News:

Herr Otto Partz says you're all nothing but pipsqueaks!

Main Menu

Cars and rules for 2023

Started by dreadnaut, November 20, 2022, 10:49:42 AM

Previous topic - Next topic

dreadnaut

Quote from: Duplode on November 20, 2022, 08:30:07 PMIt's not a priority, but to keep the flame of the perennial proposal burning, we might want to think about moving away from the linear points system. I feel such a change might make the season scoreboard more dynamic.

I've always assumed that non-linear point would make it easier for one pipsqueak to "run away", and more difficult for others to catch up, but I'm likely wrong. Could you elaborate on the differences?

In the past, ZakStunts has used these point systems:
2001-2003  —→ 12->1
2004       —→ 20, 16->6 step 2, 5->1, 0.09->0.01 step 0.01
2005       —→ 12->1, 0.18->0.01 step 0.01
2006-2007  —→ 15->1, 0.11->0.01 step 0.01
2008-today —→ 12->1, 0.10->0.01 step 0.01


alanrotoi

Points in 2001 and 2002 were 10-1 and in 2003 implemented a super complicated system (from nascar?) 275-270-265-260...

dreadnaut

Whops, then just as well that automatic calculations are not running for those early seasons 😅

alanrotoi

Is it possible to add 2001-2003 data to the rest? I would like to help about it if I can.

dreadnaut


dreadnaut

#20
Quote from: alanrotoi on December 10, 2022, 12:34:33 PMPoints in 2001 and 2002 were 10-1 and in 2003 implemented a super complicated system (from nascar?) 275-270-265-260...

The rules for 2003 mention the "175" scoring in regards to teams, but not for the main scoreboard. The scoreboard says otherwise ???   I'll update the definitions, which will help bringing the old data into the system.

[edit] Ah, now I understand: in 2003 the "pro" and "amateur" scoreboards used 10->1, but the unified board uses Nascar-style! That's tricky if not impossible to bring into how ZakStunts works post 2003. I'll see what I can do, might just copy the data and block re-calculation.

afullo

Quote from: alanrotoi on December 10, 2022, 12:34:33 PMPoints in 2001 and 2002 were 10-1 and in 2003 implemented a super complicated system (from nascar?) 275-270-265-260...
I remember similar scores from Papyrus's Nascar Racing. The only game, apart from Doom, to cause me a Venetian blind crash once...

Duplode

#22
Quote from: dreadnaut on December 10, 2022, 11:38:30 AMI've always assumed that non-linear point would make it easier for one pipsqueak to "run away", and more difficult for others to catch up, but I'm likely wrong. Could you elaborate on the differences?

I believe non-linear points would make it easier to recover from isolated bad results. For instance, suppose pipsqueak X gets a 6th place in a race won by pipsqueak Y, and that in the next several races X wins with Y 2nd. Under our current system, the first race would amount to a gap of 5 points, and X would need five races to catch up. With the 2004 system, though, the initial gap of 12 points would be bridged in just two races.

Switching to a non-linear system would also have implications for LTB. As it stands, a single LTB point cancels out the difference between 1st and 2nd place. Larger gaps for the positions near the top of the scoreboard would lessen the impact of LTB somewhat. (I still think that would be a good thing, though the new LTB system makes outsized influence of leading time less of a concern.)

One situation in which non-linear scoring should make it harder to catch up is if someone is trying to sustain a challenge mostly on the basis of regularity (e.g. lots of 2nd places but few or no wins). I'm inclined to think that would be an acceptable cost. In any case, it could be helpful to simulate past season scoreboards with different scoring systems, in order to get a better feel for what might change. 2015 and 2020 might be good places to start.



Now please excuse me while I get ahead of myself and speculate about possible non-linear systems: 

While I never got to race under the 2004 system...

