Register | Login
Attackpoint - performance and training tools for orienteering athletes

Discussion: Results for All Three Courses

in: 43rd West Point National Ranking Event (Apr 15–16, 2023 - West Point, NY, US)

Apr 17, 2023 7:01 PM # 
sshah:
Here is a link to the results from all three courses: https://sites.google.com/view/west-point-orienteer...

Due to the nature of the Long with Blue and Blue-TT, we do not have combined results for all three courses available to view.
Advertisement  
Apr 17, 2023 7:02 PM # 
sshah:
Because we had technical issues with the Sprint, there are no official split times available.
Apr 17, 2023 8:33 PM # 
nancy:
Are we going to see the total points used to calculate the awards? The formula is a bit awkward for us to figure out how things turned out.
Apr 20, 2023 12:57 PM # 
engineerexplore:
Seconding the request for overall results! It'd be great to see how it shook out across such different distances.
Apr 20, 2023 8:00 PM # 
cmorse:
Complete results, routegadget, photos and more linked from OUSA Event Recap.

https://orienteeringusa.org/2023/04/43rd-annual-we...
Apr 20, 2023 10:02 PM # 
JanetT:
Still waiting for the scoring methodology and scores. Maybe Jordan L can answer?
Apr 20, 2023 10:13 PM # 
Matt-S:
The method/formula was posted on the website (pretty sure it was on the prior Results page, but the page has been updated and it's not there. Maybe it's somewhere else.)

The formula is fairly standard and was:

1000+200*(Mean - Time)/StDev

Classes are not the right population to run this formula, but if you score the whole course and then total/compare the scores for individuals in a class it works.

Here's a TOTALLY-UNOFFICIAL quick&dirty Google Sheet that uses the WinSplits data and computes scores for each course. I've thrown a few tabs in the front that find who ran a Long-Mid-Sprint in-course/class(ish) and totals their Overall Points.

(The list is dynamic and I'm pretty sure it doesn't sort - or that's a trick I don't know in Google. But the groups that did all 3 are small enough to eyeball compare, and the official results on the WP site has who's in which class.)

I spot checked a couple...but this was jammed together quick enough there may be bugs/errors. Please let me know if you find any.

https://docs.google.com/spreadsheets/d/1yUNN7_ewMh...
Apr 20, 2023 11:37 PM # 
JanetT:
Thank you, Matt! Yes, someone referred to the formula, but I couldn't find it when I looked. Thanks for taking the time.
Apr 20, 2023 11:48 PM # 
jjcote:
There are class mismatches in combining the results on Red and Blue, because TT Men ran a longer Long course than the regular M21s, and the TT Women ran Blue instead of Red for the Long. So two additional Overall tabs are needed.
Apr 20, 2023 11:58 PM # 
Matt-S:
No classes are represented, just overall scores calculated by course. Comparisons within class won't change scoring, just ranking.

(@JJ All I really did was total overall points. The tabs are just course-not-class groupings. The scoring occurs by each course separately. The total just goes by Name.)

E.g. for Overall Blue - the formula looks for names on Long-BluTT, Long-Blu, Mid-Blu, and Sprint-Blu; and for Overall Red - Long-Blu, Long-Red, Mid-Red, and Sprint-Red.

Each person only ran 3 races wherever they were found, so their total is their total. And if someone ran a mismatch they wouldn't really have an overall anyway (this is possibly more permissive than less)

Green looks on both X & Y Longs and Mids and Green Sprint, etc. to find names.
Apr 21, 2023 7:05 AM # 
GuyO:
The formula is fairly standard and was:

1000+200*(Mean - Time)/StDev


It's standard for statistics, assuming a Normal Distribution (bell-curve), but I can't remember such a formula being used previously for multi-O-race scoring.

Mean and StDev change with every finisher, which could explain the delay in finalizing overall results. Determining whether runners still on the course could affect placing would not be a straightforward process.

