Help the New Guy - Part I

Debates and discussions on the various race scheduling methods that can be used and their fairness and accuracy in determining the winners.
Nooby
Pine Head
Pine Head
Posts: 99
Joined: Wed Aug 31, 2005 8:55 am
Location: Southwest, MI

Help the New Guy - Part I

Post by Nooby »

:x

This is my son's second year of cub scouts - and I've been tasked with running the pinewood derby this year (I assisted last year). I loved the derby when I was a kid, and it still seems like a highlight of the scout year (along with campouts).

Our setup

3 lane wood track - lanes seem very evenly matched
-manual starting gate

Electronic "judge" finishing gate - not hooked to a computer - gives 1-2-3 display. Only one race was re-run due to percieved error in the electronic scoring

Our race

60-70 scouts.
Double elimination on 3 lanes - i.e. fastest 2 cars advance

Lane rotation - manual - colored tick mark each time a car on a certain track - tried to 1) not race on the same lane twice, 2) race equally on all lanes - seemed to work as nobody complained :)

The problems:

1. Double elimination - some kids are out in the first few minutes.
I would like to see more racing for everyone

2. Our "bracket" was made by placing cars on a table - No display for the scouts or audience - hard to follow. This could easily be overcome by having some type of written bracket.

My "Proposed" Solution

I used to play in MtG tournaments, which used a swiss scoring system to match players with similar records. I'm thinking about a similar system for the derby.

Each car races - winners get 1 point first, 2 points for second, 3 points for third.

People with similar point totals race - winners race winners, losers face losers.

After X rounds (I have no idea what X should be - but I'm hoping to have at least 6 rounds) - the scores are totaled.

Because I am afraid, that the results will oscillate - ie top guy would knock number 2 and 3 down, the top 9 people would advance to an elimination round. 9 seems like a nice number for a 3 lane track. Each of the elimation races would be run 3 times - one in each lane. Each winner would advance to the finals to determine 1, 2, 3 for trophies.

Positives-
more races
equal cars race each other.

Negatives
Osillation of score (see above)
equal races may cause more disputes.
How to determne den winners - could run first few rounds only within the den (our pact awards first and second prize for each den)

I've looked at the Stern's and other methods and don't like the randomness of the matches, but like the fact that everyone races until the end.

If anyone has tried this, or sees any blatant defect - please let me know.

I'm interested in all input - because we want to have the fun derby for everyone.

Also, I need solutions to use our existing equipment - No money in the budget at the present time - maybe 2007
"Nooby"
User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6856
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Help the New Guy - Part I

Post by Stan Pope »

Given those numbers of racers all racing each other, I'd be inclined to run one of the following:

1. Multiple elimination, e.g. 4 or 5. This works best if boys stage their own cars, then 5-elim probably needs about 3 hours. Good aspect is that it can easily be done with no charting and late arrivals can be incorporated into racing with proportionate penalty. Eliminating the paperwork actually speeds up the process!

Write-up is available at http://members.aol.com/standcmr/nelim.html

2. Charted PPN. Race one or two rounds to select finalists, about twice as many finalists as you have trophies to award. Then run PN or CPN final. Best number of finalists for 3-lane track are 7 and 13. This format has one heat per round for each racer entered, i.e. 2 rounds 60 racers requires 120 heats, and each racer races once in each lane for each round. Characteristics are complete lane balance and good-as-possible opponent balance.

Charts available at http://members.aol.com/standcmr/ppngen.html

These charts are also incorporated in some race management software. Maybe Randy can tell you the name of one. :)


Some notes on your suggestions:

9 cars is not a good number to run in a final ... in a reaonable number of heats, it fails either lane balance or opponent balance. On 3 lanes either 7 or 13 cars is good. Choose a number that gives a very high probability that the fastest N (N is number of trophies) cars participate in the final. A computer simulation of accuracy of the Swill Ladder would be in order before deciding. I think that the finals group will have to be 3 or 4 times the number of trophies to assure that the deserving are in the finals. (I've run chess tournaments -- 4 and 5 round affairs -- using that method and the need to do the paperwork between rounds would be an issue for 60 or 70 entrants! Also, you lack a basis such as USCF rating for ranking the entrants within each score-group... they would probably have to be randomized.)

Getting Den Results can be added onto the PPN plan by sorting out the highest scorers in each Den or Age Group. Route them into corresponding finals, using a slightly larger ratio of finalists to trophies, perhaps 3 times as many finalists as trophies.

The nochart multiple elimination format does not lend itself to identifying Den winners.