1st  2nd  3rd  4th  5th  6th  7th  8th  9th  10th 11th 12th +
20   16   14   12   10   8    6    5    4    3    2    1    point fractions

... its 16..6 linear block looks a little too wide, and the non-linearity could be better spread across the scoreboard. My gut feeling rather favours something like Formula One scoring...

1st  2nd  3rd  4th  5th  6th  7th  8th  9th  10th
25   18   15   12   10   8    6    4    2    1

... though we'd likely want something less top-heavy and easier to relate to our current system. For that purpose, I offer the following hybrid:

1st  2nd  3rd  4th  5th  6th  7th  8th  9th  10th 11th 12th +
24   19   15   12   10   8    6    5    4    3    2    1    point fractions

One way to think about the non-linearity here is bonus points are granted to podium places by increasing the gaps, relative to the 4th-7th range: one bonus point for 3rd, two for 2nd and three for 1st (the gaps grow from 2 to, respectively, 3, 4 and 5). In such terms, the 2004 system gives two bonus points to the winner only.

As for LTB, since this system gives 24 rather than 12 points for 1st place, it would probably make sense to give +2 and +4 points instead of +1 or +2, or perhaps just give LTB points every 120 h rather than 240 h.

(A final, technical note: if you start by assigning 12 points for 1st and 1 point for 12th, then fit an exponential to figure out the points for the intermediate positions, multiply by two and round, the results from 2nd to 10th match the hybrid system above. That means the hybrid system makes the non-linearity about as smooth as it can get without over-complicating things. Below is a chart comparing the scoring systems that illustrates this point.)

You cannot view this attachment.

Daniel3D

I feel like a non linear scoring like you explain would leed to bigger gaps between the top and bottom of the scoreboard and not a fair change.
The current system that the worst 3 results don't count is better I believe.

Is it a lot of work to recalculate this season to you idea to see it's impact?
Edison once said,
"I have not failed 10,000 times,
I've successfully found 10,000 ways that will not work."
---------
Currently running over 20 separate instances of Stunts
---------
Check out the STUNTS resources on my Mega (globe icon)

Argammon

Duplode explains this really well. I would also prefer a non-linear system.

@Daniel3D As far as I see it, catching up does not become more difficult. If you become 1st, 5th, 5th and I become 5th, 5th, 1st, I catch you at exactly the same point under both systems.

It is mostly about the question whether, for example, 1st 1st 6th, indicates a better performance than 3rd 3rd 2nd. I think so, but others do not. It is all about personal preference.  :)

alanrotoi

Quote from: Daniel3D on December 11, 2022, 08:21:48 AMI feel like a non linear scoring like you explain would leed to bigger gaps between the top and bottom of the scoreboard and not a fair change.
The current system that the worst 3 results don't count is better I believe.

Is it a lot of work to recalculate this season to you idea to see it's impact?

I agree. Non linear points didn't work in the past. I'm against it because it is a little elitist. It benefits the top pipsqueaks only.

The current system prevents big gaps with a linear point system. It makes you race in every track and we have the "worst 3 result doesn't count" rule to equalize. And also we have the LTB wich never in the 21 seasons it was implemented determined a winner but always helped to the 3rd-6th positions.

Argammon

Quote from: alanrotoi on December 11, 2022, 12:25:28 PM
Quote from: Daniel3D on December 11, 2022, 08:21:48 AMI feel like a non linear scoring like you explain would leed to bigger gaps between the top and bottom of the scoreboard and not a fair change.
The current system that the worst 3 results don't count is better I believe.

Is it a lot of work to recalculate this season to you idea to see it's impact?

I agree. Non linear points didn't work in the past. I'm against it because it is a little elitist. It benefits the top pipsqueaks only.

The current system prevents big gaps with a linear point system. It makes you race in every track and we have the "worst 3 result doesn't count" rule to equalize. And also we have the LTB wich never in the 21 seasons it was implemented determined a winner but always helped to the 3rd-6th positions.

