Loading...
  OR  Zero-K Name:    Password:   

Even 1250 elo is too much for complete newbies

25 posts, 2409 views
Post comment
Filter:    Player:  
Page of 2 (25 records)
sort


11 years ago
I keep mentioning this from time to time, but I've decided to just make a thread of it, see how it goes, and then I can shut up about it forevar.

As some of us may know, newbies get an elo malus of 250 in ZK, which makes them start at 1250 elo instead of 1500. This elo malus gets smaller after each game they play (as they attain an attribute called weight/reliability), and after around 20 games, they no longer have the elo malus.

This is very cool and all, but imo the elo malus is not big enough, and 1250 expects too much from new players, making their first games often quick, steamrolley losses.

I think it'd be much better for the newbies themselves, if the game considered them to be mostly clueless about everything (because that's what they often are when they join the teamgame host) - and newbies that have taken the time to learn things before joining teamgames deserve to be rewarded with some wins anyway.

So what would be a fitting elo for a "clueless" player? I'll trow the number 1100 out there. I think this would reflect the skill level of our average newbie better, and would mean that the team should get a 1900 elo player to balance out the newb, who would be able to carry the team somewhat.

Finally, I think the elo malus should take more games to go away; around 1,5 times the current would be good. This would give newbs a little longer to be newbs. :)
+1 / -0


11 years ago
At that very time, would make smurfs able to smurf longer, and with more deleterious effects.
Aside from that, i see no flaws in the proposal.
+2 / -0


11 years ago
Smurfs are a tiny minority of the new accounts that are created, and smurfing is bannable now.

In this case, I think we shouldn't let the small number of people abusing the system deter improvements.
+3 / -0
Skasi
11 years ago
I keep mentioning this too, so I'll just do it one more time:

Gather stats on how much skill first-time players have. To do this, keep track of all wins and losses of first timers in relation to their win chance and calculate their expected skill level.

Alternatively we might want to consider taking other systems as an example. According to Wikipedia's elo rating system article, the United States Chess Federation rates beginners at around 800 Elo. Other sources I found seem to confirm this.

quote:
In general, 1200 is considered a bright beginner.
source
[quote]<1,000 - New player[/quote][url=http://www.thechessworld.com/learn-chess/18-general-information/136-deeper-look-at-elo-chess-rating-system-]source[/url]
+3 / -0
quote:
Skasi Gather stats on how much skill first-time players have. To do this, keep track of all wins and losses of first timers in relation to their win chance and calculate their expected skill level.

While a very cool idea, that requires some actual effort on the part of the implementer. Merely increasing the elo malus doesn't.

[quote]<1,000 - New player[/quote]
Interesting! Maybe even 1100 elo is too high to reflect a new player's skill.
+0 / -0
Isn't our system too small for fixed number "average" elo balancing?

We have a continues stream of new players entering ZK and most of them leave after some games or after a couple of weeks. All this time they feed elo points into the elo-system. This means that the average of 1500 doesn't hold true anymore. If you'd take the total amount of elo points of the active players and divide it over the active players you'd get a number other than 1500. Over time the 1250 starting elo will be the right figure though! :P

I'd promote a system that keeps track of the performance of new players and calculates the average win chance of a total newb. You'd have to give the new player elo penalty to reach that number.

A problem here is that you'd need to keep track of the elo penalty / level for each player. It would be determined by a players joining date.

If your statement of "1250 is still too good" is true we would see it in the results of games that include newbies. We can calculate the above mentioned figures from the difference in assumed win-chance and the real wins.


EDIT: Skasi, we think alike! XD We posted the same ideas
+0 / -0

11 years ago
1000 elo or less
+0 / -0
11 years ago
I don't even know that they should not be given a negative elo. In 10v10 (what this is about?) these < 1300 elo creatures also get their share of the team's metal. If this metal went to the good players on the team perhaps the chance of winning would go up. New players with no clue how to play can be a detriment to the team and not a benefit, however small.
+0 / -0
Skasi
11 years ago
quote:
these < 1300 elo creatures
+0 / -0


11 years ago
It doesn't make sense to add or multiply elo's by each other. As in elo has no 0, it's just an arbitrary scale. Negative elo means nothing. We could reduce all elo by 1500 and half the players would have negative elo, no change. Technically any player with elo below the team's average is 'preventing' them from winning but with the extra management that is often not the case.
+0 / -0

11 years ago
some new players play actually above 1250 elo skill level. So that averages out nicely.
+0 / -0


11 years ago
quote:
some new players play actually above 1250 elo skill level. So that averages out nicely.

The new players who win games gain elo at a very fast rate (because elo changes are bigger for new players). Thus, the few newbies that play above their elo aren't a problem, and they get proper elo quickly.

It is because of the many players who join "teams: all welcome" without any knowledge of the game, that I am suggesting making the elo malus bigger.

The current situation:
*New players who have some skills do okay.
*New players who don't have much clue get curbstomped.

What I'd like to see:
*New players who have some skills are rewarded with wins, and their elo adjusts quickly to account for this.
*New players who don't have much clue do okay.
+0 / -0
Skasi
11 years ago
Do missions give XP? If so, then missions give levels and levels decrease the -250 elo we have. That said, we could increase the 250 elo "penalty" to 500 and give bonus XP for completing tutorials and other SP missions for the first time. Then new players who completed tutorials will start close to the 1250 elo we currently use, while others will start at 1000.
+4 / -0
I don't see the point in this topic. Afterall, everyone starts at that elo right..? So 1250 is like the 0 line.
+0 / -0
Short explanation: 1500 is the "0 line". Newbs get a malus of 250 which puts them in 1250. If 1500 is considered 0, one could say newbs start at -250.

quote:
FIranksprang As some of us may know, newbies get an elo malus of 250 in ZK, which makes them start at 1250 elo instead of 1500. This elo malus gets smaller after each game they play (as they attain an attribute called weight/reliability), and after around 20 games, they no longer have the elo malus.
+0 / -0

11 years ago
I like the thought that the elo system actually also balances this out. If complete newbies are rated too high, they'll keep feeding elo into the "active" players until the elo of those players scales up, which increases the differences from the 1250 elo and thus increases the (negative) weight of new players.
+0 / -0
11 years ago
the elo malus is not a real value

if 1500 elo player loses to 1500(-250) player
the 1500(-250) player gets 50 elo making them a 1550(-200) player

the reverse:
if 1500 elo player beats 1500(-250) player
the 1500 player gets 5 elo making them a 1505 player

assuming there is at least 1 "LEET UBER SKILLZ NUB" in eatch batch of 10 "nabatron 9000's" there is a net gain of 0 elo in the system

ie. there is no elo inflation, and if you asume taht there are more then 1 skillz nubs for evry 10 no skillz nubs this is actualy causing elo deflation

tl:dr due to the elo malus avarage elo is (most likley) dropping towards 1250, not increasing
+0 / -0

11 years ago
So you would suggest newbie firstly choose their start newbieness levels?

I'm glad to see 3 sets, 1000, 1200, 1250.
+0 / -0


11 years ago
GBrankKyubey: the 1250 elo newbie would still get an elo increase in the reverse scenario (the malus decreases even if you don't win)
+0 / -0
11 years ago
O.o really, i was sure that loosing with the malus just dropped your "real" elo to compensate

ie. 1500(-250) loses -> elo becomes 1450(-200)
+0 / -0
Page of 2 (25 records)