Driver Rating System

  • Thread starter glw
  • 45 comments
  • 3,196 views
To your first point: Although I disagree with you, since I see no good reason why offline races shouldn't count towards your rating, I was just using A-spec as an example.

To your second: True. But average speed is a single number, so it's much easier to work with than a bunch of lap times for various tracks.

To your third: True, but the same goes for using lap times.

So if I want to do mayhem offline, but I'm a absolutely clean driver online, it will be recorded that I'm dirty? Count offline doesn't make sense at all.

I won't even respond to your other points, because they simply don't make any sense.
 
This is much easier than everyone is making it... When you are cruising, goofing, or prepping it's arcade mode and your driving doesn't count towards your rating. When you are racing your driving counts period.

My original premise was so that GT5 can become more of a teaching/training system like GT Academy. In my mind, using the ratings for online filtering was only a side effect.
 
glw
My original premise was so that GT5 can become more of a teaching/training system like GT Academy.

This requires even more realism from the physics. The way I drive an '05 Impreza in the game, for example, is rather different from the way I drive my '05 Impreza in real life. The steering ratios, and many other things you can't feel (even with a steering wheel,) make the entire driving experience entirely different.


So, no, I'm not sure that GT5/GT Academy can really become a teaching/training system. There's still no virtual equivalent to actual seat time, but I'll still say that iRacing's closer, and does a better job of this. The unlimited seat time in iRacing, and the hundredth-of-a-second accurate timing, that you can visually see as a colourful bar on your screen really shows you where you're fast and where you're slow. For example, I tend to focus too much on corner exits, and don't have high entry speeds (imagine merging onto a highway.) My driving style is somewhat "point and squirt." However, in GT5, I'm left with no idea where I'm losing my time, and whether I'm using my grip to the utmost efficiency.


While I agree with your idea, there are bigger fish, in my opinion.
 
I'm afraid it would work just as good as the penalty system in GT5:

Driver A gets smacked off the road by Driver B.
Driver A gets the penalty.
It's hard for an artificial intelligence to judge who was to blame for an incident (it's even hard for human intelligence sometimes) and it will certainly get it wrong lots of times.

This will lead to new levels of road rage, for Driver B did not only ruin Driver A's lap, but also his hard earned ranking.

Also, what should be recognized as bad driving and what shouldn't? Some say a little contact is fine (or even good) while others say it's strictly forbidden. What is the proper definition of corner cutting and is it always bad driving to go off the track? (If there's a stationary car on the track, might have spun around or something, and to avoid collision you need to go off track). If GT6 includes more realistic damage, is it bad driving to get a flat tyre that causes you to lose control and crash into the others?

There's a lot of parameters to take into consideration. Even if everything would be technically possible to control with 100% accracy (which it's not), the biggest problem would be that PD would have to decide the rules. And there would be 25% of players who think the rules are perfect, 50% who thinks they are okay and 25% who thinks they are the manifest of Beelzebub.

I think that the best option is to let the community regulate itself, because different drivers prefers different rules in different races. Let the host set the rules and if they see something that's not okay they get to chose how to deal with it.

My original premise was so that GT5 can become more of a teaching/training system like GT Academy. In my mind, using the ratings for online filtering was only a side effect.

I think that the license tests works well enough for this purpose. They teach you the basic theories. A better option than implementing a ranking system would in my opinion be to develop the license tests and the driving school events. Because only having a rating of the drivers would be like having a school with no classes but only grades, you'll sort the clever kids from the not so clever but no one will actually learn anything.
 
Last edited:
So if I want to do mayhem offline, but I'm a absolutely clean driver online, it will be recorded that I'm dirty? Count offline doesn't make sense at all.

I can understand that, but who does "mayhem" in their A-spec career? Normally wouldn't you go to Arcade mode for that, or at the very least just go to free run? I'm not saying that all offline driving should impact your rating, just your performance during the A-spec race events.

And in the event that maybe you do want to just goof off during an A-spec event, there could be an option allowing you to disable offline rating for a particular race so that the game knows that you're just trying something silly out (like taking on supercars with a Midget) and not to count it towards your rating.

I won't even respond to your other points, because they simply don't make any sense.

A player's average speed is a single number. That doesn't make sense?

A player will have lap time records for multiple tracks. That doesn't make sense?

Using a single number (average speed) is simpler than using multiple numbers (lap time records for multiple tracks). That doesn't make sense?

Using a slower car will result in slower lap times. That doesn't make sense?
 
A player's average speed is a single number. That doesn't make sense?

A player will have lap time records for multiple tracks. That doesn't make sense?

Using a single number (average speed) is simpler than using multiple numbers (lap time records for multiple tracks). That doesn't make sense?

Using a slower car will result in slower lap times. That doesn't make sense?

