| 1 |
@hokomoko:
If
it's
so
biased,
then
your
algorithm/system
shouldn't
have
a
problem
with
predicting
the
winner
better
than
the
current
one.
|
1 |
@hokomoko
:
If
it's
so
biased,
then
your
algorithm/system
shouldn't
have
a
problem
with
predicting
the
winner
better
than
the
current
one.
|
| 2 |
\n
|
2 |
\n
|
| 3 |
Given those games, you're also supposed to "start over" with your own elo system. It might be a bit off at the start, but the battles were predicted by the current system while player skill was also varying, so that's not an excuse.
|
3 |
Given those games, you're also supposed to "start over" with your own elo system. It might be a bit off at the start, but the battles were predicted by the current system while player skill was also varying, so that's not an excuse.
|
| 4 |
\n
|
4 |
\n
|
| 5 |
[quote]So we could assign the guys linear skill values:
|
5 |
[quote]So we could assign the guys linear skill values:
|
| 6 |
1380 -> 1
|
6 |
1380 -> 1
|
| 7 |
1500 -> 2
|
7 |
1500 -> 2
|
| 8 |
1620 -> 4
|
8 |
1620 -> 4
|
| 9 |
\n
|
9 |
\n
|
| 10 |
In a 2v2 where the teams are
|
10 |
In a 2v2 where the teams are
|
| 11 |
1500, 1500 VS 1380, 1620
|
11 |
1500, 1500 VS 1380, 1620
|
| 12 |
\n
|
12 |
\n
|
| 13 |
the linear Elo average is equal but since Elo is exponential we should use the linearized value instead which gives us unequal 2+2 vs 1+4.[/quote]
|
13 |
the linear Elo average is equal but since Elo is exponential we should use the linearized value instead which gives us unequal 2+2 vs 1+4.[/quote]
|
| 14 |
I don't think this makes sense. You are essentially adding fractions by adding the nominators and denominators separately. It may make sense here on some level, but transforming the elo value into another measure yields no reason for why team elo should be linear in [i]that[/i] measure instead of the other one.
|
14 |
I don't think this makes sense. You are essentially adding fractions by adding the nominators and denominators separately. It may make sense here on some level, but transforming the elo value into another measure yields no reason for why team elo should be linear in [i]that[/i] measure instead of the other one.
|
math bork