Random team poll

[dweiler]

Fixed Elo
the way it currently it is. Random teams with +15/-15 every game.

Team-based Elo like it currently is
What it would be like when you do !scores right now.

Tweaked team-based Elo: Stretch
This means the Elo distribution for winning and losing will be less harsh than it is now.

Tweaked team-based Elo: Cap
Same Elo distribution as now, but with a maximum (Probably around 23/7 Elo). Matchups cannot fall below this.


I am aware of the weaknesses of the poll and take them into account. Time forced me to make the poll like this

[dweiler]

Just for info, I have not submitted Don_killuminati's suggestion of random captain teams in this poll, because we can implement a !randomchallenge command if people want this. Make a new topic for that.

[Jamo]

What will the "stretch" option look like then?

[Diablo_]

!Scores isn't correct for LIHL games, see https://entgaming.net/forum/viewtopic.p ... 36#p233336

We once already got an update in the Elo calculation formula while the !scores still displays the general, public results which differ from LIHL.
So options 2, 3 and 4 are pretty wrong

If we would disable the fixed Elo now (taken the old formula is still implemented), the Elo gain/loss would differ from !scores and the distribution would be less harsh already, basically being capped at 24/6 with 1000 Elo difference between teams (I doubt the biggest difference would be any bigger than that).

Imo this Elo distribution is a good start and poll should be either 15/15, the Ilocos formula or a formula which is a bit less harsh than Ilocos (I don't like that 50 Elo difference already results in 17/13 and 100 difference in 18/12 for example, these differences can already happen due to the daily fluctuation of everyone's Elo).

[dweiler]

!Scores isn't correct for LIHL games, see https://entgaming.net/forum/viewtopic.p ... 36#p233336

We once already got an update in the Elo calculation formula while the !scores still displays the general, public results which differ from LIHL.
So options 2, 3 and 4 are pretty wrong

If we would disable the fixed Elo now (taken the old formula is still implemented), the Elo gain/loss would differ from !scores and the distribution would be less harsh already, basically being capped at 24/6 with 1000 Elo difference between teams (I doubt the biggest difference would be any bigger than that).

Imo this Elo distribution is a good start and poll should be either 15/15, the Ilocos formula or a formula which is a bit less harsh than Ilocos (I don't like that 50 Elo difference already results in 17/13 and 100 difference in 18/12 for example, these differences can already happen due to the daily fluctuation of everyone's Elo).

Okay if it differs then I will set up a new poll, leaving the technicalities behind. I will archive this and make a new poll.