worth it underlining again that both of our ideas would make all PG races rarer, as opposed to just split podium ones.)

[alanrotoi]

(Also, it's worth it underlining again that both of our ideas would make all PG races rarer, as opposed to just split podium ones.)

That's because maybe we are not facing the real problems: PG awareness and track design.

PG awareness, maybe we got too lazy or lack of time to test every possible car on every track but sometimes happens. Sometimes we should be more aware about it, for example I "got asleep" in zct252 and didn't see it was possible a better lap with Acura. How to fight it? Well we have to be more aware. Is it a solution? Not exactly because PG races still will be as often as they are.

Track design: there might be a more responsible track design preventing unwanted effects. How? A good way to prevent PG for example is creating long rivers crossing corners or crossing key parts. (Not flat track please, this is Stunts). We could be more PG restrictive from the design. It would generate also a good thing. If the bonus of gto or vett gets so high we could see races with both cars almost without PG.

But we can't ask a responsible track design because it eventually won't happen and we can't ask dreadnaut to "fix" every track because the meaning of guest tracks is to prevent him that job. So we should work in a restrictive rule.

If January starts with a -20% or -30% for every PG car it will generate two things: a delay and when a PG car appears will be exclusively because the PG. Maybe the quickest solution but in a 60% or 70%.

[Daniel3D]

(Also, it's worth it underlining again that both of our ideas would make all PG races rarer, as opposed to just split podium ones.)

That's because maybe we are not facing the real problems: PG awareness and track design.

Not everyone likes PG. And about track design. I really tried to make Monaco fast but not PG capable. But you managed to make a PG lap anyway..

[Duplode]

PG awareness, maybe we got too lazy or lack of time to test every possible car on every track but sometimes happens. Sometimes we should be more aware about it, for example I "got asleep" in zct252 and didn't see it was possible a better lap with Acura. How to fight it? Well we have to be more aware. Is it a solution? Not exactly because PG races still will be as often as they are.

Extra awareness could be part of a solution, though I agree with @Daniel3D that it is tricky to get it right all the time. It's not just that some lines are really easy to overlook, but also because sometimes, given a certain track concept, there is just no budget for adding an extra PG breaker section without ruining the flow of the track.

A different sort of awareness we might want to further cultivate is that of bonus percentages, in particular to make sure we take full advantage of the single-track bonus adjustments available to designers. By the way, I think we might have enough bonus data by now to make it worth looking at the historical trends, looking for the cutoffs at which various cars become favoured. Hopefully I'll get to do some work on that before the new season.

P.S.: Here is an update (up to Z257) of those car usage stats I mentioned earlier:

Code Select Expand

Car Points 1-6 Points 1-10 1st 2nd 3rd Podiums Full Podiums Occurr. Seasons Notes
PMIN 85 263 6 1 2 7 1 18 4 PG
VETT 87 274 5 3 3 5 3 17 4 PG
ANSX 82 273 4 2 4 5 2 20 4 PG
P962 103 448 3 3 5 5 2 50 4
PC04 86 354 3 3 3 3 3 39 4
COUN 85 321 3 4 2 4 2 30 4
AUDI 77 273 3 4 2 4 1 25 4
LANC 78 308 2 4 4 4 2 30 4
JAGU 68 282 2 3 3 3 2 29 4
LM02 55 220 2 2 2 3 1 23 4
SUKA 46 173 2 2 1 2 1 18 2 19, 20
RANG 45 178 2 2 1 2 1 18 1 19
FGTO 36 108 2 2 1 3 1 7 4 PG
DAUD 29 102 2 0 1 3 0 9 2 19, 20
DBMW 60 272 1 3 4 4 1 34 2 19, 21
CDOR 51 255 1 2 2 2 1 32 2 21, 22
ZTST 28 120 1 1 1 1 1 13 2 20, 21
ZLET 26 101 1 1 1 1 1 10 1 21
ZCS7 26 109 1 1 1 1 1 12 1 22
CERV 26 101 1 1 1 1 1 10 1 22
ZF40 21 71 1 1 1 1 1 6 2 19, 20
DMCB 32 165 0 2 2 2 0 21 2 20, 21
STRA 16 75 0 1 1 1 0 8 1 22
NSKY 0 2 0 0 0 0 0 1 1 22

Over the last four seasons, we've had:

• 17 PG wins, with 7 single-car PG podiums. (Among the mixed PG podiums, only Z227 featured two different PG cars.)
• In particular, 6 Indy wins but only one Indy full podium.
• As for the non-PG cars, there were 31 wins and 22 single-car podiums.

