Time Trial Discussion

  • Thread starter seadog777
  • 27,765 comments
  • 1,750,603 views
No luck with GW in my last trial yesterday eveninig.
I really think 55' was in my capabilities, but I was not able to put a lap togheter.
It is what it is....
As a lesson learned I think I spent too much time to figure out different lines in couple of turns. The exercize itself is ok, the way I approached was wrong, giving only couple of tries to a lane before moving to the other. I really spent time going back an forth while probabily sticking with one and perfecting it would have been better, even if the choosed one is not the absolute best.

I'm quite happy with the my trend over the TTs. My bronzes are usually Nordschleife and Gr1/F1 cars, so probably another bronze in on its way with the brand new TT

1688050788075.png
 
Well against my better judgement and doing this instead of working, I present to you my testing results (Minus the Jag!! I totally forgot to screenshot my FL and Opt time and while I have the FL I'm too lazy to update the table and graphs + Apparently skipped over the Genesis G7 SMH)! I'd like to update this post, potentially, if you all have some suggestions for comparisons using the data I have - paging @half_sourly for some ideas!

Gr 4 Table.png


  • I was quite surprised and happy to see that I would have gotten gold on all but 4 of the 29-cars (shows 30 cuz I did the 458 twice)!
  • Based on the linear trendline for fastest lap I have a hunch that I would have gold'ed the Viper & 650S had I tested them later in the process as the R-squared value was 0.0347 (0.0885 filtering out the god-awful M4 that @Evilmuffin somehow gold'ed)
  • Another reason for thinking that was when I looked and ranked the split between FL & Optimal, those two ranked 1st and 2nd
  • All but the M4 have gold optimal times!
  • Obvious, but backed by data, the more laps you drive, the more consistent you are driving the track. In addition to the R-Squared value showing that correlation a good example is the Ferrari 458 Italia (noticed I misspelled in the table). It's the 3rd & 30th car I tested. 1st test ended with 6th FL and last ended with 1st FL, improving 0.544-seconds but when looking at optimal rank, it ended 2nd and 1st with only 0.255-second improvement.
  • I also graphed the FL to Optimal splits and the R-Squared value was a whopping 0.2025! That's some fairly strong correlation and probably a better indicator of track memory improving and resulting in smaller splits as my testing continued.
  • Don't let the aliens who set a top 1,000 times "in only 5-6 laps" when a new TT comes out lol. Think you suck? They just play more than you or have been investing in their skills for a much longer time period.
  • One variable that I didn't control or keep track of at all is the # of laps for each test. They varied quite a bit with some cars getting >20 and some <6-7 - Some felt like they had more potential and others felt like it would take a bunch of laps to maybe get close to PB's. Bias for driving style preference
  • Another takeaway is try different cars out! The majority of the cars I tested had 0-miles on them and probably 0-miles on the road version as well. The RCZ was such a gem to find and felt made for this track, wish I would have put more laps in with it - Same with the C7-Corvette, Hyundai Genesis & Lexus RF
Graph showing FL & Optimals in the order tested along with linear trendlines
Testing Graph 1 with BMW.png


Without the BMW M4!
Testing Graph 2 no BMW.png


Optimal Splits Graph
Optimal Split Graph.png


Will have to edit and update this with all the pics of the cars with liveries on them from my phone
 
Saargasm
Well against my better judgement and doing this instead of working, I present to you my testing results (Minus the Jag!! I totally forgot to screenshot my FL and Opt time and while I have the FL I'm too lazy to update the table and graphs + Apparently skipped over the Genesis G7 SMH)! I'd like to update this post, potentially, if you all have some suggestions for comparisons using the data I have - paging @half_sourly for some ideas!

View attachment 1269138

  • I was quite surprised and happy to see that I would have gotten gold on all but 4 of the 29-cars (shows 30 cuz I did the 458 twice)!
  • Based on the linear trendline for fastest lap I have a hunch that I would have gold'ed the Viper & 650S had I tested them later in the process as the R-squared value was 0.0347 (0.0885 filtering out the god-awful M4 that @Evilmuffin somehow gold'ed)
  • Another reason for thinking that was when I looked and ranked the split between FL & Optimal, those two ranked 1st and 2nd
  • All but the M4 have gold optimal times!
  • Obvious, but backed by data, the more laps you drive, the more consistent you are driving the track. In addition to the R-Squared value showing that correlation a good example is the Ferrari 458 Italia (noticed I misspelled in the table). It's the 3rd & 30th car I tested. 1st test ended with 6th FL and last ended with 1st FL, improving 0.544-seconds but when looking at optimal rank, it ended 2nd and 1st with only 0.255-second improvement.
  • I also graphed the FL to Optimal splits and the R-Squared value was a whopping 0.2025! That's some fairly strong correlation and probably a better indicator of track memory improving and resulting in smaller splits as my testing continued.
  • Don't let the aliens who set a top 1,000 times "in only 5-6 laps" when a new TT comes out lol. Think you suck? They just play more than you or have been investing in their skills for a much longer time period.
  • One variable that I didn't control or keep track of at all is the # of laps for each test. They varied quite a bit with some cars getting >20 and some <6-7 - Some felt like they had more potential and others felt like it would take a bunch of laps to maybe get close to PB's. Bias for driving style preference
  • Another takeaway is try different cars out! The majority of the cars I tested had 0-miles on them and probably 0-miles on the road version as well. The RCZ was such a gem to find and felt made for this track, wish I would have put more laps in with it - Same with the C7-Corvette, Hyundai Genesis & Lexus RF
Graph showing FL & Optimals in the order tested along with linear trendlines
View attachment 1269161

