Thread Rating:
  • 0 Vote(s) - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
AI testing results
#11
If it shows he's in a Swift then all sweet. trying to figure out my formula for doing it 1st. want it so Matsushita doesn't win (as per real life) but don't want it so spread out that it's always the same 6/7 cars in the top 5 each race either. thinking all the guys that had a pole/win with highest values (which is obvious) then low values represent how consistent they are. eg Zanardi '98 has same high value as other top guys but smaller value is a bit higher than everyone else for example. have an idea on how Brack '01 can be better on ovals but not so much road courses but have yet to run tests on how the values may look like.
Reply
#12
Quote:........so Matsushita doesn't win (as per real life)......


"Hiro!.... you so srow! you awrays finish rast! - very shameful dispray!"
Reply
#13
Hey now... watch how you're talking about King Hiro! Smiley
Reply
#14
You can calculate average time differences in qualifying and standard deviations of those to make the qualifying results more realistic. Qualifying times/average speeds can be found at YouTube. In fact I made some calculations for the 1995 season though I'm not sure how useful they are. The qualifying results can be found for all the 1995 races except Portland.
Reply
#15
(12-13-2018, 11:16 PM)Purojannu Wrote: You can calculate average time differences in qualifying and standard deviations of those to make the qualifying results more realistic. Qualifying times/average speeds can be found at YouTube. In fact I made some calculations for the 1995 season though I'm not sure how useful they are. The qualifying results can be found for all the 1995 races except Portland.
This website has the times Smiley

http://www.race-database.com/results/res...eries_id=4

This website has the ave start etc stats & it allows you to put them in order so could use that as a base with the in game values? but unsure what parameters etc for each but will hopefully find a good formula for it.

https://www.racing-reference.info/cart.htm
Reply
#16
(12-14-2018, 11:40 AM)JR82 Wrote: This website has the times Smiley

http://www.race-database.com/results/res...eries_id=4

This website has the ave start etc stats & it allows you to put them in order so could use that as a base with the in game values? but unsure what parameters etc for each but will hopefully find a good formula for it.

https://www.racing-reference.info/cart.htm

Thanks for the first link. I didn't know about that site. Average start is good info, but I think it's even better to have the exact times/average speeds. So here are the stats I calculated, average loss to the fastest guy (%) and its standard deviation:

Villeneuve 0,77 1,08
Andretti 0,82 0,80
Fabi 1,22 1,18
Vasser 1,32 0,78
Gordon 1,33 0,75
Gugelmin 1,39 0,73
Tracy 1,40 0,96
Pruett 1,45 0,82
de Ferran 1,47 1,09
Unser 1,64 1,24
Rahal 1,66 1,03
Ribeiro 1,75 0,92
Boesel 1,80 0,80
Herta 1,84 1,56
Fernández 1,92 0,63
C. Fittipaldi 1,92 0,92
E. Fittipaldi 1,95 1,01
Sullivan 2,42 1,44
Johansson 2,56 1,23
Cheever 2,99 1,13
Salazar 4,10 1,47
Zampedri 4,29 2,16
Bachelart 4,38 2,11
Greco 4,55 1,44
Guerrero 4,69 2,70
King Hiro 5,49 1,99

Only regular drivers are included. Hopefully someone can make some use of that information.
Reply
#17
Yer it's very handy for this kind of stuff, got no idea at this stage how best to figure it out with the in game #'s

Using real times would be a good way to go but i don't have an accurate way to do it without it taking too long to do may try using rough gaps with parameters & see how that goes.

More testing, did a high car v low car (gaps of 100) with the same gap (100) for all parameters & found =

Interestingly the gaps at Michigan at bigger in qual (with default times at start of session) but smaller in the race than Milwaukee. seems 200 gaps might be the way to go overall with overlapping done depending on how competitive each car was in a given season. Maybe stretch that to 250/300 for early 90's carsets when field was more spread out. This may mean driver parameters need to be say 75 apart but Zanardi '98 have them 50 apart for example?

high car v low car

100 gaps

= 1.3/2.3 qual 2.2/2.5 per lap in race belle isle/elkhart

= 3.5/5.3MPH qual .6 per lap race both milwaukee/michigan

200 gaps

= 2.7/4 sec qual 4.2/4.6 race belle isle/elkhart

= 6.8/10MPH 1.3/1 sec per lap milwaukee/michigan
Reply


Forum Jump:


Users browsing this thread: 1 Guest(s)