Total Posts:6|Showing Posts:1-6
Jump to topic:

70th Percentile

TUF
Posts: 21,309
Add as Friend
Challenge to a Debate
Send a Message
12/29/2013 6:21:53 AM
Posted: 2 years ago
Lol wierd. Might wanna ask ore ele about that, he's the stat guys around here ha
"I've got to go and grab a shirt" ~ Airmax1227
Bullish
Posts: 3,527
Add as Friend
Challenge to a Debate
Send a Message
12/29/2013 11:25:18 AM
Posted: 2 years ago
There is also no percentile below 66.04%

http://www.debate.org...

This is because percentiles are calculated by some algorithm with total wins, losses, ties, and debates that are still in voting period, which can cause a lot of ties.
0x5f3759df
Ore_Ele
Posts: 25,980
Add as Friend
Challenge to a Debate
Send a Message
12/29/2013 1:20:47 PM
Posted: 2 years ago
At 12/29/2013 11:25:18 AM, Bullish wrote:
There is also no percentile below 66.04%

http://www.debate.org...

This is because percentiles are calculated by some algorithm with total wins, losses, ties, and debates that are still in voting period, which can cause a lot of ties.

Actually, it is mostly just based on wins. It is still off of our old system. That is why Dairygirl, with 25 wins and 202 loses is still a 98.99% percentile, while The_Chaos_Heart is only 98.96 with 24 wins and only 1 loss.

Basically, yours wins alone indicate what tier you are in, then your losses indicate where within that tier you are. So even if you have 200+ loses, you will be in a higher tier than someone with just one loss less than you (though you will be at the bottom of that higher tier), for the most part (this gets mixed up when you get closer to the top)

The size of the tier is dependent on the number of people in that tier, but the breadth of the tier is an average of all the tiers.

To make this easier, if we have 20 different tiers (with a total membership of 1,000), then the average will be 5% per tier, However if 400 people are in that bottom tier, that tier will start at 40%, but the bottom of the tier will only be down to 35%. Now, when we get to the top, lets say that very top tier has 10 people, in it, That represents only 1% of the population (so the tier right below them will start at 99%), but that top tier will still have the breadth of the 5%, so the top tier will be between 100% and 95%, while the second tier begins at 99% and ends at 94%.

We can see these large gaps when two near tiers both have a significant number of people in them. So lets say that the second to bottom tier has 300 people in it. It will start at 70% and go as low as 65%, while the tier below it starts at 40%, meaning there is no 41% - 64% range. There is also the delay in updating with has an impact.
"Wanting Red Rhino Pill to have gender"