Without the BMW M4!
View attachment 1269162

Optimal Splits Graph
View attachment 1269163

Will have to edit and update this with all the pics of the cars with liveries on them from my phone
Click to expand...
As a data guy I approve of your dedication but as one that still needs a morning coffee, this made my head hurt! LOL.
 
half_sourly
Here are some statistics for TT31 at Watkins Glen Long Course in Gr.4 cars.
Click to expand...
Outstanding work as always!
I have an idea or request if you will.

Can we have a table or leaderboard of some sorts with times achieved by thread regulars at the end of each TT? It shouldn't be hard to filter out using our nicknames, right?

My reasoning is twofold:

1. Not everyone has everyone in their friends list on PSN and I think it would be interesting to see how the times spread out over the whole community.

2. Even if someone has everyone in their friends list, the friends leaderboard vanishes once the time trial is over, which is rubbish if you ask me. So I can't see if some of my friends did a last minute push to slid into gold like it happened this morning and how the leaderboard finally shaped out. Looking through an archive of past TTs would be very interesting as well.

It would require consent from each person to be published in such organised manner, but I think most of you wouldn't mind that, would you?

I, myself don't mind at all, my GT7 nickname is tmtk.
 
Top Dog
I just realised that the Valkyrie has an overtake button, an extra thing to contend with. Just gone from a 1:18.9 to a 1:18.1 lol.

Smooth inputs is the way with this beast. Overall fun to tame and satisfying when you get it right.
Click to expand...
Good to know about the overtake button but as I'm alone on the track, I need a button to gain 2 secs :)
#Askingforafriend
 
Saargasm
Wait a minute, this says you're in Wisconsin! Aren't you central time aka 1-hour before me?? Must be living the good life needing morning coffee at near 10:30 AM on a Thursday lol. I have done exactly 3-work related things this morning and its almost lunch... 0-dials too :banghead:
Click to expand...
Yes. But I still have some British roots so a cup of tea first thing and then usually coffee around 9:30... Running late this morning....

Oh, and central is one hour BEHIND eastern. My my, Floridians, the sun must have gone to your head...
 
Last edited:
@half_sourly i really think you need a massive shout out for the TT data and data vis work you are doing!

This Aston TT is quite the test especially when the car will aggressively wheel spin at such high speeds. I think I may tire of this one quite quickly.

I can't use my main account until Saturday but early tries this morning on my new account were very mixed and inconsistent for me (A/B S) I seem to fair better in the road cars and GR3/4 cars than the odd ball or hyper and LMP cars.

Will see how it goes I'm just glad I managed to get gold times before my suspension on both accounts (annoyingly faster on my alt/new account :) )
 
tmtk
Outstanding work as always!
I have an idea or request if you will.

Can we have a table or leaderboard of some sorts with times achieved by thread regulars at the end of each TT? It shouldn't be hard to filter out using our nicknames, right?

My reasoning is twofold:

1. Not everyone has everyone in their friends list on PSN and I think it would be interesting to see how the times spread out over the whole community.

2. Even if someone has everyone in their friends list, the friends leaderboard vanishes once the time trial is over, which is rubbish if you ask me. So I can't see if some of my friends did a last minute push to slid into gold like it happened this morning and how the leaderboard finally shaped out. Looking through an archive of past TTs would be very interesting as well.

It would require consent from each person to be published in such organised manner, but I think most of you wouldn't mind that, would you?

I, myself don't mind at all, my GT7 nickname is tmtk.
Click to expand...
It is more difficult than you think, but I can do it. The data is spread among many workbooks because the file sizes get massive. And nicknames can be tricky, as there can be multiple instances of each.

There is the added complication in future that I might have to cut down what I do. Today I had to pay for the data scrape for TT31 because Sony/PD/whoever runs the gran-turismo.com site has blocked my normal scraper. That is why the results were delayed. I spoke with the scraping firm and they say sites sometimes increase their blocking intensity, hence me having this problem for the first time. The solution is to pay for the pro version and take advantage of their IP address rotation tool, but that is north of $100 p.m. so clearly I am not doing that. Other suppliers are similar in price.

