Page 1 of 1

Team score modifications (Activity factor) and JrTeams

Posted: Sun May 14, 2006 12:45 pm
by [dmp]
Tokimi made interesting read. Not that I agree with it in full deatils, but he got alot of good points. Thanks for sharing tokimi. While reading it I got an idea. So here is my few cents.

Team score modification (Activity factor) (Was discussed on IRC)
Rewarding active teams for activity is to be an issue on alot of peoples mind. I think the activity-factor is an easy way to do this. Every team got an activity factor (1 = all active, 0 = dead team). There ELO score (as we use now) is then multiplied with the activity factor and teams are sorted by the result of the multiplication. The only issue is how this activity factor should be calculated (but it should be pretty easy to come up with some formula). Codewise the idea is pretty straightforward to implement and if it dosnt work as expected (we could have a trail periode to see how it works out) then just remove it again.

Jr. teams
Anohter idea I had while reading, was to formalize the "Jr." team concept as used by "DUB jr". A team is a "Jr" only if all members (excluding the teamleader) are new members (registered within say 6 months). The reason for the teamleader is excuded is to encourage regulars to start up new teams - a mentor like role.

The "Jr" label will be automatically assigned to new teams, and removed when:
1) a too old player (more than 6 months) joins.
2) they reach an aim such as playing 10 matches. If they reach this (and still is a Jr team) they will be rewarded with a system message saying congrats. We could even go further and post it on the news page of the league.

A Jr. label could also be added to players to make it easier to see how is candidates for a Jr team). One could even go abit further and keep track of players you brought up new teams and show it on their players page. So active people are rewarded. (Just see how much people care about teamscore - one could only hope that "personal stats" would get same attention.)

Jr-teams should be listed as a seperate section (like "Inactive teams", "Never played a match") and perhaps without a (visible) score. (Score are calculated, but is first shown publicly once it qualifies as a normal team)

This is basically an attempt to move focus (for new teams) from the el score onto qualifying (by playing) as a fullblown team. Once they upgrade, they play as we do now. Score is the aim for the team.

And it will also reward regular players for leaving the "secure" environment of old timers and starting up something new.

So aims for new player would be:
1) join or create a (Jr-) team
2) fight for getting rid of the "Jr"
3) fight for elo points and fame
4) when getting good, create a new team, find new players and earn "karma" for bringing up a new team. Goto 3

Codewise, will same as before. Its pretty straightforward (its adding a few new values and making a task that enables/disables the Jr-state of a team). And if people dont care about the Jr-teams concept, they can ignore it.

Posted: Sun May 14, 2006 10:53 pm
by Discovery
I agree, and like it!

Posted: Mon May 15, 2006 5:56 am
by SportChick
Interesting ideas, dmp.

I had heard the activity factor one before. It certainly has possibilities. I'd be interested in hearing more what others think about its usefulness.

On the Jr. teams, I was actually thinking of something similar when reading tokimi's post. Your idea is much more thought out, and more structured than mine. I was thinking that it would be nice if some of the larger, more established teams adopted or mentored a new team - either using it as a farm team (to train and recruit new talent) or just to mentor and help the team until that team became more established and capable of standing on its own. I am not sure exactly how Jr. Dub works, but that is always the way I had thought of it.

Posted: Mon May 15, 2006 9:17 am
by CannonBallGuy
This is a little radical, but would help introduce new blood ( :twisted: ) to the new league like meacan's Jr Team idea:
Every established team (one that has been active for 6 months (for example) and/or has a rating of 1200 (for example) or higher) must recruit at least one "Jr. Player" (someone that joined the site within the last 6 months (also for example)) every 2 months (you guessed it... for example) and they must keep this Jr. Player for at least 2 months to give them a fair chance to play some matches and show the team what they are made of.

Posted: Mon May 15, 2006 1:16 pm
by SportChick
CannonBallGuy wrote:This is a little radical, but would help introduce new blood ( :twisted: ) to the new league like meacan's Jr Team idea:
Every established team (one that has been active for 6 months (for example) and/or has a rating of 1200 (for example) or higher) must recruit at least one "Jr. Player" (someone that joined the site within the last 6 months (also for example)) every 2 months (you guessed it... for example) and they must keep this Jr. Player for at least 2 months to give them a fair chance to play some matches and show the team what they are made of.
I thought of that actually. The problem is team size. Particularly if, at some point, smaller teams are decided upon. A few teams are already pushing the limit of team size.

