Re-Organizing DSB

General discussion

Moderators: Beer Hunter, Tembest, Entr0py

Post Reply
Tembest
Posts: 692
Joined: Sun Dec 18, 2011 10:12 am

Re: Re-Organizing DSB

Post by Tembest »

Entropy, the problem is smurfing. Beastie cannot prac with anyone but smurfs to increase his rating, as for anyone with those 1700 numbers.
They have reached a number that is not possible after we disallow smurfs. Hence, the top will remain the same forever unless they on purpose start to
practice with those nicks, which will every single time cause them to lose rating. Not even Beastie has the slightest chance of winning a prac versus 3 average
players if he has to play with 2 newbies with rating of like 700 to balance the total ratings.

As you could see from my prac scores... I played around 100 pracs on my smurfs and lost only 10 of those. And that is with a rating of 1000. That's the only
way the best can improve their ratings anymore. Play with a good smurf on your team to decrease the expected winning margin, then rape by 10k. That's
the problem if we do not reset those. We just encourage them to never play on those nicks, because it becomes literally impossible for them to increase
the ratings once smurfing will be disallowed. The numbers are unrealistic.

Not to mention the other problem. It's supposed to be a one nick game, and now there are like 10 names per player in the top lists. Unrealistic numbers that
cannot be achieved when one cannot play with smurfs anymore, and same names in top lists who are just aliases of the same player.

The reason to reset those is to motivate people to play. People might actually care about their lifetime rankings if we made them matter at all.
Because currently, no one cares... They're messed up.

Edit: Oh and so far Ent, you have been the only one resisting the reset.

User avatar
falconeer
Posts: 205
Joined: Wed Oct 09, 2013 12:24 pm

Re: Re-Organizing DSB

Post by falconeer »

Does no one agree with me that a rating system for individual players, can't be based on wins or losses when it's a team game. We aren't rating squads here. Wins and losses should only be a stat. Can you remove rating wins and losses completely.

Also how about making the bot do automated teams based on player rating. So people with high ratings are always on opposite teams. Set the bot to detect aliases, and IP addresses so smurfs receive the rating of their original name. Then add an option to allow players to !register a new nick name. Players can have a maximum of two nick names. Upon !registering a third nickname their old nickname is removed from the ratings list. So I can technically start over and get a new rating under a new name. But yeah I think you can make it detect all of this. The best thing you could do ent, is code all of this in Python and make it a module. Even if its the old bot with no upgrades. The server can switch to ASSS, and then snake can improve your bot afterward. But I understand this is like 40 hours of work. A whole week of coding..

I agree with Entropy that Lifetime ratings is a good thing. We need a bot/module that can rate a players skill accurately, even if they are on a losing team. (IE WINS OR LOSSES DONT MATTER). And then we need a bot that uses that information to make teams that are NOT stacked. Thus game play remains fun and challenging, and not a torture for new players or a bore fest for old players.

What I disagree with is what is displayed. The life time ratings don't need to be shown. Only monthly and yearly need to be displayed in game. Lifetime is for the website. We also don't want to EVER show match predictions (IE freq 1 has 2000 rating advantage). That information is not necessary unless players use !teams option. The bot will make the teams. With an option !teams for players to make their own teams if they disagree with the bot. But the bot will make teams based on ratings so that the teams are even.

I know my ideas seem convoluted, but they are highly automated, and customized. This is a good thing, if anyone can code it and work their mind around the coding of it all it would be a very nice system.

jim the chin
Posts: 315
Joined: Sat Dec 24, 2011 1:17 pm

Re: Re-Organizing DSB

Post by jim the chin »

Entr0py wrote:I should point out here that lifetime ratings are not a "league". They are only supposed to be a reflection of how good someone is and help with match outcome predictions. There is no "winner" for lifetime ratings. Rather, the focus is supposed to be on the ratings for the current league, for which there is a winner.
But if everyone's score is reset to 1000, then why do we want the bot to predict that some players will win their matches? The players who are predicted to win will start off with a disadvantage won't they?

If you have to keep the existing lifetime ratings for calculation purposes (though I don't see why), then why not keep them on the bot, but without making them accessible for people to view with a command. Just make them some kind of hidden database for calibrating the bot.

Even if they are `comparable', I don't think anyone will want you to confound the new ratings with the old. The command !lifetimestandings will need to be cleared out (reset), such that for the first season of this league, !lifetimestandings and !standings are identical.
Please delete this account. I want nothing to do with this place any more.

