Warrior General Games

Originally posted by NiborRis
But that's the point - elo handles record AND strength of opponent. Based on who you played and the results you got, if you had started on a level playing field you'd be ranked 4th (as of when the OP ran his rankings). And then you'd get matched up with top 10 teams (most likely) and we'd find out for sure if Iowa could play with them. Your performance, compared to the performance of other top teams, would put you in the top 10. Maybe you wouldn't stay there when matched up against other top 10 teams, but at least then we'd know.


Gotcha...yeah...well Im sure I will eventually get the chance if Im worthy...were slowly making our way up there. Guess I was just more concerned with League play than ladder rankings. Ill watch a little closer now...but Im a little flattered

Originally posted by Riddler08
Gotcha...yeah...well Im sure I will eventually get the chance if Im worthy...were slowly making our way up there. Guess I was just more concerned with League play than ladder rankings. Ill watch a little closer now...but Im a little flattered

You're 18 right now. You'll get your chance.

Originally posted by Jampy2.0
not sure what all this means thunderoo,

but can u just tell me why http://glb2.warriorgeneral.com/game/team/74 isn't up there.......

Got you guys at 57 with mid teams at 1200 elo start and 51 with mid teams at low start. Around 1288.95 rating.

Edit: There are probably 15 teams in the top 50 that should be switched to near 1200 rating. Hmm this could be done automatically if they're all in the planet leagues.

I don't know why everyone wants age separate rankings. The problems merging them as teams progress would be a nightmare. Have everyone enter at the same elo (make this entry point the average rookie elo after season 2). And then just keep things static, this will ensure that new rookie teams are playing with teams around their level (ie won't take long to move down) and then no one has to stuff around trying to filter teams in to the ladder rankings as they move in to veteran.

Also just display the elo as well; it provides way more information than a ranking as if you're using an elo system then there are going to be big gaps which won't be obvious. For example right now 1 and 2 are far enough in front that it'd take a few losses for them to move down.

Oh got CSV files of the rankings and also of all of the games up until the 2nd of Feb if anyone wants them (also code to generate them too).

Definitely agree that midseason teams from last year have some issues climbing, takes way longer.

Originally posted by Thunderoo
Got you guys at 57 with mid teams at 1200 elo start and 51 with mid teams at low start. Around 1288.95 rating.

Edit: There are probably 15 teams in the top 50 that should be switched to near 1200 rating. Hmm this could be done automatically if they're all in the planet leagues.

oh ok cool

Made some changes and now have teams in solar system leagues being dropped (need to test to work out correct drop point). All mid season teams also get roughly same elo. Interesting seeing the spread though, gap from 1 to 10 is the same as the gap from 10 to 43 which when you know its elo you expect but I bet it's going to cause a lot of questions in the coming seasons.

Also Jampy you guys move up to 38.

You should also remove the claim that these are the numbers that we use, because they aren't.

Interesting that the ranks generally align with our rank though.

OG's rank over SF is awesome. imo.

Not that surprising, the starting rating for the S1D1 teams is irrelevant for the final rank, so it doesn't matter if he got that different, and k-value differences are going to be a pretty small factor with the small number of games you have. The biggest differences are when he's taking different guesses into the midseason team starting ratings, the S2 team starting ratings, and the rookie S2 team starting ratings. And whether or not he resets the teams that dropped down at the right time and to the right rating.

There's a bunch of other factors not being talked about but it's unlikely they're different - the factor of 400 can be set arbitrarily to whatever you like but it's not that useful to change it. The distribution used is usually logistical, but a normal distribution could be used instead as well (and originally was by Elo). Probably a logistical distribution is more correct here, but I dunno. Feels correct so far though.

Originally posted by Corndog
You should also remove the claim that these are the numbers that we use, because they aren't.

Interesting that the ranks generally align with our rank though.

It says this is an estimate in the OP.

Originally posted by Parab00n
It says this is an estimate in the OP.

I just changed it shh

Originally posted by Homage
Originally posted by bhall43

OG's rank over SF is awesome. imo.



lold

I'm guessing the ranking that's shown on Unaffiliated teams is just the ranking they last had when they were included?
http://glb2.warriorgeneral.com/game/team/27

For sure they aren't in the current rankings (and shouldn't be). Another case where having the actual rating might be interesting (but maybe not, because ratings may not be very comparable over time; ratings inflation and/or deflation is likely to occur in this group).