CrossFit Discussion Board  

Go Back   CrossFit Discussion Board > CrossFit Forum > Competitions
CrossFit Home Forum Site Rules CrossFit FAQ Search Today's Posts Mark Forums Read

Competitions Competitions, contests & challenges

Reply
 
Thread Tools
Old 07-20-2010, 10:10 PM   #91
Justin McCallon
Member Justin McCallon is offline
 
Profile:
Join Date: Oct 2007
Location: Atlanta  GA
Posts: 551
Re: 2010 games scoring - impossible to catch up?

Quote:
Originally Posted by Joe Mercurio View Post
This just creates a new problem. Imagine you have 1 competitor who scored 12th, 12th, 12th, and 12th in 4 events before a cut and you had another competitor who scored 1st, 1st, 1st, and 45th (last) with both having a total of 48 points. Then cut the field in half and reassign points. Now the 2nd guy scores 1st, 1st, 1st, and 23rd for a total of 26 points and leaps ahead of the 1st competitor.
In this case, the guy that consistently got 12th probably was beat by a few of the people cut each time, too.

But yeah, you've highlighted one of the problems with rank-scoring. It's a really non-ideal scoring system and this whole argument wouldn't be an issue if you scored based on proportions and then used standard deviation.
http://www.board.crossfit.com/showthread.php?t=58723
__________________
My Workout Log
  Reply With Quote
Old 07-20-2010, 10:15 PM   #92
Justin McCallon
Member Justin McCallon is offline
 
Profile:
Join Date: Oct 2007
Location: Atlanta  GA
Posts: 551
Re: 2010 games scoring - impossible to catch up?

Quote:
Originally Posted by Josh Crawford View Post
Scale would be a more accurate term. Scale the points as follows. 45 competitors 45/45 = 1 point per place, 24 competitors 45/24 = 1.875 points per place, 16 competitors 45/16 = 2.1825 points per place. Each event gives you the same potential for gaining or losing overall placement without throwing away the previous performance like some of the scoring methods effectively do. That is how I rescored them with my original spread sheet.
I think this still has problems (maybe I haven't thought it through enough?). I really don't think there's a perfect answer for fixing rank-scoring with cuts. Your idea is maybe an improvement from doing nothing, but I think the other solution is better.

This year's Games make a pretty good example. For the most part, the top 24 people were beating the remaining 21 people pretty consistently, before the cut. So, let's say all of the top 24 were in the top 24 each event before the cut. 45/45 = 1 point per place. So they were only getting 1 point. Then, there's the cut. So at that point, 45/24 = 1.875 points per place. So, now, the later events are effectively more important than the early events. Imo your system is not ideal unless the people cut were scoring in the top 24 just as often as the people not cut. And that's necessarily not going to be the case.
__________________
My Workout Log
  Reply With Quote
Old 07-20-2010, 10:32 PM   #93
Joe Mercurio
Member Joe Mercurio is offline
 
Profile:
Join Date: Feb 2009
Location: Oklahoma City  OK
Posts: 81
Re: 2010 games scoring - impossible to catch up?

Scoring by standard deviation isn't the answer either. Take a look at the 1RM push jerk. One guy got a zero creating a huge outlier that made everyone else's margins smaller than an event with a more compact distribution. Do you plan on multiplying each event with a modifier to account for the different shapes of data?

Not to mention that many competitors failed to finish the events in the time allotted. How can you calculate standard deviations when half the competitors have a score in seconds and the other half are scored by reps?
  Reply With Quote
Old 07-20-2010, 10:34 PM   #94
Katherine Derbyshire
Member Katherine Derbyshire is offline
 
Profile:
Join Date: Jan 2009
Location: Seattle  WA
Posts: 7,596
Re: 2010 games scoring - impossible to catch up?

Quote:
Originally Posted by Justin McCallon View Post
Umm... not sure if you read the discussion. He was saying that all the pre-cut WODs were better measures of fitness than the post-cut WODs. My point was that the RHSPU workout was almost identical to the muscle-up workout....
I read it, but not closely. My point was that muscle-ups are more of a core CF task, so suitable for a "screen" to weed people out.

Katherine
  Reply With Quote
Old 07-21-2010, 04:48 AM   #95
Andy Gann
Member Andy Gann is offline
 
Andy Gann's Avatar
 
Profile:
Join Date: Apr 2008
Location: Kodak  TN
Posts: 1,998
Re: 2010 games scoring - impossible to catch up?