Organizers for 2024 might want to consider something more simple; eg, scoring based on only the winning time, or top 3 times.

IIRC, OE(M) supports the scoring system in which the fastest time is assigned 1,000 pts, and lower places are scored as 1000*(Fastest Time/Time).
Apr 21, 2023 11:53 AM # 
jjcote:
No classes are represented, just overall scores calculated by course.

I think you're missing what I'm saying. On OVERALL-RED, for example, both Isabel Bryant and Jonathan Campbell appear. But they did not run the same course on Sunday. Izzy ran Blue, while Jon ran Red.

I have seen this formula used before for orienteering, at least once. I'm pretty sure it was at a multi-day in Alberta in 2002, and it had the additional complication that you counted your best 4 of 6 runs. I was surprised to see it used again, because it wasn't that well-received. (The results at that meet were taking a long time to compute, and I remember tracking down a statistics professor to get the formula for standard deviation so that we could figure out whether the F12 we were traveling with would be getting an award, or whether we could hit the road for our long drive to Wyoming.)

I can't find the place where the meet info listed this formula (I know I saw it before), so I can't tell whether they specified if the distribution would be applied to each class or to each course as a whole (as is done in the spreadsheet).
Apr 21, 2023 1:25 PM # 
Matt-S:
I got you, JJ.

I think we may be on the same page, or nearly there. You're right, Izzy and Jon rand different Long courses, so I'll use them as an example:

OVERALL-RED is just a collection of people _it is not a ranking and has nothing do to with classes_...there is no scoring done on that sheet or any way that those people are compared to each other other than they did the same sprint and maybe the same middle so the formula could find their 3 races. Every score for every participant is calculated by the course they ran (on the other tabs.)

Izzy ran Long-Blu (1084.22), Mid-Red (1135.43), and Sprint-Red (1097.52) for a total that is added up and can be found on the OVERALL-RED tab (3317.16) and has absolutely nothing to do with Jon other than his total for his courses (2 the same and 1 different) can also be found on the OVERALL-RED tab.

The OVERALL tabs are just finding runners who did 3 races and totaling their scores from those races. (E.g. GRN and BRN look in both X and Y races)

The reason why this formula is no good calculated by class is that the populations are way to small: It's job is to weight how much does a minute difference on the Sprint compare to a minute difference on the Mid or the Long.

With a really small population (e.g. 2 or 3 runners in a class did all 3 races) they could come in 1 second apart on the Long, 30 seconds apart on the middle and 2 minutes apart on the Sprint and you end up with the exact opposite of what you're using the formula to determine. Using the whole-course population to calculate the scores is the only consistently reliable way to come up with the intended outcomes.

So if you're trying to rank a class - here's what the scores should be. You could create a Temporary Filter (allowed in view-mode) that includes the class-members and sort.

E.g. Jon ran M45+: (On OVERALL-RED tab "Create Temporary Filter View" [the funnel-looking-button] and Select Jon, Magnus, and Justin [the 3 M45+ with times/scores for all 3 races] and you get this:




Name Overall-Score
Jonathan Campbell 3,494.68
Magnus Bjorkman 2,681.84
Justin Brown 2,301.53


Hope it helps. Even though I didn't make a tab for every class (yuk) if the WinSplits by Course data included class-info for each runner (like it does club) I could have pre-made filters, but maybe "next time".
Apr 21, 2023 3:48 PM # 
cmpbllv:
I appreciate all the thought you are putting into this. I suspect there won't be any further results published from West Point - they just needed enough to distribute awards, and as you've all figured out, this is a beast to calculate (and impossible on-site with people wanting awards so they can go home). And the cadets are in academic crunch-time right now, too.

In retrospect, most of the people in each class had a pretty good idea of who had won 2 out of 3 races in their class and should receive the 1st place coin, and who were the most likely runners-up. We should probably have asked classes to self-identify, done a quick results check if there was any question, and then saved the crazy calculations for just the classes with lots of people that had less clear results.

