| 1 |
[quote]Only on a few occasions has anyone suggested that perhaps some matches would be better off with a somewhat less balanced predicted win probability in exchange for an apparently-better distribution of low-skilled players so as to not constantly create unenjoyable games where the high-skill players are all on one side along with all the low-skilled players, effectively playing 1v2 against the medium-skilled players.
|
1 |
[quote]Only on a few occasions has anyone suggested that perhaps some matches would be better off with a somewhat less balanced predicted win probability in exchange for an apparently-better distribution of low-skilled players so as to not constantly create unenjoyable games where the high-skill players are all on one side along with all the low-skilled players, effectively playing 1v2 against the medium-skilled players.
|
| 2 |
\n
|
2 |
\n
|
| 3 |
I think this point deserves a lot more discussion than it's gotten so far.[/quote]
|
3 |
I think this point deserves a lot more discussion than it's gotten so far.[/quote]
|
| 4 |
\n
|
4 |
\n
|
| 5 |
That's actually what I've been talking about in the entire first page of the thread... The need to introduce a tradeoff (= new formula) between elo difference between teams and intra-team-elo-variance.
|
5 |
That's actually what I've been talking about in the entire first page of the thread... The need to introduce a tradeoff (= new formula) between elo difference between teams and intra-team-elo-variance.
|
| 6 |
[i]Then[/i]
we
diverged
on
how
the
formula
of
@Brackman
(
or
any
such
formula,
really)
makes
the
brute
forcing
a
lot
slower,
and
then
we
got
to
common
optimization
techniques
and
@Aquanim's
algorithm.
|
6 |
[i]Then[/i]
we
diverged
on
how
@Brackman's
formula
(
or
any
such
formula,
really)
makes
the
brute
forcing
a
lot
slower,
and
then
we
got
to
common
optimization
techniques
and
@Aquanim's
algorithm.
|
| 7 |
\n
|
7 |
\n
|
| 8 |
About that:
|
8 |
About that:
|
| 9 |
Yes, a decent starting guess and some sort of short hillclimbing (i.e. try swapping random players, accept it if the solution got better, stop if it didn't succeed x times in a row) should work pretty well. But the random swapping might increase elo variance again, potentially to the point where it's back to the old problematic cases.
|
9 |
Yes, a decent starting guess and some sort of short hillclimbing (i.e. try swapping random players, accept it if the solution got better, stop if it didn't succeed x times in a row) should work pretty well. But the random swapping might increase elo variance again, potentially to the point where it's back to the old problematic cases.
|
|
|
10 |
\n
|
|
|
11 |
PS: I decided to stop caring about the apostrophes breaking player links. Blame the one who broke it...
|
Balancer broken?