Will Perfect-N Work Well

Debates and discussions on the various race scheduling methods that can be used and their fairness and accuracy in determining the winners.
Post Reply
User avatar
PWD_addict
Master Pine Head
Master Pine Head
Posts: 712
Joined: Mon Jan 22, 2007 11:35 am
Location: Middle River, Maryland

Will Perfect-N Work Well

Post by PWD_addict » Mon Feb 01, 2010 2:01 pm

Our Pack runs timed races with the lane rotation method. Average time with slowest tossed out determines winners. I know that Perfect-N is more exciting for the boys since they don't race against the same ones each time. We like to run the ranks through first, and then advance the fastest from each to the Grand Finals. Is this doable with Perfect-N scheduling?

The primary challenge that I see is that staging the cars is not as predictable as lane rotation. Is that correct? Anything else I'm missing?

Does Stan have a good post explaining how Perfect-N works? I can't access his website from my work computer. :(

Oh yeah--about 50 Scouts.



User avatar
Darin McGrew
Pine Head Legend
Pine Head Legend
Posts: 1735
Joined: Fri Jul 18, 2003 1:23 pm
Location: Mountain View, CA
Contact:

Re: Will Perfect-N Work Well

Post by Darin McGrew » Mon Feb 01, 2010 2:43 pm

First, a few definitions:

Perfect-N (PN) - each car races the same number of times in each lane, and the same number of times against every other opponent

Complementary Perfect-N (CPN) - Perfect N, with the additional requirement that any race between two cars has a corresponding race in the schedule with the two cars racing again with the lanes reversed

Partial Perfect-N (PPN) - each car races the same number of times in each lane, and races against different opponents as much as possible

If you're just trying to race the cars against different opponents, then PPN is what you need. PN and CPN apply additional constraints (which are necessary when you're using finish order to determine results, but not when you're using times).
PWD_addict wrote:We like to run the ranks through first, and then advance the fastest from each to the Grand Finals.
Is there anything different about the times you collect during the Grand Finals? Does each car race more times than in the qualifying races? Do you combine the times from the qualifying races with the times from the finals so you have more times (more data samples, and thus, more accurate results)?
PWD_addict wrote:The primary challenge that I see is that staging the cars is not as predictable as lane rotation. Is that correct? Anything else I'm missing?
Yes, but you can generate the schedules in advance. A plus of CPN-like scheduling over lane rotation is that you can avoid racing any car in two consecutive races, which allows you to start staging the next race before the kids retrieve their cars from the finish line.
PWD_addict wrote:Does Stan have a good post explaining how Perfect-N works?
Yes, there are good articles describing the various scheduling systems in Pope's Pinewood Pages Portal.



User avatar
PWD_addict
Master Pine Head
Master Pine Head
Posts: 712
Joined: Mon Jan 22, 2007 11:35 am
Location: Middle River, Maryland

Re: Will Perfect-N Work Well

Post by PWD_addict » Mon Feb 01, 2010 3:37 pm

Darin McGrew wrote:
PWD_addict wrote:We like to run the ranks through first, and then advance the fastest from each to the Grand Finals.
Is there anything different about the times you collect during the Grand Finals? Does each car race more times than in the qualifying races? Do you combine the times from the qualifying races with the times from the finals so you have more times (more data samples, and thus, more accurate results)?
The Grand Finals are a separate round so the previous qualifying round doesn't count for anything but to determine Grand Finalists. Everyone races in each lane one time. No different than the qualifying rounds.

I don't know that I'd agree that your definition of "More Accurate" is necessarily a better way to hold the races. Maintaining a consistent speed seems to be more challenging than having a really fast car for a few races, which then tapers off in speed. In our Pack and District races, there are very few that can manage the former, but many that manage the latter.
Darin McGrew wrote:
PWD_addict wrote:The primary challenge that I see is that staging the cars is not as predictable as lane rotation. Is that correct? Anything else I'm missing?
Yes, but you can generate the schedules in advance. A plus of CPN-like scheduling over lane rotation is that you can avoid racing any car in two consecutive races, which allows you to start staging the next race before the kids retrieve their cars from the finish line.
Since we race by rank, we never have more than 16 cars at a time in a particular rank round. But, if we did, that would be a much quicker way to race. It's the running the cars from the finish to the start line that takes a fair amount of time. We could be staging the next race while that was going on. Oh, and we have an 8 lane track. That's significant to the explanation.
Darin McGrew wrote:
PWD_addict wrote:Does Stan have a good post explaining how Perfect-N works?
Yes, there are good articles describing the various scheduling systems in Pope's Pinewood Pages Portal.
I know there are lots of good things on Stan's website but it is blocked from work and I am into instant gratification and wanted to see something NOW. :D
But, you explained it perfectly. Thank you Darin.



