Lets Beef - Battle Rap Forums - View Single Post - The Greatest Of: Champions (Kiwi's version)

**Kiwi Peewee** · #29 07-09-2019, 11:29 PM

@NOBLE @Phil Banks @Jimmy Hoffa

Regarding adjusting scores based on time period: If it's available, the raw data on this would be so helpful. The problem is idk if it can be found, but if it can, here's what I would do: Take the top 10,000 most highly ranked accounts on the site in text and maybe the top 5,000 in audio. Sort them by years active, so you'd exclude people like ALL-BORO whose runs transcend across multiple eras to make it easier. Maybe just split them into people who were primarily active in 2006-2009, 2010-2015, 2016-2020. Then, you work out how many points under this system everyone in the rankings would have, and use that to format a half-bell-curve (90% or more would have 0 points obviously so it wouldn't be a full bell-curve). You could then work out how the score distributions differed by era, so for instance "the top 0.5% of battlers in the 2000s got >20 points, the top 5% got >5 points, the top 10% got >0 points; whereas from 2010-2015 the top 0.5% got >30 points, the top 5% got >15 points and the top 10% got >3 points" and then use some form of mathematical modelling or something to adjust each battler's scores based on their time period.

However, I think that this kind of system is more like a "pound for pound greatest boxer of all time" discussion. The thing about "pound for pound" type discussions is that its always going to be more effective to make a subjective call based on talent rather than trying to use statistics.

Also, no-ones mentioned it, but I think that whether achievements were in audio or text needs to be taken into account as well, relative to the time period in which they were earned. In 2006, the audio GC I think carried much more weight than the text, because the text side of the site was really underdeveloped. THA DUKE's accomplishment should probably be worth more than SpleNDiD. However, in 2019, I think a text win is probably more noteworthy than an audio win. For instance, in the latest audio GC, there were 15 people signed up, almost half of whom no-showed in the first round or something like that, and it reached the point where I actually signed up for it purely because I felt bad that such a prestigious tournament couldn't find 16 participants. Not to take anything away from Boro's phenomenal achievement, because the quality of the battlers that remained was impressive, but the competition for text titles is currently much more heated than audio imo. Just my thoughts though, I'd be interested to see what people have to say.

**Kiwi Peewee** · 07-09-2019, 11:29 PM

@NOBLE @Phil Banks @Jimmy Hoffa

Regarding adjusting scores based on time period: If it's available, the raw data on this would be so helpful. The problem is idk if it can be found, but if it can, here's what I would do: Take the top 10,000 most highly ranked accounts on the site in text and maybe the top 5,000 in audio. Sort them by years active, so you'd exclude people like ALL-BORO whose runs transcend across multiple eras to make it easier. Maybe just split them into people who were primarily active in 2006-2009, 2010-2015, 2016-2020. Then, you work out how many points under this system everyone in the rankings would have, and use that to format a half-bell-curve (90% or more would have 0 points obviously so it wouldn't be a full bell-curve). You could then work out how the score distributions differed by era, so for instance "the top 0.5% of battlers in the 2000s got >20 points, the top 5% got >5 points, the top 10% got >0 points; whereas from 2010-2015 the top 0.5% got >30 points, the top 5% got >15 points and the top 10% got >3 points" and then use some form of mathematical modelling or something to adjust each battler's scores based on their time period.

However, I think that this kind of system is more like a "pound for pound greatest boxer of all time" discussion. The thing about "pound for pound" type discussions is that its always going to be more effective to make a subjective call based on talent rather than trying to use statistics.

Also, no-ones mentioned it, but I think that whether achievements were in audio or text needs to be taken into account as well, relative to the time period in which they were earned. In 2006, the audio GC I think carried much more weight than the text, because the text side of the site was really underdeveloped. THA DUKE's accomplishment should probably be worth more than SpleNDiD. However, in 2019, I think a text win is probably more noteworthy than an audio win. For instance, in the latest audio GC, there were 15 people signed up, almost half of whom no-showed in the first round or something like that, and it reached the point where I actually signed up for it purely because I felt bad that such a prestigious tournament couldn't find 16 participants. Not to take anything away from Boro's phenomenal achievement, because the quality of the battlers that remained was impressive, but the competition for text titles is currently much more heated than audio imo. Just my thoughts though, I'd be interested to see what people have to say.

07-09-2019, 11:29 PM	#29
Kiwi Peewee Autumn Champion Rookie Champion Ranked Text Record 57 Won / 17 Lost Join Date: Feb 2019 Voted: 49 audio / 843 text Posts: 1,225 Mentioned: 392 Post(s) Tagged: 23 Thread(s)	@NOBLE @Phil Banks @Jimmy Hoffa Regarding adjusting scores based on time period: If it's available, the raw data on this would be so helpful. The problem is idk if it can be found, but if it can, here's what I would do: Take the top 10,000 most highly ranked accounts on the site in text and maybe the top 5,000 in audio. Sort them by years active, so you'd exclude people like ALL-BORO whose runs transcend across multiple eras to make it easier. Maybe just split them into people who were primarily active in 2006-2009, 2010-2015, 2016-2020. Then, you work out how many points under this system everyone in the rankings would have, and use that to format a half-bell-curve (90% or more would have 0 points obviously so it wouldn't be a full bell-curve). You could then work out how the score distributions differed by era, so for instance "the top 0.5% of battlers in the 2000s got >20 points, the top 5% got >5 points, the top 10% got >0 points; whereas from 2010-2015 the top 0.5% got >30 points, the top 5% got >15 points and the top 10% got >3 points" and then use some form of mathematical modelling or something to adjust each battler's scores based on their time period. However, I think that this kind of system is more like a "pound for pound greatest boxer of all time" discussion. The thing about "pound for pound" type discussions is that its always going to be more effective to make a subjective call based on talent rather than trying to use statistics. Also, no-ones mentioned it, but I think that whether achievements were in audio or text needs to be taken into account as well, relative to the time period in which they were earned. In 2006, the audio GC I think carried much more weight than the text, because the text side of the site was really underdeveloped. THA DUKE's accomplishment should probably be worth more than SpleNDiD. However, in 2019, I think a text win is probably more noteworthy than an audio win. For instance, in the latest audio GC, there were 15 people signed up, almost half of whom no-showed in the first round or something like that, and it reached the point where I actually signed up for it purely because I felt bad that such a prestigious tournament couldn't find 16 participants. Not to take anything away from Boro's phenomenal achievement, because the quality of the battlers that remained was impressive, but the competition for text titles is currently much more heated than audio imo. Just my thoughts though, I'd be interested to see what people have to say.
Offline