The GGR forum has reached it's end, but posts are still available read-only for posterity. We invite you to make posts and discussions on the GGR Facebook page.

Why do base points differ so much from current PAX index

Use this forum to discuss the Points Proposal

Moderator: David Leong

Post Reply
User avatar
Bill Dally
Posts: 18
Joined: Tue Jun 28, 2005 9:31 pm
Location: Palo Alto

Why do base points differ so much from current PAX index

Post by Bill Dally » Thu Aug 18, 2005 1:38 pm

The current PAX index seems to be quite fair as evidenced by different cars competing effectively for PAX points. However the current proposal gives base points that are quite different from the balance achieved by PAX -- penalizing the mid-engine cars. For example a 914/2.0, a 69 911E, and a 924 turbo all get 300 base points, but the PAX indices for these three are .8810, .8856, and .8916.

What is the rationale for deviating so much from an index that has proven its value over time?

dtfastbear
Posts: 89
Joined: Wed May 11, 2005 10:21 pm
Location: San Anselmo, CA
Contact:

Post by dtfastbear » Mon Aug 22, 2005 2:20 pm

Bill,

What do you believe a reasonable spread in base points should be given that in a normal 50 second autocross course the current PAX index spread you mention results in about half a second difference across those three cars?

Do you believe that these cars could not compete against each other fairly in a single class, but that handicapping one of them by half a second and the other by a quarter of a second makes it an obviously fair battle that has been "proven over time"?

That seems to be splitting hairs, don't you think?

Dean

BTW, I significantly overhauled the PAX system in 2004, so the current indexes are only two years old.
Faster than the average bear...

User avatar
Bill Dally
Posts: 18
Joined: Tue Jun 28, 2005 9:31 pm
Location: Palo Alto

The issue is ranking, not granularity

Post by Bill Dally » Thu Sep 29, 2005 6:56 pm

Dean,

The issue I'm bringing up with points vs PAX is one of ordering or ranking of cars, not one of granularity.

If we aren't worried about 0.01 point differences in the PAX, we can improve competition a lot by collapsing adjacent classes that differ by less than this. B,F, and G would become one class, C,D,E,H,J, and K would become one class, L, M, and T one class, etc...

The PAX system and current rules system orders cars in a way that is consistent with results seen in events. The new rules system seems to ignore this ordering in assigning base points.

The deviation seems limited to mid-engine cars, the only ones that have different TT and AX points.

====Bill

dtfastbear
Posts: 89
Joined: Wed May 11, 2005 10:21 pm
Location: San Anselmo, CA
Contact:

Post by dtfastbear » Thu Sep 29, 2005 8:35 pm

Hey Bill,

Sorry I missed you point the first time - I see what you're saying, now.

The GGR rules state that "Indexes shall be based on car class historical data collected from past autocrosses with all times corrected to Modified and Exhibition cars times. " I created the current index two years ago using a combination of historical GGR data from the previous 3 years, common sense and a highly technical, scientific process I like to call "fooling around with Excel until the results looked right to me". :)

I'm proud of that index and the close competition it has fostered among what I believe to be some of the best drivers in our region. But, one of the reasons this system works well for GGR is that it was created FOR GGR specifically. As much as I wanted PAX to be a car handicapping system, the rules state that it be at least partially based on historical data, which means that drivers matter, making it a driver handicapping system. The result is that classes where we don't have much data with outstanding drivers get a slightly favorable index while classes with one or more very, very good drivers get a slightly tougher index. Having so few cars in many of the classes exasperates this. While I took some liberty with smoothing and adjusting the indexes to eliminate what was once called the "Brooksby effect" (because Glenn Brooksby's fabulous driving gave Li one of the highest indexes of any class), driver performance is still influentual. And hence, you can expect close PAX competition in GGR.

So, here's my point: the points proposal is our attempt at making a CAR handicapping system while the PAX index is largely a GGR AX DRIVER handicapping system. I fully believe that if we gave our rules and PAX index to a Northeast region, for instance, where there is a different mix of cars and drivers, the index would fall apart quickly.

So, while I agree with you that collapsing the classes you mentioned would result in close GGR AX results, I don't agree with the fundamental premise that this is a fair way to group or order cars, necessarily.

I don't know if you've seen all the analysis that Andrew Blyholder has done on the AX results to date. The results based on the points system show have a tighter spread between finishers than the current system - the competition is close!

Thanks for taking the time to express your opinions and ideas, Bill. I hope that the points system is adopted and that the general club will turn its collective wisdom and attention to tuning and improving it. I believe it is already vastly superior to the current system, and with more tuning it will only get better.

Dean
Faster than the average bear...

Post Reply

Who is online

Users browsing this forum: No registered users and 3 guests