Something else not in this table that caught my eye is how recovery periods after mixed podiums can be quite short. For instance, after a mixed podium win on Z252, the Acura reemerged for a strong 3rd place lap on Z255.

[Cas]

I would like to point out the difference between two things here. When you guys say "a PG win", do you mean just a win with a PG-capable car?  Or do you specifically mean a win in which the PG feature was fundamental?  For cars like the GTO, probably these two are the same, but say, the March Indy is a fast car and may be picked because of that and maybe at some points of the race, you do use PG, but still the race wasn't won because of PG. I think this difference should be taken into account.

As regards a negative bonus, I agree that this would have a negative impact on the choice of the car even when the track is not PG-friendly. If the bonus is too strong, PG cars will simply never come out and if it's subtle, we'll get what Duplode says: we'll have a negative impact on the cases in which PG wasn't fundamental, but it won't be negative enough when it was, so we'll be doing the opposite of what we want to achieve. This is all analysis, of course, and maybe trying it would yield something different, but well, good to consider.

I like what the guys point out about awareness. Seeing a PG lap coming is kind of like recognising the opportunity for a shortcut and nobody questions shortcuts here. In fact, they're essential in ZakStunts. On track design, yes, we can do more. Not that Dreadnaut will have to fix all tracks, but if he sees something and wants to warn us in case we want to change something, I think that's fair. I'm glad to accept suggestions and I'll try to be aware of these PG opportunities in my designs.

[Duplode]

I would like to point out the difference between two things here. When you guys say "a PG win", do you mean just a win with a PG-capable car?  Or do you specifically mean a win in which the PG feature was fundamental? 

That's a good point. For the stats above I counted any win with a PG car as a "PG win", regardless of what the racing line was like. After doing a quick review of those 17 PG wins, I'd say only ZCT230 counts as a non-PG lap (while there's short PG section, it isn't fundamental in the way you describe). You might also want to regard ZCT227 and ZCT242 a bit differently, as they seem clearly meant as Acura-friendly PG speedways from the start, but that's about it.

Also, just to be 100% clear: I do enjoy PG races, and would miss them very much if they were no longer featured on ZakStunts. The goals I have in mind here are merely making sure that non-PG cars aren't crowded out of the scoreboards, and that not too many of those PG races that do happen turn out to be PG surprises.

[Daniel3D]

Ok. Personally I don't like PG or PG races. In the CCC i try hard to prevent it. But it is indeed part of Zakstunts.
It's I think ultimately up to the track designer to make a track PG friendly it not.

In crazy eight i used Dual ways and water, not effectively apparently. It had a PG win.
Monaco had a lot of scenery, but also the PG win.
So I have to try harder.

Maybe we can open a topic with design tips for if you want to make a certain type of track.

[Cas]

I like PG races too (not that this is of any convenience to me, since I'm not good at exploiting PG) and I think that ZakStunts is about freedom, so they should be allowed, but that's my personal point of view. Race For Kicks, for instance, is more about hard rules, so there's something for everybody.

I also like Daniël's point. If you don't like it, you try to prevent it on your track design. And like Duplode mentioned, what we're trying to prevent here is more the "surprise", which splits a race into two. If everybody knows we're racing PG, then it's not a big deal.

[Duplode]

Maybe we can open a topic with design tips for if you want to make a certain type of track.