I encourage all clubs to use existing software that automates the results process, or to write your code *in advance* and test it *in advance.* As anyone who's ever hosted Jr. Nats knows, calculating results that could change with each finisher with limited time available and getting it right on the first go is really, really hard.

Thanks to everyone who jumped in to help with the excel formulas to try to automate this on the fly!
Apr 21, 2023 5:53 PM # 
GuyO:
At least with Jr Natls, only finishers that place in the top 3 will affect scores of those who have already finished.

While not as simple as with totalling times, determining when none of the people still on each course could affect the final placing is relatively straightforward.
Apr 21, 2023 7:20 PM # 
Matt-S:
Tori and all the folks trying to crunch the stats real-time deserve their own medals. Big thanks to them; 100%.

Guy, you've got me curious what other models would work well for Overall totaling.
It seems like anything that's linear would not be as effective at the variation between races, but maybe it's close enough.
Apr 21, 2023 7:33 PM # 
GuyO:
@Matt-S: Quick Q...
Who are you?
Apr 21, 2023 7:47 PM # 
Matt-S:
Hi, Guy.

I'm Matt. uh - I'm a Cancer, I like long walks (and Middle runs)... I'm not very funny. My brain gets curious about weird things. I'm a Cybersecurity geek in D.C., Eagle Scout, QOC member & course designer, just added to the NT Selection Committee, was at West Point and had a really fun time both not doing as well as I should have and most of all following the exceptionally competitive TT field and how those results worked out. Idk what else?
Apr 21, 2023 8:05 PM # 
vmeyer:
Here is a PDF of the score table...there is a separate class for each of M/F-21+ and M/F-21+ TT.

There was a little fiddling required to create separate entries for people who ran different classes over the three events, but I think I got it mostly correct. I let OE2010 do the rest of the heavy lifting on the calculations.

This is the scoring for best 2 of 3 races:
vmeyer.net/ostuff/points2.pdf

Was that what the awards were based on?

I encouraged the cadets to set up the event as separate stages, since you can not have multiple classes for an entrant without creating a separate entry for the person. It is easier to administer each event this way, but this caused the issue with not having easily accessible overall results. Something to think about in the future...
Apr 21, 2023 8:25 PM # 
cmpbllv:
Awards: races were scored and then totaled for those who had all 3 races in the same class…with exceptions for the TT group that had an ultra long Sunday (made them their own category).

And yes, that was one of the existing scoring options I was thinking of, Valerie. Thanks for running that!
Apr 21, 2023 10:22 PM # 
EricW:
Guy and others- I am not sure why Matt-S is reluctant to give up his last name, and I am always skeptical of people who don't want to introduce themselves, but I'll vouch for him, since he sounds like the guy I worked with recently on the NT Selection committee, very capable and with good intentions. I believe Peggy D will vouch for Matt-S with even more familiarity.
Apr 21, 2023 10:37 PM # 
vmeyer:
If awards was based on all three results in the same class, then the results would look like this

vmeyer.net/ostuff/points.pdf

The formula used is a common one, and it is available within the scoring software. It is also common to use it for multi day events where people run the same classes throughout the event. Additionally, it is used at large events where each class is essentially a course of its own, so, there isn’t the issue of not enough results within a class.

For the Sprint, Middle, Long scenario, we would be better off with cross country scoring, or something simpler.
Apr 21, 2023 11:14 PM # 
GuyO:
@Matt-S: I thought you might be one of the cadets who helped organize the event.

Eric's endorsement has value, but, based on the info you offered, I suspect you were on a registration list I recently managed... ;-D
Apr 22, 2023 12:37 AM # 
peggyd:
Yes, Matt-S is Matt SMITH. I think he went to more NRE races last year than anyone but Clint M ... plus the Billygoat, plus he volunteered at the World Games. He's one of the people I'd call a student of the sport, learning about everything from mapping and course setting to who the international elite runners are. I was very pleased to have him on the NT Selection Committee.

This discussion thread is closed.