Quote:
Originally Posted by Joe Mercurio View Post
This just creates a new problem. Imagine you have 1 competitor who scored 12th, 12th, 12th, and 12th in 4 events before a cut and you had another competitor who scored 1st, 1st, 1st, and 45th (last) with both having a total of 48 points. Then cut the field in half and reassign points. Now the 2nd guy scores 1st, 1st, 1st, and 23rd for a total of 26 points and leaps ahead of the 1st competitor.
You don't just cut their points in half. You reassign their position based on the remaining competitors after the cut.
__________________
The greatest of journeys began with a single step. LOG
  Reply With Quote
Old 07-21-2010, 05:14 AM   #96
Ewen Roth
Member Ewen Roth is offline
 
Profile:
Join Date: Mar 2007
Location: Geneva  Switzerland
Posts: 205
Re: 2010 games scoring - impossible to catch up?

Quote:
Originally Posted by Joe Mercurio View Post
Scoring by standard deviation isn't the answer either. Take a look at the 1RM push jerk. One guy got a zero creating a huge outlier that made everyone else's margins smaller than an event with a more compact distribution. Do you plan on multiplying each event with a modifier to account for the different shapes of data?

Not to mention that many competitors failed to finish the events in the time allotted. How can you calculate standard deviations when half the competitors have a score in seconds and the other half are scored by reps?
It seems to me that straight scoring by percentage of the top performer solves most problems. No need for standard deviations.

- This works for times, reps, and weights, and even DNF's (Starting from the winning performance, you can calculate the percentage score of a theoretical individual that finishes the Rxd rounds at the cut-off time, and work from there; someone who fails all their lifts in a max weight-type event gets a big fat 0.)

- Size of distribution doesn't matter, because if everyone is performing near the level of the event winner, then that's how they should be scored, IMO. If scores are all over the place, big deal. That's the makeup of the competitor pool for that event.

- Having cuts or not at various stages isn't an issue anymore. If you destroy the last WOD and someone else does poorly, the scores will reflect that, whether there are 5 competitors remaining or 100.

- Ties would be virtually impossible (not that it's been a problem up to now).

- I think this would give a better overall view of the competitors' fitness levels compared to each other (subject to how well you think the events measure overall fitness), and therefore would be a better way to crown the "fittest person".
  Reply With Quote
Old 07-21-2010, 06:48 AM   #97
Aushion Chatman
Affiliate Aushion Chatman is offline
 
Aushion Chatman's Avatar
 
Profile:
Join Date: Dec 2004
Location: San Diego  CA
Posts: 3,342
Re: 2010 games scoring - impossible to catch up?

Quote:
Originally Posted by Thomas Green View Post
she aint "small" lol
Not small...TINY!
__________________
My Youtube Channel
CFSD Yelp
  Reply With Quote
Old 07-21-2010, 10:35 AM   #98
Joshua Refenes
Member Joshua Refenes is offline
 
Joshua Refenes's Avatar
 
Profile:
Join Date: May 2009
Location: Mount Prospect  IL
Posts: 173
Re: 2010 games scoring - impossible to catch up?

Quote:
Originally Posted by Ewen Roth View Post
It seems to me that straight scoring by percentage of the top performer solves most problems. No need for standard deviations.

- This works for times, reps, and weights, and even DNF's (Starting from the winning performance, you can calculate the percentage score of a theoretical individual that finishes the Rxd rounds at the cut-off time, and work from there; someone who fails all their lifts in a max weight-type event gets a big fat 0.)

- Size of distribution doesn't matter, because if everyone is performing near the level of the event winner, then that's how they should be scored, IMO. If scores are all over the place, big deal. That's the makeup of the competitor pool for that event.

- Having cuts or not at various stages isn't an issue anymore. If you destroy the last WOD and someone else does poorly, the scores will reflect that, whether there are 5 competitors remaining or 100.

- Ties would be virtually impossible (not that it's been a problem up to now).

- I think this would give a better overall view of the competitors' fitness levels compared to each other (subject to how well you think the events measure overall fitness), and therefore would be a better way to crown the "fittest person".
I agree that this is a good way to score events, and you wouldn't have to take into consideration the amount of competitors remaining.