From what I'm reading it seems you want to set a skill level based on average speed, which doesn't make any sense. If someone races on K-cars and is a great driver, it'll show that he's worse than one who simply just races NASCAR?
 
I would agree with most replying, that to do a system like Iracing wouldn't work for the casual racer. It would be great to see, but I doubt it would make it.

I'd rather see just some driver stats for online racing. So instead of your welcome message when you join a room, you get something like this.

Driver Name: DS3/Wheel, A-spec Level, Licenses G/S/B, # online races finished/# online races started, average finish position, number of penalties (online career), average # of penalties per race (online career), number of penalties last 10 races.

~~~chat box example~~~
Driver Stats: Wheel, 39, 32/28/0, ORF/ORS .902 , AFP 5.75, TP 35, APR 0.171, PL10 3
~~~~~~~~~~~~~~~~~

Also you could also look at each other racer's stats while in the lobby by going to their name on the driver lists. So if someone had started off bad in racing online, they could at least have something to their defense instead of just the numbers that come out when you join a room. Maybe we could see more info from their last 'x' races. (Started, Finished, Penalties.)

~~~Detailed Example Using 10 races as a filter~~~
Controller Method: DFGT
A-Spec Level: 39
Licenses: 32 Gold, 28 Silver, 0 Bronze
Career Races Started: 205
Career Races Finished: 185
Career Races Finished/Started .902
Last 10 Races Finished/Started 1.00
Career Average Finishing Position 5.75
Last 10 Average Finishing Position 4.33
Career Total Penalties 35
Last 10 Total Penalties 3
Career Average Penalties Per Race 0.171
Last 10 Penalties per Race 0.3
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

Granted I just made up numbers. Something like this, of course things could be added/tweaked would be nice to see when someone joins the room.

There would always be the people that would abuse it. I could run races with very few people to make my AFP look better. More so if I was looking at someone joining a room I was in, I'd look at a few key numbers.

Races Finished/Races Started
Total Penalties
Average Penalties per race
# Penalties last 10 races.

The last 10 races, could be extended to cover 20-50 races, but I felt 10 was a good number.

Sure people would get screwed by a few things:

A collision penalty when they didn't do anything (ie, punted into another car)
A shortcut penalty that's to close to the track edge (Fuji has bad penalties in my opinion)
Disconnects

That sucks, and life sucks sometimes.

There would still be an option to have online rooms not stat track.

Sorry for the length, but my two cents on the matter.
 
From what I'm reading it seems you want to set a skill level based on average speed, which doesn't make any sense. If someone races on K-cars and is a great driver, it'll show that he's worse than one who simply just races NASCAR?

What I'm saying is that skill ratings based on lap times would be just as biased in favor of people who drive faster vehicles as ones based on average speed, so if either method should be used, it should be using average speed since it's simpler.

But I understand your concerns, and fortunately there's a possible way to reduce the bias against slow cars. You just have the player's rating based off a ratio between their average speed with a vehicle and the top speed of the vehicle. For example, a person who does an average speed of 60 with a vehicle capable of doing 110 would be rated the same as a person who does an average of 90 with a vehicle capable of 165.

Of course now you may be thinking something like "well now you've gotta keep track of average speed for every single vehicle the player drives. How is that any better than having to take into account lap times for various tracks? There's way more cars than there are tracks..."

But you wouldn't need to keep track of each vehicle individually. You just need three numbers to keep track of a player's performance in this way: Global average speed, Global average vehicle top speed, and total number of miles driven.

At the end of a race, you just use the amount of miles they drove in the race and the amount of miles they had driven before the race to average in the player's average speed for the race and the top speed of the vehicle they used in the race with the global averages used for calculating the player's rating.
 
Last edited:
How would that be any different that they way you explain it before.

If I do 20 laps on SSRX in a car with max speed of 200mh, then my average for that race is 200mph.

Next race, 20 laps on London with a 400pp car that might get 100-120mph top speed, but I won't get that on that track?

So let's say on the first race I would average exactly 200mph speed, because I'd have to start, but 20 laps of SSRX would be pretty close.

We'll say 190mph average speed
200mp top speed
18.817 miles x20=376.34 miles driven on 20 laps SSRX

Then we go to london.

Average speed on a 400pp car at london? 60? Let's say 60
Top Speed? 80-90
1.19 miles x20=23.8 miles driven on 20 laps at London.


SSRX/London
190/60 average speed
200/85 top speed
376.34/23.8 miles driven

I guess I don't understand how this method would you be able to determine anything about the player's ability. Without having to track each car on each track separately.

1072 cars (from GT5 might not be including DLC)
81 track configurations (from GT5 might not be including DLC)

86,832 track/car combinations. If my math is correct.

Granted we're talking about GT6 and maybe we won't have that many cars, but more tracks.

I guess I just don't understand.
 
How would that be any different that they way you explain it before.

If I do 20 laps on SSRX in a car with max speed of 200mh, then my average for that race is 200mph.

