2-day points event with different courses

Keith Quistorff · **Posted:** Thu Jul 05, 2012 4:01 pm

Following up on James' request in the other (now locked) thread for feedback on best way to handle calculating ones' best run in a 2-day autocross with different courses -

First, thanks, James, for caring enough to propose a solution to the points issue. Also, the efforts of everyone involved in getting the new site(s) and making adjustments needed to keep the Jimmy V event weekend on track are significant, appreciated, and should not go unrecognized.

As for feedback, I cannot think of a better or simpler way to do this (points for 2-day event with different courses) than what James has already proposed in the other thread, assuming his proposed method for the calculation & final results isn't too difficult for someone to handle, as it sounds above my head. I agree that the fairest method for calculating the appropriate Saturday/Sunday time offset would be to utilize only each days' fastest times for only the drivers that attend both days, and I would suggest, if not too difficult, dropping any driver's data from that calculation where that driver clearly has a large discrepancy in relative driving performance between the two days.

JamesShort · **Posted:** Thu Jul 05, 2012 4:55 pm

Thanks Keith. I'll mock up some data and come up with criteria for outliers and post later.

RobLupella · **Posted:** Thu Jul 05, 2012 6:35 pm

I seem to remember us doing that at one of the other events with the format Keith suggested. I think it was one of the Jimmy V events. I'll look back and see if I can find it.

Rob

JamesShort · **Posted:** Thu Jul 05, 2012 6:52 pm

Example:

Only 2 day people are included in the averaging. Since the 79.xxx is more than 2 standard deviations away from the average, it is thrown out of the calculation for the average for that day.

For the 2 people who only came to day 1, they'd *add* 14.1s to their time to get their day 2 time. For the 2 people who only came on day 2, they'd *subtract* 14.1s from their day 2 time to get their day 1 time.

The lowest summed time across both days will be counted for points.

Steven Carter · **Posted:** Thu Jul 05, 2012 7:18 pm

Sounds like how a Tour event is judged....I'm OK with that, FWIW.

Seems like the lowest and highest times should be tossed in order to calculate the "other" time for the 1-day attenders, in a fashion similar to diving, freestyle skiing, gymnastics etc. Why did you opt for just the slowest time?

Edit....I see you went with Std Dev. I am not a stats guy, but if that is what works best, then I will go with it.

JamesShort · **Posted:** Thu Jul 05, 2012 8:22 pm

Steven Carter wrote:

Sounds like how a Tour event is judged....I'm OK with that, FWIW.

Seems like the lowest and highest times should be tossed in order to calculate the "other" time for the 1-day attenders, in a fashion similar to diving, freestyle skiing, gymnastics etc. Why did you opt for just the slowest time?

Edit....I see you went with Std Dev. I am not a stats guy, but if that is what works best, then I will go with it.

That's a reasonable request (remove highest and lowest). The 79 was removed because I forced it to be 2+ stddevs away, but if it was a time that was 2 stddevs faster than avg, I'd throw it out too.

So options now:

-Throw out any score that is higher or lower than 2 standard deviations from the average
or
-Throw out only the highest and lowest times and then average.

Steven Carter · **Posted:** Thu Jul 05, 2012 8:41 pm

I am not trying to split hairs, and certainly not intending to make things difficult...it was an obsevration more out of curiosity than anything else. Go with what is most efficient to create in your spreadsheet as it could be a lot of work.

JamesShort · **Posted:** Thu Jul 05, 2012 8:45 pm

Steven Carter wrote:

I am not trying to split hairs, and certainly not intending to make things difficult...it was an obsevration more out of curiosity than anything else. Go with what is most efficient to create in your spreadsheet as it could be a lot of work.

No problem, I'm glad you are interested and offering feedback. I'm thinking maybe a combination of drop highest lowest and then dropping large differential outliers too.

BriceJohnson · **Joined:** Thu Apr 13, 2006 7:33 am **Posts:** 2230

The problem with this approach is that it is impossible to know where you stand in real time, which is sort of a bummer. This calculation is even more convoluted than pax. I would also like to drive a different car for the fun event, but that is just me being selfish.

Rain, particularly quickly passing rain, can totally screw this algorithm up too. I would prefer to keep it simple, even if it bites a couple people who can't make it both days, so that the competition is still transparent.

JamesShort · **Posted:** Thu Jul 05, 2012 9:36 pm

Brice, I know it's not scientific in the least, just a proposal.

JohnByers · **Posted:** Thu Jul 05, 2012 11:08 pm

IMHO this is making something more difficult and confusing than needs be. Use the past as a guideline for points and let the rest be gravy. I know I enjoyed driving my RX8 on Saturday and Frank on Sunday last year. Smarter not harder, repeat over and over again!

And yes, thanks to those who do the grunt work :thumbsup:

Steven Carter · **Posted:** Fri Jul 06, 2012 9:04 am

BriceJohnson wrote:

The problem with this approach is that it is impossible to know where you stand in real time, which is sort of a bummer. This calculation is even more convoluted than pax. I would also like to drive a different car for the fun event, but that is just me being selfish.

Rain, particularly quickly passing rain, can totally screw this algorithm up too. I would prefer to keep it simple, even if it bites a couple people who can't make it both days, so that the competition is still transparent.

I don't understand, Brice....this is a calculation for non-attenders. Those attending will see their times posted via Carlton's SuperAwesomeOneofaKindInstaResultApp just like always. Those who don't attend a day will get the "average" time for the non-attended day. Granted that is a potential big penalty for fast drivers in a class with a large range (TIR, as an example) of times and would strongly favor the 2-day attenders....it also penalizes consistent strong drivers in a large class with a close range of times (eg, Zach is always at the top in STR. If he doesn't attend one of the days, is it realistic to saddle him with an average STR time? I lean towards "no.")

Have we considered this scenario:
---each day is considered a separate points event (5A and 5B)
---for season points standings each driver must drop one of the two results.
This has two effects: First, it keeps 1-day and 2-day attenders in competition based on actual driven times, not manufactured averages. Second, if you score well on Day1, Day2 can be used for fun, testing or another shot at improving your standing depending on personal approach.

JamesShort · **Posted:** Fri Jul 06, 2012 9:07 am

Steve, I mentioned making it 2 independent points events with the caveat that you'd have to drop 3 events lowest instead of 2 and Les seemed to think I was nuts, so I'm not sure. Anyone else like that idea?

Steven Carter · **Posted:** Fri Jul 06, 2012 9:11 am

James, I see what you are getting at---my proposal is narrower in that you MUST drop one of the two results from points #5, not 3 drops across the entire season of events. That, I think, is a distinction that differentiates what you said earlier (although I may have misread what you stated.)

JamesShort · **Posted:** Fri Jul 06, 2012 9:20 am

Steven Carter wrote:

James, I see what you are getting at---my proposal is narrower in that you MUST drop one of the two results from points #5, not 3 drops across the entire season of events. That, I think, is a distinction that differentiates what you said earlier (although I may have misread what you stated.)

Ahhh, I see. Sounds reasonable.

Others?

THSCC Forums

2-day points event with different courses

Who is online