The only problem I potentially see happening is that a "specialist" wins a workout by such a huge margin, the lead becomes insurmountable for the rest of the competition. This situation is really only a problem for Regional and Sectional competitions where there's only 3-5 events. You can mitigate this possibility by having 1) a large number of events, and 2) testing either extremes of the endurance/strength spectrum.
__________________
29|6'4"|230 lbs|Started CF 5/18/09
CFT:975
  Reply With Quote
Old 07-21-2010, 11:27 AM   #99
Justin McCallon
Member Justin McCallon is offline
 
Profile:
Join Date: Oct 2007
Location: Atlanta  GA
Posts: 551
Re: 2010 games scoring - impossible to catch up?

Quote:
Originally Posted by Joe Mercurio View Post
Scoring by standard deviation isn't the answer either. Take a look at the 1RM push jerk. One guy got a zero creating a huge outlier that made everyone else's margins smaller than an event with a more compact distribution. Do you plan on multiplying each event with a modifier to account for the different shapes of data?

Not to mention that many competitors failed to finish the events in the time allotted. How can you calculate standard deviations when half the competitors have a score in seconds and the other half are scored by reps?
Good point about someone getting "0" messing up the scoring. The solution is to take out the bottom 20% when calculating standard deviation.

And you times convert into scores easily. Just add x seconds per rep short (they had already done this in the previous scoring before assigning ranks).
__________________
My Workout Log
  Reply With Quote
Old 07-21-2010, 11:33 AM   #100
Justin McCallon
Member Justin McCallon is offline
 
Profile:
Join Date: Oct 2007
Location: Atlanta  GA
Posts: 551
Re: 2010 games scoring - impossible to catch up?

Quote:
Originally Posted by Ewen Roth View Post
It seems to me that straight scoring by percentage of the top performer solves most problems. No need for standard deviations.

- This works for times, reps, and weights, and even DNF's (Starting from the winning performance, you can calculate the percentage score of a theoretical individual that finishes the Rxd rounds at the cut-off time, and work from there; someone who fails all their lifts in a max weight-type event gets a big fat 0.)

- Size of distribution doesn't matter, because if everyone is performing near the level of the event winner, then that's how they should be scored, IMO. If scores are all over the place, big deal. That's the makeup of the competitor pool for that event.

- Having cuts or not at various stages isn't an issue anymore. If you destroy the last WOD and someone else does poorly, the scores will reflect that, whether there are 5 competitors remaining or 100.

- Ties would be virtually impossible (not that it's been a problem up to now).

- I think this would give a better overall view of the competitors' fitness levels compared to each other (subject to how well you think the events measure overall fitness), and therefore would be a better way to crown the "fittest person".
They addressed the issue with percentage scoring in their scoring post. They were right.

Here's the issue. Take two events: Running 5k and Rowing 5k.
At the elite level, the top 50 5k runners are going to be spread out by maybe 1 minute or so (worst to best) with probably a normal distribution among the top 50. So, let's say #1 is 13:00 and #50 is 14:00. When calculating the last place person's score, 13/14 = .928.

Then, compare a 5k row at the elite level. The times are much closer together. The top place might be 15:30, but the #50 time is probably around 15:50. So, 15.5/15.83 = .979

.979 vs .928. That's why you need to use standard deviation. In the CrossFit world the difference is probably even more drastic. I would guess that the top rower in most affiliates is around 18:00. Probably the 10th best rower is around 19:30. For runners, I bet the top runner is around 18:30, and the 10th best is closer to 22:30. 4 minutes vs 1:30. And other tasks are even more severe. 10th best "Fran" time is probably half as fast as #1 "Fran" times in most gyms (4:30 vs 2:15 sounds reasonable).
__________________
My Workout Log
  Reply With Quote
Reply


Thread Tools

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

vB code is On
Smilies are On
[IMG] code is Off
HTML code is Off
Forum Jump

Similar Threads
Thread Thread Starter Forum Replies Last Post
Scoring in CF Games Comps Justin McCallon Community 8 07-18-2010 07:02 AM
2010 CF Games Volunteers Britt Dowling Competitions 0 06-28-2010 06:33 AM
2010 Crossfit Games Scoring Rolf Whitney Competitions 2 05-05-2010 01:19 PM
Games Scoring Tom Seryak Competitions 112 02-01-2010 12:27 PM


All times are GMT -7. The time now is 01:46 AM.


Powered by vBulletin® Version 3.6.8
Copyright ©2000 - 2019, Jelsoft Enterprises Ltd.
CrossFit is a registered trademark of CrossFit Inc.