All-Star Let's Discuss Worlds Scoring

[King]

Here is IOC 5 Day 2 with the US teams removed (because the US teams were already ordered). If completely random we would not expect a bias based on time. Instead for international teams you will see a 15-20 point advantage to competing later.

 

[BlueCat]

This could be done with Day 1 at NCA as well, which is theoretically random order.

 

[King]

This could be done with Day 1 at NCA as well, which is theoretically random order.

Oh please share. Where is that data?

 

[King]

This could be done with Day 1 at NCA as well, which is theoretically random order.

However NCA does try to limit division size to combat this (which is honestly a good solution).

 

[Emily]

However NCA does try to limit division size to combat this (which is honestly a good solution).

And they have a scoring rubric.

But with those two differences to worlds it would be really interesting to see if NCA actually does avoid judge bias.

 

[King]

Google doc someone?!?!


The Fierce Board App! || iPhone || Android || Upgrade Your Account!

 

[Peter El]

Umm... there is so much money in big data and statistical analysis right now. I don't know how long this bubble will last, but I would guess at least for another 30-40 years. Were just at the beginning of using all this stuff. Nate Silver is one of my heroes.

As a mathematician, I thoroughly agree with you. Thank you for doing this analysis. If I wasn't so busy with my day job I'd probably play along. I may run some scripts on the big worlds spreadsheet in the future, and will post them if/when I do. Quite willing to do things more than just linear regressions as well. I'll keep an eye on this thread.

 

[Ashley]

Google doc someone?!?!


The Fierce Board App! || iPhone || Android || Upgrade Your Account!

2013 NCA & NDA All-Star National Championship

Have fun.

 

[King]

As a mathematician, I thoroughly agree with you. Thank you for doing this analysis. If I wasn't so busy with my day job I'd probably play along. I may run some scripts on the big worlds spreadsheet in the future, and will post them if/when I do. Quite willing to do things more than just linear regressions as well. I'll keep an eye on this thread.

I'd love to measure deviation from the trend line for each team. And then look at teams that are over a certain percentage of deviation and see if the stand out as true spikes. Are the lows not as lows if they are later?


The Fierce Board App! || iPhone || Android || Upgrade Your Account!

 

[SarahS]

Umm... there is so much money in big data and statistical analysis right now. I don't know how long this bubble will last, but I would guess at least for another 30-40 years. Were just at the beginning of using all this stuff. Nate Silver is one of my heroes.

I'm taking a break from doing my real-world regression analysis of aircraft values and fuel costs in order to check out FB... and I'm confronted with more regression analysis. Love it!

 

[BlueCat]

Big NCA Spreadsheet. (not up-to-date)

NCA 2012

NCA 2013

All is very raw and incomplete, which absolutely makes me cringe to even think about. I still have the interest, just not the time to keep it up-to-date. Please, feel free to rearrange, normalize, etc. . . . Just don't don't delete data in the process. I'm sure that someone resourceful could fill in many of the information gaps as well.

Also, some of it is a little heavy on the Cheer Athletic-y side, but hey, it was my spreadsheet. I must be getting more mellow in my old age, because I would have kept all of this top-secret back in the day. (My preciousssss)

 

[King]

Big NCA Spreadsheet. (not up-to-date)

NCA 2012

NCA 2013

All is very raw and incomplete, which absolutely makes me cringe to even think about. I still have the interest, just not the time to keep it up-to-date. Please, feel free to rearrange, normalize, etc. . . . Just don't don't delete data in the process. I'm sure that someone resourceful could fill in many of the information gaps as well.

Also, some of it is a little heavy on the Cheer Athletic-y side, but hey, it was my spreadsheet. I must be getting more mellow in my old age, because I would have kept all of this top-secret back in the day. (My preciousssss)

Holy cow NCA scoring looks fantastic from a statistical point of view.

NCA Small Senior Day 1. No advantage.

Day 2 after the teams are organized from day 1 we see a solid straight climb (exactly like we would expect).

 

[King]

Now there is much more bias in Small Senior Coed. And there were also a lot more teams. The more teams, the more bias.

Day 1

Day 2

 

[King]

BTW all I am doing is plotting the scores on a line graph over time, then seeing if there is a trend. Some of the terms we are throwing around make it sound more difficult than it is. With this type of data its less about the math or technique, but knowing what and how to analyze.

 

[BlueCat]

This is the kind of information that EPs are afraid would be available to coaches if they release more data. We may find that there are systemic issues with their judging processes. I would argue that finding that out is better than not knowing it, but hiding it takes less effort on their part.

I would love to go all "Moneyball" on the scoring systems, but the EPs just don't release that much useful information. (NCA/Varsity is coming around, to their credit.)