That could be a good topic for the Wiki as well! Meanwhile, here are a few words on thing to keep an eye for when it comes to PG prevention:

• The most reliable way of limiting PG probably is using water to force corners (flat or bridge) to be driven OWOOT. There are of course plenty of downsides to flooding tracks -- it limits a lot the possibilities in terms of shortcuts, terrain and even aesthetics. That being so, when I opt for this approach I prefer shorter, strategically placed water sections. (For instance, in ZCT243 the pair of corners after the tunnel and the northwest bridge were, in part, meant as PG breakers.)
• Keeping in mind the differences between the PG cars helps with identifying what you actually need to deal with on a given track. (GTO and Vette: "rigid", hard to reach PG, with powerslides; Indy: "flexible", easy to reach PG, with powerslides; Acura: "flexible", easy to reach PG, no powerslides.)
• Loops should be handled carefully. With Indy or Acura a loop often means instant PG. Furthermore, if the loop comes early enough on the track, even Vette and GTO might be launched into race-upturning PG lines, thanks to Rotoi's Trick ™ (for a recent example, see ZCT240).

In any case, I feel it is important to avoid putting too much weight on the shoulders of track designers. I'd much rather have the occasional PG surprise than see tracks constrained into sameness by strict expectations of what they should look like. That's why it would be nice to be able to shift some of the load to the coefficient system, even if it's just by getting better insight on how to use track-only custom bonuses in an effective way.   

[Duplode]

By the way, I think we might have enough bonus data by now to make it worth looking at the historical trends, looking for the cutoffs at which various cars become favoured. Hopefully I'll get to do some work on that before the new season.

Update: I have done some preliminary analysis along those lines. The results give a better sense of how large the difference in behaviour between PG and non-PG cars is.

Here is a brief summary of my strategy here. Given any two cars and their bonuses, we can calculate the advantage one has over the other. Morally, the advantage corresponds to the ratio between the coefficients, though the actual calculation is slightly more complicated: subtract the percentages from 100%, linearise by taking the (base 2) logarithms, and finally take the difference between the values. Now, to see how these advantages get reflected on the racetrack, we can  take every race in which at least one of the cars has made it to, say, the top 6, and arrange them in increasing order of advantage:

You cannot view this attachment.

In this chart, blue means Lancia won, and orange, Carrera. The columns display the Carrera advantage at each race; that being so, negative values mean the Lancia had a higher bonus than the Carrera, and vice versa.

The key point here is how there is a clear separation between the blue and orange regions, making it easy to tell at which point the Carrera pulls ahead of the Lancia. Other combinations of non-PG cars often lead to scenarios like this one, with little or no overlap and clean cutoff points -- and generally the more similar the cars are, the cleaner the cutoff gets. If we replace the Lancia with the GTO, though, things get quite different:

You cannot view this attachment.

As we know, powergear introduces a lot of variability to car performances -- is there be a PG line on the track? if so, how effective it is, and will people be able to find it? This is reflected in the chart, which has a large overlap between blue and orange regions (look at the span between Z81 and Z128, which takes pretty much all of the middle of the chart). To make it easier to picture how large this overlap is, we can convert the advantages back to bonuses. Here I have done that by keeping the Carrera bonus fixed at 25% and calculating the bonus percentage for the other cars from the values in the charts:

Code Select Expand

Min. bonus Max. bonus
Car w/PCO4 loss w/PC04 win

ANSX 11.2 18.7
AUDI 16.6 20.0
COUN 22.5 23.0
FGTO 4.1 20.5
JAGU 1.7 7.2
LANC 20.0 20.8
LM02 29.8 32.0
P962 -1.5 2.3
PMIN -45.5 -20.8
VETT 9.5 22.8

The values in the table correspond to the rightmost blue column and the leftmost orange column in each car's chart against the Carrera. Given a bonus percentage in between them, we know by experience that, under the right conditions, either car could win. The intervals for PG cars are on average much larger, often larger than 10%. (And for the Indy, the intervals are much larger still, not just because it has the most powerful PG, but also because the effect of percentage differences increases with higher bonuses: the difference between 20% and 30% bonus is much more significant than that between -20% and -30%.) That explains why we as track designers can almost never fully relax about PG. Under the current rules, a car can lose at most 15% bonus after a race. That being so, with enough (bad) luck, a PG car might get a full podium, lose 15% and still, given a favourable track, remain in PG viability range.

[Daniel3D]

Under the current rules, a car can lose at most 15% bonus after a race. That being so, with enough (bad) luck, a PG car might get a full podium, lose 15% and still, given a favourable track, remain in PG viability range.

Then it will have a an other 15%.
Is it an option to make a car after a full podium drop to the bottom care bonus panalty? In the current case to -24%