Alternatively, you can race by age group using PPN or whatever. Then hold a "race of champions" pitting the 1st place cars from each age group for a single trophy. This doesn't work if you must find more than one "fastest car in the pack."
Stan
"If it's not for the boys, it's for the birds!"
Nooby
Pine Head
Pine Head
Posts: 99
Joined: Wed Aug 31, 2005 8:55 am
Location: Southwest, MI

Re: Help the New Guy - Part I

Post by Nooby »

Stan,

I think the multiple elimination is more along the lines of what we're looking for - except for the lack of picking the den champions.

I really like the scoring simplicity.

PPN and Sterns just seem like scheduling nightmares, whereas the multiple elimination seems pretty simple.

However, I'm not seeing the advantage of having the boys stage their own cars - seems like a lot of kids walking around the track - with all the potential of knocking the track out of level.

Now, How do we pick Den champions?

With 5 to 12 scouts in a den (last year's numbers). Could we run say 3 rounds within the den, determine the den champion and runner up. Afterwards, the den scores would be used as a starting point in the pack race

I need to think some more about this
"Nooby"
User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6856
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Help the New Guy - Part I

Post by Stan Pope »

For small (den size) groups, I like final standings like PPN. Note that PPN and Stearns are similar in concept, but PPN has fewer "bad schedules". PPN is generally well behaved.

Such charts are not "scheduling nightmares." Prestaging heats by car number allows races to be run and scored expeditiously. Showing scoring by projection keeps everyone aware. (I used transparencies and overhead projector on the gymnasium wall!)

We've run our district races for as long as I know (at least 20+ years) with the Cub Scouts (Tigers and up!) staging their own cars. Track bumping has not been an issue. It is important to have the Cubs as participants rather than spectators! Currently we provide bleacher seats for family on one side of the track and an infield for groups of "drivers" opposite, mostly sitting on the floor. Participants in the current heat have staged their cars and are awaiting them at the finish line (best seat in the house!) Waiting participants in the current round are queued (in as random order as possible) leading to the pit / starting line area.

For "pack winners"... how many do you need to identify? e.g. three fastest in the pack? Are there any limits on how many pack winners can come from various age groups? If you need just one "fastest in the pack" then run the den winners to a final. If you need "fastest three in the pack" then running the three fastest from each den can produce some anomalies such as the third-fastest Tiger taking first in the "fastest in the pack" competition.
Stan
"If it's not for the boys, it's for the birds!"
RMoose
Master Pine Head
Master Pine Head
Posts: 125
Joined: Sat Apr 17, 2004 9:30 pm
Location: North Carolina

Re: Help the New Guy - Part I

Post by RMoose »

Nooby,

I would agree with Stan that PPN would be the best way to go with small groups, in fact we are considering that approach for our race this year. Nonetheless, you can use multiple elimination with smaller groups.
Last year we ran four groups of boys and girls using triple elimination. We were looking for the top three in each age group. The numbers went like this:
boys grades 1-3 = 21 participants
boys grades 4-6 = 7
girls grades 1-3 = 10
girls grades 4-6 = 8

Three of these groups fall into your range of 5-12 members. We ran each group separately using no chart 3E based on Stan's methods, awarding the top three places in each group. We did not do an overall grand champion or anything like that. We finished in just 2 hours and the race went well - received lots of positive feedback from parents.

From my experience I would say if you do not feel comfortable with PPN and like the sound of multiple elimination give it a try! I would also second the idea of a projector to get the progress and results up where everyone can see. This is a big help! Also, be sure everyone who will be helping run the race is on the same page. We did a dry run last year before our race and this proved to be very helpful to us.

Happy Racing! whatever method you decide on!
We are Ambassadors for Christ
User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6856
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Help the New Guy - Part I

Post by Stan Pope »

RMoose wrote:Also, be sure everyone who will be helping run the race is on the same page. We did a dry run last year before our race and this proved to be very helpful to us.
Excellent suggestion! Seems obvious, but so many folks skip over this one! Even if you give the staff detailed oral or written instructions, hands-on training / experience is so much more powerful.

With larger groups, we run three cars at a time. With smaller groups, that pares the groups sizes down too quickly, and you should consider running 2 cars at a time instead.

You can analyze it by looking at how quickly the group of unbeaten cars reduces to a single car. Here's some examples:

Racing 3 at a time:
Rounds - car numbers
1 - 1 to 3
2 - 4 to 9
3 - 10 to 27
4 - 28 to 81
5 - 82 to 243

Racing 2 at a time:
Rounds - car numbers
1 - 1 to 2
2 - 3 to 4
3 - 5 to 8
4 - 9 to 16
5 - 17 to 32

So, if you are running a 5-elim with 6 cars, the last unbeaten car will face 2 competitive heats if the cars race 3 at a time and 3 competitive heats if the cars race 2 at a time.