Next race, 20 laps on London with a 400pp car that might get 100-120mph top speed, but I won't get that on that track?

So let's say on the first race I would average exactly 200mph speed, because I'd have to start, but 20 laps of SSRX would be pretty close.

We'll say 190mph average speed
200mp top speed
18.817 miles x20=376.34 miles driven on 20 laps SSRX

Then we go to london.

Average speed on a 400pp car at london? 60? Let's say 60
Top Speed? 80-90
1.19 miles x20=23.8 miles driven on 20 laps at London.


SSRX/London
190/60 average speed
200/85 top speed
376.34/23.8 miles driven

I guess I don't understand how this method would you be able to determine anything about the player's ability. Without having to track each car on each track separately.

1072 cars (from GT5 might not be including DLC)
81 track configurations (from GT5 might not be including DLC)

86,832 track/car combinations. If my math is correct.

Granted we're talking about GT6 and maybe we won't have that many cars, but more tracks.

I guess I just don't understand.

So... if I'm understanding your example correctly, your concern is that some tracks are faster than others and thus would have a greater effect on a player's rating?

This would be a correct assessment. But your solution of having to keep tabs on each car/track combination possible is incorrect. A more ideal solution is to come up with an algorithm for calculating a car's approximate top average speed theoretically possible on any given track and use that instead of the car's actual top speed when calculating the ratio which'll be factored into the player's rating. This way, we don't have to keep track of each individual car/track combination, all we need is global average speed, total miles driven, and a clever algorithm for calculating a car's theoretical max average speed possible on a given track.
 
here's more of what i was thinking...

each track would have a predefined racing line with acceleration and braking points for each of the PP car classes. your adherence to the racing line (and other negatives like collisions) are used to create the rating for that driver during that race with that PP class car. i thought the gt academy concept could be used to show a coach/judge explain key points in your race by comparing your line and pace vs the defiined racing line... even show (through creative replays) how your line differed and what to change so that your next attempt is better - like a training session or academy.

also, a running average driver rating over the most recent X races would be your current driver rating and could be used 1) to measure your progress, 2) by online races to filter drivers, 3) whatever else the creatives at PD could come up with.
 
The way for implementing time trials that influence rating is just like the gt5 online special events (afaik how they are): one car that cannot be tuned in any way, fixed assists (hopefully none including abs off), best time in one lap. One event per each car type (400pp sports cars, gt500, f1, etc.).

Those cannot be the only factor to be considered though, as lots of people will want to skip those then win all races against noobs, but this is one of those things in which copypasting other people's ideas works best (iracing and other gaming genres too).
 
A more ideal solution is to come up with an algorithm...

Rather than that, why not use the proven strategy of bell curves? Sure, the first few days to 1st couple of weeks will be wonky, but after that you have a reliable self sustaining system.

I still don't like the driver rating idea though.

glw
here's more of what i was thinking...

each track would have a predefined racing line with acceleration and braking points for each of the PP car classes. your adherence to the racing line (and other negatives like collisions) are used to create the rating for that driver during that race with that PP class car. i thought the gt academy concept could be used to show a coach/judge explain key points in your race by comparing your line and pace vs the defiined racing line... even show (through creative replays) how your line differed and what to change so that your next attempt is better - like a training session or academy.

also, a running average driver rating over the most recent X races would be your current driver rating and could be used 1) to measure your progress, 2) by online races to filter drivers, 3) whatever else the creatives at PD could come up with.

You had better pray that all your races are boring then. Get into a competitive race, and there goes your rating. Also, try not to go as fast as you can. If you find a line better than the predefined line, you're just going to be punished for it.
 
Infinital-NG
Sure people would get screwed by a few things:

A collision penalty when they didn't do anything (ie, punted into another car)
A shortcut penalty that's to close to the track edge (Fuji has bad penalties in my opinion)
Disconnects

That sucks, and life sucks sometimes.

There would still be an option to have online rooms not stat track.

Sorry for the length, but my two cents on the matter.

Yes, getting penalties you didnt deserve will always happen, but in the long run the clean drivers will easily separate themselves from the bad drivers with their penalty per race rating. If they are intentionally wrecking others (and giving both penalties), or just flat out wall surfing or cutting the course, in time its going to catch up with them where their penalty rating is so bad its hard for them to find a race that accepts thier poor penalty per lap rating.

Of course the option to reset your penalty-per-race rating should always be present if you want a fresh start.
 
I think this is an important topic to bump... we need something that's roughly along the lines of driver safety rating or some kind of ranking system to be implemented in GT6.
 
Back