User avatar
falconeer
Posts: 205
Joined: Wed Oct 09, 2013 12:24 pm

Re: Re-Organizing DSB

Post by falconeer »

Ballerina wrote:If you have to keep the existing lifetime ratings for calculation purposes (though I don't see why), then why not keep them on the bot, but without making them accessible for people to view with a command. Just make them some kind of hidden database for calibrating the bot.
This. +1

See I have a sense of humor..

jim the chin
Posts: 315
Joined: Sat Dec 24, 2011 1:17 pm

Re: Re-Organizing DSB

Post by jim the chin »

Tembest wrote: As the BONUS comes from kills, it directly increases the value of a kill. It does not increase the value of a flag.
Yes, the `Killbonus' increases the value of a kill with flag points, specifically 5*flagcount. Here's some examples with a base kill value of 80. The only thing I'll change in each example will be the flag victory (in bold).

Team 1 wins flags 12:8 but loses kills 70-80 in east.
Team 1 kill points = 5600
Team 1 kill bonus = 4200
Team 1 flag dings = 5400
Team 2 kill points = 6400
Team 2 kill bonus = 3200
Team 2 flag dings = 3600
SCORE: Team 1 wins by 15200 - 13200.

Note how Team 1's 70 kills are worth more than Team 2's 80 kills because of the flag points that have been added through the `kill bonus'. Lets see how narrow a flag victory is actually needed to defeat Team 2's kill advantage:

Team 1 wins flags 11:9 but loses kills 70-80 in east.
Team 1 kill points = 5600
Team 1 kill bonus = 3850
Team 1 flag dings = 4950
Team 2 kill points = 6400
Team 2 kill bonus = 3600
Team 2 flag dings = 4050
SCORE: Team 1 wins by 14400 - 14050.

Seems that winning flags 11:9 is still better than having 10 extra kills. Let's try an even smaller flag victory.

Team 1 wins flags 10.8:9.2 but loses kills 70-80 in east.
Team 1 kill points = 5600
Team 1 kill bonus = 3780
Team 1 flag dings = 4860
Team 2 kill points = 6400
Team 2 kill bonus = 3680
Team 2 flag dings = 4140
SCORE: Team 1 wins by 14240 - 14220.

That's the narrowest win you can have. In other words, you would only need a flag victory of 1.6 flags to offset the value of 10 more kills for the other team. That makes it way too easy to win by flagging in my opinion. A base kill value of 80 is too low.

If we do the same process with a base kill value of 120 points. We get this:

Team 1 wins flags 11.1:8.9 but loses kills 70-80 in east.
Team 1 kill points = 8400
Team 1 kill bonus = 3885
Team 1 flag dings = 4995
Team 2 kill points = 9600
Team 2 kill bonus = 3560
Team 2 flag dings = 4005
SCORE: Team 1 wins by 17280 - 17165.

That's the narrowest win in this case. So here you need a flag victory of 2.2 to offset losing by 10 kills. This seems a bit more realistic to me.

If you do away with the kill bonus altogether, then setting the base kill value to 80 points will probably work fine. With the kill bonus, it needs to be about 120.
Please delete this account. I want nothing to do with this place any more.

Tembest
Posts: 692
Joined: Sun Dec 18, 2011 10:12 am

Re: Re-Organizing DSB

Post by Tembest »

You're making the assumption that the teams actually hold BF's throughout the game. That's a rather silly assumption.
But, IF they managed to do that, shouldn't the victory be OBVIOUS? Really.. You manage to hold 12 flag average throughout the game
and lose by 10 kills? That sounds like a very well justified clear victory to me.

Remember..

"That's the narrowest win you can have. In other words, you would only need a flag victory of 1.6 flags to offset the value of 10 more kills for the other team. That makes it way too easy to win by flagging in my opinion. A base kill value of 80 is too low."

1.6 flags throughout the game. Now let me ask you, which is harder, to get 10 kills more or hold 1.6 more flags for 20 minutes?!?!?
If you say 10 kills is harder then... we cannot continue this discussion.


Should we perhaps just ditch the bonus? We would not really need it if kills and flags were balanced... and suiciding for flags wasn't
smart unless you obtain at least 3 flags upon your death. So as long as the kill value is higher than 2 bf's, I don't think we really need
the "hidden" bonus, which is annoying also because it cannot be seen in the statistics.

jim the chin
Posts: 315
Joined: Sat Dec 24, 2011 1:17 pm

Re: Re-Organizing DSB

Post by jim the chin »

Of course I'm making the assumption that teams will hold those flags throughout the game. If a team averages 12 bell flags at the dings, the common sense assumption is to say that they'd average 12 bell flags during the whole game too. Why would I assume something different? I think you're calling it silly because you don't want to consider the outcome of the scores.
Tembest wrote: 1.6 flags throughout the game. Now let me ask you, which is harder, to get 10 kills more or hold 1.6 more flags for 20 minutes?!?!?
If you say 10 kills is harder then... we cannot continue this discussion.
Really? Then I guess this discussion is over because it's easy to average 1.6 more flags than the other team over the course of a game, but getting 10 more kills than the other team demonstrates a genuine difference in ability.
Please delete this account. I want nothing to do with this place any more.