The other issue is how many total heats will be needed to finish racing, and there is a javascript on my webpage to compute that for you.
Stan
"If it's not for the boys, it's for the birds!"
Nooby
Pine Head
Pine Head
Posts: 99
Joined: Wed Aug 31, 2005 8:55 am
Location: Southwest, MI

Re: Help the New Guy - Part I

Post by Nooby »

Historically, our pack has awarded the following:

1st and 2nd in Den based on Double elimination

1st, 2nd and 3rd in pack based on double elimination of the top 2 from each den. And yes, I know that double elimination does not accurately determine the 3rd place finisher. - but I didn't set up the system.

While it may appear anomolous, I personally would not have a problem with a 3rd place in the den taking 1st in the pack. To me, that would just mean that we have an extremely competitive field.

I observed a causual pinewood derby this summer. The 3rd and 4th place cars were about as evenly matched as possible. A really awful scheduling method was used (don't ask) whereby these cars raced each other at least 6 times prior to the finals. In fact, they only raced separately once (20 cars participated). They split first and second 50-50%. In the finals, they were in a run-off to determine 3rd place and it came down to a 2-1 decision.

It was more exciting than the championship race. :P
"Nooby"
User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6856
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Help the New Guy - Part I

Post by Stan Pope »

Nooby wrote:While it may appear anomolous, I personally would not have a problem with a 3rd place in the den taking 1st in the pack. To me, that would just mean that we have an extremely competitive field.
What this usually shows is inaccuracy in one (or both) of the racing methods used. It might also show differences in lubrication life profile.
Stan
"If it's not for the boys, it's for the birds!"
Nooby
Pine Head
Pine Head
Posts: 99
Joined: Wed Aug 31, 2005 8:55 am
Location: Southwest, MI

Re: Help the New Guy - Part I

Post by Nooby »

Definitely a flawed race method. The race used a scoring system similar to to mulitiple elimination, but did not seem to match cars in any order other than random - with many nearly identical heats. I think they were trying to use a program called Derby Master, but failed to get it running, and instead tried to mimic the program manually without understanding how the program worked.

While I imagine there are some variable such lubrication life , lane selection and car staging - I believe that there is a significant amount of randomness that can make racing a toss up between 2 given cars.

Also, I observed one of the racers being quite rough on his car between heats. The race was run outside in a tent, and the scout was rolling his car on an asphault surface - which may have degraded the performance of that car.
"Nooby"
User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6856
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Help the New Guy - Part I

Post by Stan Pope »

Nooby wrote:Definitely a flawed race method. The race used a scoring system similar to to mulitiple elimination, but did not seem to match cars in any order other than random - with many nearly identical heats. I think they were trying to use a program called Derby Master, but failed to get it running, and instead tried to mimic the program manually without understanding how the program worked.
This situation is instructive! In years gone by, Cub Scout Leader Basic Training emphasized the concept of "Plan B" which, in summary, recognized that something is apt to go wrong with the primary plan and that leaders should have a well-developed alternative plan. I have not seen this emphasis in more recent training.
Nooby wrote:While I imagine there are some variable such lubrication life , lane selection and car staging - I believe that there is a significant amount of randomness that can make racing a toss up between 2 given cars.
A lot depends on the cars and the track. At the Sycamore Invitational a few years ago, two high-quality cars ran 4 heats on alternating lanes. The cars were staged by their owners. The same car won all four heats with winning margins of about 1/8" to 3/8", measured by eye. The heats were conducted using a well-finished 32' Piantedosi wooden track.
Nooby wrote:Also, I observed one of the racers being quite rough on his car between heats. The race was run outside in a tent, and the scout was rolling his car on an asphault surface - which may have degraded the performance of that car.
We try to emphasize careful handling of the cars, but sometimes it falls on deaf ears. :(
Stan
"If it's not for the boys, it's for the birds!"
ronin718
Journeyman
Journeyman
Posts: 40
Joined: Tue Sep 13, 2005 11:39 am
Location: Sterling, VA

Re: Help the New Guy - Part I

Post by ronin718 »

Nooby... I ended up running or helping to run two PWDs for a couple of packs last spring. One pack had something like 40 boys, the other only 13. The pack with 40 ran on a four-lane track, the one with 13 had a three-laner.

Rather than run elimination races, we had an Excel program that randomized the lane assignments, ensuring each car ran on each lane. The results of each race were recorded, and the program calculated the results based on the finishes of each heat. The 40-boy pack did it for each rank (Tiger, Wolf, Bear, Web I, and Web II), and the top three from each rank then went to the Finals. The 13-boy pack all competed against each other, and because the numbers were so low, the program was unable to break it down beyond a three way tie for first. These three then ran a final.