That said, I'm happy to help, but as this already takes a whole chunk of my time, let me propose the following:
  • - You make a list of all users who wish to be included.
  • - You get their consent for me to slightly process their data (I already have it for some due to an earlier exercise I did - option for you to search for that on this thread or just ask them again).
  • - You get their GT7 nicknames and ensure they are the right ones by getting their GT7 profile page link (example here is mine: https://www.gran-turismo.com/gb/gt7/user/mymenu/acfa6a32-1c74-4c0b-852e-b8a1bf152d08/profile)
  • - Share with me that list of nicknames and profile page links
  • - Tell me what data you want, for example, time, position?
  • - Then I will create the table and update after each TT, as long as I can resolve the scraping matter.
 
TheNormsk
My my, Floridians, the sun must have gone to your head...
Click to expand...

I blame the evening storm they are having almost every evening :D

half_sourly
Today I had to pay for the data scrape for TT31 because Sony/PD/whoever runs the gran-turismo.com site has blocked my normal scraper. That is why the results were delayed. I spoke with the scraping firm and they say sites sometimes increase their blocking intensity, hence me having this problem for the first time. The solution is to pay for the pro version and take advantage of their IP address rotation tool, but that is north of $100 p.m. so clearly I am not doing that. Other suppliers are similar in price.
Click to expand...

It might have to do with how fast the scraping tool is hitting the site. I had built a web scraper for my job before..and it was hitting the site so fast they though they were under attack...they (government agency) called our manager and it was a fun conversation. I ended up having to build a delay and slow down the hits.
 
Last edited:
half_sourly
It is more difficult than you think, but I can do it. The data is spread among many workbooks because the file sizes get massive. And nicknames can be tricky, as there can be multiple instances of each.

There is the added complication in future that I might have to cut down what I do. Today I had to pay for the data scrape for TT31 because Sony/PD/whoever runs the gran-turismo.com site has blocked my normal scraper. That is why the results were delayed. I spoke with the scraping firm and they say sites sometimes increase their blocking intensity, hence me having this problem for the first time. The solution is to pay for the pro version and take advantage of their IP address rotation tool, but that is north of $100 p.m. so clearly I am not doing that. Other suppliers are similar in price.

That said, I'm happy to help, but as this already takes a whole chunk of my time, let me propose the following:
  • - You make a list of all users who wish to be included.
  • - You get their consent for me to slightly process their data (I already have it for some due to an earlier exercise I did - option for you to search for that on this thread or just ask them again).
  • - You get their GT7 nicknames and ensure they are the right ones by getting their GT7 profile page link (example here is mine: https://www.gran-turismo.com/gb/gt7/user/mymenu/acfa6a32-1c74-4c0b-852e-b8a1bf152d08/profile)
  • - Share with me that list of nicknames and profile page links
  • - Tell me what data you want, for example, time, position?
  • - Then I will create the table and update after each TT, as long as I can resolve the scraping matter.
Click to expand...
I'd happily contribute $5 a week to keep you doing this mate, just another $90 to go (you gotta have skin in the game to;) h
 
newmedia_dev
I'd happily contribute $5 a week to keep you doing this mate, just another $90 to go (you gotta have skin in the game to;) h
Click to expand...
Very kind of you but I will sort something out. Paying for one-off scrapes as I did today was only $6.

In the meantime I'll investigate ways around the site's blocking attempts. Using a proxy server is supposed to be one way but I tried that today and it didn't work.
 
half_sourly
Here are some statistics for TT31 at Watkins Glen Long Course in Gr.4 cars.

Highlights:
  • Participation was up again and well above the average to date.
  • This was another easier TT, possibly 2nd easiest to date, with over 25% of drivers getting gold (2nd easiest to date) and almost 39% getting silver (again 2nd easiest). Only 3.3% of drivers got no reward, the smallest proportion to date under normal rules.
View attachment 1269081
View attachment 1269082
View attachment 1269076
View attachment 1269156

View attachment 1269155

Drivers with gold in all TTs
The number of drivers with gold in all TTs to date remained at 126.

As usual I've split the graphs because the high early numbers mean the later low numbers are barely legible when all the data is on one graph.
View attachment 1269153
View attachment 1269154
View attachment 1269075
View attachment 1269152

View attachment 1269077
View attachment 1269080
Yes.
View attachment 1269078View attachment 1269079
My results:
  • Time: 1:53.957 - Gold streak continues
  • Position (percentile): 352/131,890 (Top 0.27%)
  • 0.99% off #1 and 0.3% off #100.
Click to expand...
Recently, I haven't been able to spend too much time on the TTs, but despite dropping in performance a fair bit I managed to stay in gold. Strongest TTs would be N24. I'd be interested in the overall personal statistics, so if you find the time I'd love to take a look. Cheers and thanks for constantly providing us with this interesting data :)
 

Latest Posts

Back