To be fair, that has always been the majority opinion. We had exactly the same discussion in the old days, and I have always been in the minority. I just do not have mainstream taste. 8)

afullo

In my opinion nonlinear scoring could in the case work for rac3rs, but not for teams, because it would advantage those having a single strong member, rather than two or more collaborative mid-level pipsqueaks.

So, if we switch to nonlinear, then two different score systems (one for rac3rs, one for teams) can be considered.

Duplode

Wow, this does look like a contentious issue! Let me quickly address some of the points here.

@alanrotoi -- I genuinely don't think of non-linear scoring as elitist. From my point of view, the absolute value of the total scores is not very important, and the key point being how easy it is to bridge gaps on the season scoreboard. A difference of, say, six points might look small to the eye, but that doesn't matter much if it likely takes half a season of hard work to overcome it.

@afullo -- Good point about team scoreboards; we might indeed want to consider the implications for them separately.

@Daniel3D -- It shouldn't be too much trouble to get the simulations done, considering that I already have the historical race standings compiled in table form. I'll try to get it done today, or failing that over the next few days.

Also, I'm not at all suggesting we should stop discarding the three worst results. While the discards also help with lessening the impact of unusually bad results, their main point is allowing people to miss a few races without worrying about the season scoreboard. That makes the discards too important to give up IMO.

Duplode

#29
Quote from: Daniel3D on December 11, 2022, 08:21:48 AMIs it a lot of work to recalculate this season to you idea to see it's impact?

I have now ran a few simulations, and it looks like those of you who criticised the proposal were right in being suspicious. With respect to making season scoreboards more dynamic and competitive, the results are mixed, and I don't feel I can recommend the non-linear system on those grounds. In particular, I'm now warming to @alanrotoi 's view of LTB being, after the 2020 changes at least, a benign balancing factor. With a non-linear system, it seems we'd have to choose between LTB being relatively small next to the podium place gaps -- which could lower competitiveness over the top positions at least as easily as it could raise it -- and LTB being large in absolute terms -- which risks unbalancing everything, specially the season scoreboard midfield.

Below are some results from recalculations done for the 2021, 2020 and 2015 seasons. A few notes about the tables here:

  • "Current x2" values were calculated with the 2022 ZakStunts rules, except that I have doubled the points for positions 1-12 and LTB to make comparisons easier. "Hybrid" values were calculated with the non-linear system I presented above.
  • "Real" is the score with discards from the season scoreboard.
  • In the 11th round tables, "Available" is the maximum points the pipsqueak could get in the final race. It is not simply the points for 1st place plus maximum LTB because discards have to be accounted for. (For instance, in the last race of 2008 Ayrton only had the LTB points available, as winning all previous races meant there was nothing left to be discarded.)

I have also attached the source workbook for 2021, so that you can explore the data. There are instructions on how to set up the workbook in the first sheet ("parameters"), and the raw data for other seasons is included as well.



2021 -- There's a lot going on between 3rd and 8th place here:

  • There is a very close fight between Overdrijf and Zapper. The system change has little effect over it.
  • Under the current rules, CTG (6 missed races out of 11) comes back for the final race and, with a 2nd place and maximum LTB, overtakes dreadnaut, Zapper and Overdrijf to end up in 4th place overall. With the non-linear system, though, CTG is already 4th after the eleventh race, and gets to overtake KyLiE for 3rd in the final round.
  • Under the current rules, dreadnaut ends the eleventh round one position ahead of Akoss (6 missed races out of 11), who would go into the final round with a challenging gap to bridge. With the non-linear system, the roles are reversed.
  • Something not in these tables: the 2021 season was decided after ZCT243, the tenth round (real gap between me and Alan of 16 points, while I had only 14 points available from the two final races). With the non-linear system, though, the championship would still mathematically be in play after the tenth round, if only barely (real gap of 19 points, with 22 points available for me, and 31 for Alan). The difference can probably be attributed to the lessened influence of LTB near the top of the season scoreboard.

