Total Posts:5|Showing Posts:1-5
Jump to topic:

99'er's win ratio

Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
6/2/2011 12:44:55 PM
Posted: 5 years ago
Something that I find rather interesting is that while doing my statisitcal tracking of the 99'ers (yes, the authoritarian nut job is watching you), is that the win % has remained almost complete flat.

Over the two months, it is at the same point it started at 77.86% (with a maximum of 77.90% and a minimum of 77.70%). This includes changes of Hello Orange leaving, then coming back (about 3 weeks later), which bumped Brittwaller up into the 99'ers (then back out), as well as Social Pinko's closed account, and now Darth Grevious becoming a 99'er, and the completion of about 200 debates by 99'ers.

After all of this, this win ratio remains almost insanely flat, compared to the win ratio of various political ideaologies.

The ALC started at 80.38% and is now at 80.15% (max - 80.38%, min - 79.67%).
The CSL started at 64.56% and is now at 65.76% (max - 65.76%, min - 63.54%).
"Wanting Red Rhino Pill to have gender"
RoyLatham
Posts: 4,488
Add as Friend
Challenge to a Debate
Send a Message
6/18/2011 5:42:38 PM
Posted: 5 years ago
Suppose the 99s have collectively completed 3000 debates. I don't know the real number. Then suppose that the have won 2310 of them, so the probability of winning is 0.77. To move the percentage from 0.77 to 0.78 in one month, there would have to be 30 more wins that loses among the 99s in that month. To have a total of 3000 debates, there are probably about 100 debates per month. So instead of 77 wins, there would have to be 107, but that's impossible since there are only a hundred debates. To move down by 0.01, there would have to be only 47 wins rather than 77.

The variance of the binomial distribution is np(1-p) = 100 x 0.77 x 0.23 = 17.71. and so the standard deviation is 4.2. Getting 30 more losses than wins is a seven sigma event. That's way less than one chance in a million.

The reason the 99s win percentage is so stable is that their win percentage is so high and the relatively large umber of monthly debates makes large departures unlikely. The 99s should be roughly twice as stable as other groups.
askbob
Posts: 7,254
Add as Friend
Challenge to a Debate
Send a Message
6/18/2011 5:54:01 PM
Posted: 5 years ago
At 6/18/2011 5:42:38 PM, RoyLatham wrote:
Suppose the 99s have collectively completed 3000 debates. I don't know the real number. Then suppose that the have won 2310 of them, so the probability of winning is 0.77. To move the percentage from 0.77 to 0.78 in one month, there would have to be 30 more wins that loses among the 99s in that month. To have a total of 3000 debates, there are probably about 100 debates per month. So instead of 77 wins, there would have to be 107, but that's impossible since there are only a hundred debates. To move down by 0.01, there would have to be only 47 wins rather than 77.

The variance of the binomial distribution is np(1-p) = 100 x 0.77 x 0.23 = 17.71. and so the standard deviation is 4.2. Getting 30 more losses than wins is a seven sigma event. That's way less than one chance in a million.

The reason the 99s win percentage is so stable is that their win percentage is so high and the relatively large umber of monthly debates makes large departures unlikely. The 99s should be roughly twice as stable as other groups.

This is why we love roy.
Me -Phil left the site in my charge. I have a recorded phone conversation to prove it.
kohai -If you're the owner, then do something useful like ip block him and get us away from juggle and on a dofferent host!
Me -haha you apparently don't know my history
Kohai - Maybe not, but that doesn't matter! You shoukd still listen to your community and quit being a tyrrant!
Me - i was being completely sarcastic
Kohai - then u misrepresented yourself by impersonating the owner—a violation of the tos