User avatar
Entr0py
Posts: 135
Joined: Sat Dec 17, 2011 7:54 pm

Re: Re-Organizing DSB

Post by Entr0py »

Update:

1. The carry stats are now fixed for the individual sectors. Note that the "Total" line is not necessarily the sum of the carries from the 4 sectors, since other "sectors" are included in that total as well (e.g., TCL, DSBL, etc.).

As a side note, ignore the "1000" for winR/powR in the individual sectors. The bot does not track winR/powR for individual sectors, so those numbers are kind of placeholders. I didn't have the energy last night to make those spots blank.

Tembest
Posts: 692
Joined: Sun Dec 18, 2011 10:12 am

Re: Re-Organizing DSB

Post by Tembest »

jim the chin wrote:Of course I'm making the assumption that teams will hold those flags throughout the game. If a team averages 12 bell flags at the dings, the common sense assumption is to say that they'd average 12 bell flags during the whole game too. Why would I assume something different? I think you're calling it silly because you don't want to consider the outcome of the scores.
Tembest wrote: 1.6 flags throughout the game. Now let me ask you, which is harder, to get 10 kills more or hold 1.6 more flags for 20 minutes?!?!?
If you say 10 kills is harder then... we cannot continue this discussion.
Really? Then I guess this discussion is over because it's easy to average 1.6 more flags than the other team over the course of a game, but getting 10 more kills than the other team demonstrates a genuine difference in ability.
I would assume the flags held at other times would be 10-10 unless given a reason to believe otherwise.
It's the only assumption we can have if we want an unbiased calculation... That's why I'm calling it silly. And the truth is quite funny as well because often
the team winning bell flags is the "tubing" team, which is actually the team that holds most of the time only 8 flags during the prac. So it's turned around.
Anyhow... as we cannot make such assumptions... it needs to be considered 10-10.

I hope you get the point. They are like separate games, holding flags, bell flags and kills. Three different games, and because we are
discussing the relation between kills and bell flags, you should consider holding flags 10-10. You cannot assume either team does it better, or you cannot
compare the absurd relation between kills and bell flags.

So... your first calculation is actually supposed to look like the following:

Perhaps you finally understand what's wrong

Team 1 wins flags 12:8 but loses kills 70-80 in east.
Team 1 kill points = 5600
Team 1 kill bonus = 3500
Team 1 flag dings = 5400
Team 2 kill points = 6400
Team 2 kill bonus = 4000
Team 2 flag dings = 3600
SCORE: Team 1 wins by 14500 - 14000.

Now with your 120 points...

Team 1 wins flags 12:8 but loses kills 70-80 in east.
Team 1 kill points = 8400
Team 1 kill bonus = 3500
Team 1 flag dings = 5400
Team 2 kill points = 9600
Team 2 kill bonus = 4000
Team 2 flag dings = 3600
SCORE: Team 1 wins by 17300 - 17200.

One kill would change the winner in your system....

It's getting really frustrating soon :) I'm sorry man but I just cannot understand how you don't realise the how the bonus acts as its own game kind of
and the only assumption you can make is that it is 10-10. Either it's 10-10 or not included in the calculations at all... Your call.

But what I just illustrated above should now clearly explain how ridiculous 120 points would be. It would not change the game in absolutely any way from
how it works right now. Only kills matter. Bell flags are meaningless. With 120 kills, 50% victory in bell flags can be overcome by a 14.5% victory in kills.
Talk about that...

Tembest
Posts: 692
Joined: Sun Dec 18, 2011 10:12 am

Re: Re-Organizing DSB

Post by Tembest »

Oh and to simply demonstrate it further...

Here is how it currently works:

Team 1 wins flags 12:8 but loses kills 70-80 in east.
Team 1 kill points = 10500
Team 1 kill bonus = 3500
Team 1 flag dings = 5400
Team 2 kill points = 12000
Team 2 kill bonus = 4000
Team 2 flag dings = 3600
SCORE: Team 2 wins by 19400 - 19600.

The difference between your 120 points system and the current is that Team 1 victory (100 points) is changed to Team 2 victory (200 points).
That could be kind of called a 300 points difference. Do you still think 120 kills would change absolutely anything?

With 150 points kills Team 2 wins if they get 10 kills and with your 120 points Team 2 will win with 11 kills.

There is a one kill difference between what you're proposing and the current system.... which is not a system at all. It's just a kill fest.

Post Reply