2021, current rules (x2), after the 11th round (ZCT244):

Real LTB Available
Alan Rotoi 230 24 8
Duplode 212 6 6
KyLiE 158 0 14
Overdrijf 136 0 18
Zapper 128 0 20
dreadnaut 116 0 18
CTG 114 10 28
Akoss Poo 100 0 28
2021, current rules (x2), final results:

Real LTB
Alan Rotoi 234 24
Duplode 212 6
KyLiE 160 0
CTG 140 14
Zapper 138 0
Overdrijf 138 0
dreadnaut 116 0
Akoss Poo 100 0

2021, hybrid non-linear, after the 11th round (ZCT244):

Real LTB Available
Alan Rotoi 216 24 13
Duplode 197 6 9
KyLiE 107 0 20
CTG 97 10 28
Overdrijf 85 0 23
Zapper 81 0 24
Akoss Poo 77 0 28
dreadnaut 64 0 23
2021, hybrid non-linear, final results:

Real LTB
Alan Rotoi 225 24
Duplode 197 6
CTG 120 14
KyLiE 109 0
Zapper 89 0
Overdrijf 86 0
Akoss Poo 77 0
dreadnaut 64 0

2020 -- A season which featured a three-way battle for the title that went to the final round:

  • With the non-linear system, the title would have been decided at the end of the eleventh round! If the change of balance would have delayed the decision in 2021, here it would have anticipated it, robbing us of a three-way decider.
  • With the current rules, Overdrijf and dreadnaut reach the final round drawn (in both real and available points!) while fighting for 4th place. Under the non-linear system, Overdrijf would have a rather significant advantage of three points (which amounts to more than a position anywhere below the podium).

2020, current rules (x2), after the 11th round (ZCT232):

Real LTB Available
Duplode 210 2 8
Seeker1982 204 16 12
CTG 202 14 10
Overdrijf 162 2 14
dreadnaut 162 6 14
2020, current rules (x2), final results:

Real LTB
Duplode 214 2
CTG 210 18
Seeker1982 208 16
Overdrijf 166 2
dreadnaut 164 6

2020, hybrid non-linear, after the 11th round (ZCT232):

Real LTB Available
Duplode 199 2 13
Seeker1982 172 16 18
CTG 171 14 16
Overdrijf 114 2 20
dreadnaut 111 6 20
2020, hybrid non-linear, final results:

Real LTB
Duplode 208 2
CTG 182 18
Seeker1982 177 16
Overdrijf 118 2
dreadnaut 113 6

2015 -- one of the closest title battles ever:

  • The non-linear system would remove some of the sting of the big final round battle, as Akoss and me wouldn't go into it drawn. Still, Akoss would have been able to win the title without depending on my result (9 available points to bridge a gap of 6, while a perfect race by Akoss would leave me with at most 2 more LTB points, as getting anything else would require discarding a 2nd place). It was a pretty close season for the two of us in terms of LTB, so the balance is not affected all that much despite the old LTB rules.
  • AbuRaf versus Usrin for the 5th place would see little change under the non-linear system, though the final gap would be a bit larger in relative terms.

2015, current rules (x2), after the 11th round (ZCT172):

Real LTB Available
Duplode 236 24 6
Akoss Poo 236 28 6
Marco 174 2 12
AbuRaf70 144 0 16
Usrin 138 0 18
2015, current rules (x2), final results:

Real LTB
Duplode 242 28
Akoss Poo 238 30
Marco 180 2
AbuRaf70 148 0
Usrin 146 0

2015, hybrid non-linear, after the 11th round (ZCT172):

Real LTB Available
Duplode 230 24 9
Akoss Poo 224 28 9
Marco 127 2 18
AbuRaf70 92 0 22
Usrin 86 0 23
2015, hybrid non-linear, final results:

Real LTB
Duplode 239 28
Akoss Poo 226 30
Marco 136 2
AbuRaf70 96 0
Usrin 93 0