In the finals for both packs, the same process was done. Each car ran on each lane, and the results of those heats determined the overall winners. This method allows the boys to run at least as many races as you have lanes, and nobody gets the "two and out" toss.
User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6856
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Help the New Guy - Part I

Post by Stan Pope »

Ronin718, you are on your way toward good handling of a "final standings" method. I don't know the details of your specific EXCEL spreadsheet, although some, such as implemented by Cory Young, go beyond simply achieving complete lane balance and approach complete opponent balance. He did this by having the spreadsheet acquire the race chart from the PPN Chart Generator, AKA Young&Pope.

Even with these charts, the rankings at the end of a reasonable duration race are subject to some errors. When opponint balance is incomplete, the error rate is somewhat higher. What this means is that for many of the combinations of number of lanes and number of racers, the final standings scores are only approximate.

The selection of finalists should try to include those who, by virtue of "bad match-ups", fell somewhat short of the top scores. If you are planning to award three trophies within a category and would like to award those to the three most deserving, then you must take the accuracy of the method into consideration when choosing the number of finalists. Choosing only the three tied racers as finalists does not achieve this goal. With PPN charts, to get high probabilities (into the high end of the 90% range) of including the N fastest cars into the finals, it is usually necessary to include approximately 2N (2 times N) highest scoring cars in the finals.

I've written quite a bit about this issue and how to address it analytically at http://members.aol.com/standcmr/pwmevalc.html
Stan
"If it's not for the boys, it's for the birds!"
Nooby
Pine Head
Pine Head
Posts: 99
Joined: Wed Aug 31, 2005 8:55 am
Location: Southwest, MI

Re: Help the New Guy - Part I

Post by Nooby »

I've looked at the PPN and Sterns, and I dont think it gives us quite the race day atmosphere we are seeking.

Since we are coming from double elimination system, it seems like the competitive nature of chartless elimination keeps that "tradition" while providing more racing for all the scouts.

I do like the fact that the schedule could be prepared in advance - thus letting each of the scouts know exactly when their car will race - something that cannot be done with elimination methods.
"Nooby"
Nooby
Pine Head
Pine Head
Posts: 99
Joined: Wed Aug 31, 2005 8:55 am
Location: Southwest, MI

Re: Help the New Guy - Part I

Post by Nooby »

Here is my idea for selecting Den champions using multiple elimination.

First, each den races 2 elimation rounds which are recorded.

The top racers then compete to determine first and second within the den. These scores would not be recorded for the purpose of the pack race. (Don't ask me how to do this, I'm still in the concept stage).

After all the Den races, the cars are raced 2 or 3 more rounds (incorporating the first two "den only" scores) to determine the cars with 0, 1, 2, 3, and maybe 4 losses.

The finals would be run as normal.

While there may be some logistical hurdles, but I think that this could give the early inter den competition along with the race of champions.

Let me know what you think.
"Nooby"
User avatar
Stan Pope
Pine Head Legend
Pine Head Legend
Posts: 6856
Joined: Sat Jul 05, 2003 7:01 pm
Location: Morton, Illinois
Contact:

Re: Help the New Guy - Part I

Post by Stan Pope »

Nooby wrote:I've looked at the PPN and Sterns, and I dont think it gives us quite the race day atmosphere we are seeking.

Since we are coming from double elimination system, it seems like the competitive nature of chartless elimination keeps that "tradition" while providing more racing for all the scouts.

I do like the fact that the schedule could be prepared in advance - thus letting each of the scouts know exactly when their car will race - something that cannot be done with elimination methods.
PPN is valuable when groups are small (e.g. den size) OR when lane differences are great enough to be the deciding factor in many match-ups.

PPN charts can guarantee each racer at least 2 runs in each lane against varied competition. (There are a small number of combinations of lane and car counts for which only one round charts are presently identified.)

The nochart multiple elim format does not work well for a small number of racers (around den size), since each score group very rapidly reduces to a single member. Heats matching racers from adjacent score groups messes up the accuracy of the method.

Elimination methods also have accuracy problems when significant lane differences are present. PPN, on the other hand, tends to push final scores closer together when there are significant lane differences.

Some folks expand Double Elim by making each "DE heat" a best 2 out of 3, running in alternating lanes. This gets more runs, but does very little to improve accuracy, the effects of lane differences, or the number of different opponents that each car faces. A slower car is still out after racing 2 other cars.

"Atmosphere of competition" depends a lot on how well the participants understand what is going on. Can they "measure their progress" during the competition. Does elimination come as a surprise? Racing by elimination, by points or by time is less important in atmosphere than participation (Does the Scout race his own car or watch an adult / track official race the car for him?) and information about results and progress (Do they know how each heat finished, and can they estimate overall standings?)

One "idea" that I've read from time to time is to keep the overall standings (running Stearns charts) under wraps until the racing is done, and then announce the winners. I don't understand how this maintains the excitement level during the racing. I prefer that each participant have a good idea of where he stands during the competition.
Stan
"If it's not for the boys, it's for the birds!"
Post Reply