Posted: Mon May 15, 2006 6:04 pm
by CannonBallGuy
Then an official team-size-limit of 10 players (for example, hehe) could be set..?

Posted: Mon May 15, 2006 9:05 pm
by Admirarch
I think forcing people to take in members under any conditions is a mistake, it's far more likely to just end up annoying the existing members than building up the new ones. Teams have a right to only let in players who they believe they will have fun playing with and forcing them to be continually on the lookout for new talent who would fit in the team would be a burden.

By far the most effective way for new and active teams to start up is for experienced players to go out on a limb and commit themselves to teams full of people just starting to play. I've been involved in such teams in both roles and I don't think it would be possible to force anybody to make such a scheme work.

Posted: Mon May 15, 2006 10:36 pm
by [dmp]
What are you commenting Admirarch ? Both suggestions? (CBG and mine?)

Posted: Tue May 16, 2006 9:44 am
by Admirarch
Mainly CBG's, yours I'm still undecided as to whether I like it or not. I'm still convinced the best way to accomplish the goal it sets out to achieve is just for experienced players to start teams but this may well be impractical in which case Junior teams may work in some form.

Posted: Fri May 19, 2006 8:18 am
by coincoin
Hi all,

Even if i m only a gu player, i think this interesting and could be apply to gu.

i agree with the idea to take into account the activity of a team in the score calculation by giving them a bonus. It will encourage to play. But i think this "bonus" should be something like :

activity > 2 matchs a day => *y or +x
1< activity <2 => *y/2 or +x/2
activity < 1 => *0 or +0

this idea of make categories of activity because : some teams have more players than other, so it s easier for them to play. Some player can only play at week end.

hf all,

P.S. : sorry for my bad english, i hope it s understandable

Posted: Sun May 21, 2006 11:16 am
by zongo
Hey all,

Thx [dmp] for your post.
I won't comment on the activity thing but since i am a pioneer in the business of Jr teams lemme say a word here :)

My original idea was to give the opportunity to newbies to play in a farm team, without caring about team score, until the reach a certain level. Then, they could stay in the team, create a new one, or try to "upgrade" to well established teams. The main problem i observed in the league being that newbies hardly have any chance to be part of "good" teams.
On the other hand, as we know, making a new team out of blue is not always easy and takes lots of time. Thats how came the idea of Jr DUB.

Here are some issues I encountered:

- lack of time: taking care of a bunch of new players required more time than i thought. You have to sort players, decide who to accept etc. Organise matches as well can be painful, explaining tactics, tricks and so on.

- level of activity: many new players are active at the very beginning and then lose interest. Some very motivated players are disappointed because they dont see teamies online etc.

- Jrteamleader: the leader of the Jr team should be a very active player, and normally not involved in any other team (like i was). My main mistake was to be both captain of DUB and Jr. DUB. Don't do the same and be totally involved in the Jr team ;)

On the other hand, i saw many players in Jr. DUB (at least 30 players all in all) and i'm happy to see that some of them now are nice players that found a place in good teams (I could name them but can't remember em all ;) )

So I think that [dmp]'s "formalism" for Jr teams is good, it could help a lot to make things clear (at the time Jr DUB was often considered as a DUB2 team...) and hopefully it could help new players to integrate into the league!


Posted: Sun May 21, 2006 8:06 pm
by wizard
What about this:
We will only add the icon for the rank (the stars/stripes) after a team won their first match.

Also: For new teams with uncertain strength it's ok to sort them into the elo system by a modified and faster algorithm:
If you win against a team you get their strength index. If you lose you get the strength index of the loosing team.
So a new team would drop very quickly to the bottom of the league by losing against another newcomer team. While a new top team would rise quickly to their real rank.

The idea is that one of the worst moments for a new team is to see their score constantly dropping in the first few month.

Another modification could be a seperate ladder for teams that didn't win any match yet. Maybe the ladder "did not play any match" could be reused as did not win any match.

I believe this would make new teams value their first win much higher because it's their entry to the real ladder while a loss doesn't really change anything.

A problem is: How should a match affect the score of an established team? Just like a normal game? Not at all? Partly?

Only when a team has at least one loss and one win it will continue with normal elo based rating.