User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6888
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Will Perfect-N Work Well

Post by Stan Pope » Mon Feb 01, 2010 8:18 pm

Darin told the story well. Thank you, Darin!

Sadly, the online page at http://www.stanpope.net/ppngen.html and (I think) the DDL version that is included in GPRM, et.al., only support up to 6 lane tracks. :( Since Cory Young is the "keeper of the code" for the DDL version, I'll have to let him give the final say on the DDL version capabilities... or, maybe, someone who has GPRM readily accessible on their computer!

A subtle value to the PPN approach for timed racing is that, absent errors in timing, there is a high correlation between "finish order" dependent results (points) and time ranking. This high correlation is a result of the nearly equal opponent mix that each racer faces. Since finish order can be "validated" by the audience, the audience can satisfy themselves that points results are reached accurately. If they can keep up with the points stats, then they can compare points ranking with time rankings and develop confidence in the time rankings. (As one whose district was bitten badly by timing errors by a supposedly experienced crew, I put a pretty high value on the audience being able to validate the official results from their own observations.)

Simple lane rotation, on the other hand, does not have a high correlation between ranking by heat points and ranking by heat times. Consequently, it is harder to show that timing is consistent throughout the races.

In between these extremes, a wide variety of chart generator vectors could be used which provide more balance in the competition than does simple lane rotation, but less balance than proper PPN. Even randomly produced generators (provided some simple criteria are adopted to exclude certain random choices) would do far better than simple lane rotation!

If you have consistent timing apparatus and well trained operators, then 8 times turned in for each racer in age-group prelims should be neither more nor less valid than eight times recorded by each participant in the finals. (To the extent that there is random variation in the times, then the times in the finals will moderate from their prelim results.) Even if there is no more validity, it is reassuring to see prelim times reaffirmed by times in the finals. The "clinker" in this is the pattern of run times as teh cars accumulate heats. Some degrade more rapidly than others, and the cars with "staying power" will tend to do better if there are finals than if their prelim times are compared.


Stan
"If it's not for the boys, it's for the birds!"

User avatar
PWD_addict
Master Pine Head
Master Pine Head
Posts: 712
Joined: Mon Jan 22, 2007 11:35 am
Location: Middle River, Maryland

Re: Will Perfect-N Work Well

Post by PWD_addict » Tue Feb 02, 2010 7:14 am

Stan Pope wrote: The "clinker" in this is the pattern of run times as the cars accumulate heats. Some degrade more rapidly than others, and the cars with "staying power" will tend to do better if there are finals than if their prelim times are compared.
That seems to be what happens with my son's cars. It is able to maintain it's times throughout the races. #2 had an average speed of 4.1409s during prelims. Son had 4.1461 during prelims. In Grand Finals, son had 4.1300. #2 was 4.1326. We drop the slowest time.

Now, during my son's rank races, there was a bad joint in lane 4 where his car jumped lanes twice in a row. Second time, it didn't hit any other cars so we let the time stay. Obviously, this was the time that was dropped. After his rank, we repaired the joint (foil tape). During the Grand Finals, he had an average run in that lane. The #2 boy raced after we repaired the track so it would not have been fair if we would have counted the times from the preliminary heats towards the final standings.

We have a very rough track and my son's standard deviation (dropping the aberration) was 0.017s. The #2 guy's stdev was 0.025s.

Honestly, with the new wheels, I expected a significant increase in speed. We didn't see it. Perhaps it's because we didn't go as far with axle prep as we have in the past but I only suggested my son go as far as I thought he was willing. ;)



Post Reply