I think this modification can be an addition to a parallel rating which rates the activity. We could use the existion activity index to sort the teams by activity to get an activity related ladder.

And a ladder which lust lists the number of wins in the last three month. It would be easier for a good team to rise on that ladder but every team can compensate with activity.

Posted: Sun May 21, 2006 10:19 pm
by [dmp]
If we need to have the teamscore to change quicker for new teams, why not change the
. It states:
If the K-factor coefficient is set too large, there will be too much sensitivity to winning, losing or drawing, in terms of the large number of points exchanged. Too low a K-value, and the sensitivity will be minimal, and it would be hard to achieve a significant number of points for winning, etc.
We currently use 10 as K-factor. Why not just increase this K-factor for new teams and put it back to 10 when the team is more mature (which could be : played 10 matches or similar)? That would give the desired effect too - and we keep the same formula :)

Posted: Tue May 23, 2006 1:38 pm
by SportChick
Several of us had a discussion yesterday. As Admirarch is preparing for exams, I agreed to post this on his behalf (this is taken from ##essy if you care to read the entire conversation):

Admirarch said: "What I favour:
1) Keep the ELO at all costs
2) Make a prominent 1vs1 leagueish bi-monthly ladder thing
On a side note, I like the idea of having who played in matches recorded whatever happens."

Regarding a player-based ELO system (as proposed by [dmp]):

"How do we handle subs? And people dropping in worse players when the match is in hand in order to increase the swing. Also, teams gaining points from inactive players. I'm trying to work out just how much of a difference it would make and whether it really seems worth it."

[dmp] said, "the big difference is the shift of focus. Now players can fight for ego-reasons. To improve their own rating."

Admir said "It's definitely worth thinking about."

On the idea of an overall system, I commented "I am sort of thinking of a combination of [player ELO] plus mistake's seasonal ladder."

Admir's response: "I think seasonal ladder is pretty much decided upon. This would be a direct replacement for the traditional system. I think closing seasonally is better [than a rolling ladder]. I think a rolling ladder doesn't really make any sense. It doesn't actually represent very much."

Posted: Tue May 23, 2006 2:53 pm
by Chestal
I always wanted to have a way to get a meaningful ELO value for new teams. As is, all new teams just start at 1200 which will usually put them somewhere below the middle of the active teams. This rating can be far too high or far too low and both has some problems:

If the new team is a newbie team, the older teams will be eager to play it before their ratings drops because they know they are easy to beat and will give good points. While it is good that the new team gets activity this way, it's mostly frustrating activity because they won't stand a chance against the superior teams. IMHO matches should preferrably played between teams of a similar skill level/ELO rating. Also, as wizard pointed out, the new team will see a steady decline of their rating.

If the new team consists of veterans, it will also start at 1200 and noone will want to play it because there is nothing to gain, but much to lose.

So it would be good to have a rough estimate for the strength of a new team. This has to be based on the result of their first few matches, though. So I agree with wizard, that a new team should not have any rating at all, but be marked as 'Jr.', 'novice' or something. After a sufficient number of matches (which would have to be defined), their initial rating is estimated. This rating is then used a posteriori to calculate the ELO shifts for the teams they have played.

This will only solve some minor problems, however, not the main one with activity or the problem that a team which is not (yet) so good might get frustrated and leave. It's a sad fact that almost all teams which have been in the lower third of the ladder didn't last very long. Which also means, that a new team will have the same problem, because those teams that are at about the same skill level have given up already, so they have to play vastly superior ones, which is rarely fun.

Posted: Sun Jun 25, 2006 10:20 am
by sussi
1) a too old player (more than 6 months) joins.
incorporate that---rooky player? what means every player who is shorter than 6 months in the league, is a rookie player?! I wlike that idea, but the time is to long, 3 months are long enough.
2) they reach an aim such as playing 10 matches.
That could be corupt by playing just 10 matches, better is to introduce some handwork for this topic.
They played 10 official matches the teamleader can apply for remove the Jr. tag. Obvious ..the council decide, or at least 3 members voting for. The 3 have some special resposibility for that team.
Jr-teams should be listed as a seperate section (like "Inactive teams
is a little unclear-- only active teams should be on the team-tab. I think Jr and regular teams are active, so both together on a single tab. Inactive and player without could move to the same tab, but not together with the active teams.