Massive Assault

New Glicko Ratings Page
Page 2 of 4

Author:  Artanis [ Sat May 29, 2004 5:16 pm ]
Post subject: 

I think the maps should be equally weighted if for no other reason than that they weight themselves. Due to luck, it's much harder to get a very high or very low rating on a small map than it is on a bigger one: you're much more likely to get layout-screwed out of a win on Antarcticus than on New Paradise, after all. Thus over time, small-map matches will average out closer to 1500 than big-map matches, keeping it from affecting the score as much. My gut instinct is that this will weight the maps for randomness and amount of skill better than by throwing out arbitrary numbers.

...I probably botched what I was trying to say with this, but I hope it got through. Try to see what I mean, instead of what I say :wink:

Edit: grammar

Author:  Maelstrom [ Sat May 29, 2004 5:34 pm ]
Post subject: 

This is true in the case there are as many games on each map. But because players don't get as many games in on the big maps, they haven't always solidified to perfectly accurate results.

On Wasserland, the maximum number of games anyone has had is 16. While the maximum number of games that people have had on Antarticus is over 200. You can play several smaller maps in the time it takes to finish one of the larger maps.

Author:  Quitch [ Sun May 30, 2004 12:37 pm ]
Post subject: 

For it to be Glicko it needs to use Glicko methods, and weighting is just something we shove in because "it feels right". Just like we can't take into account a bad SA setup, I don't think we can start making calls on map difficulty, one persons difficult map is another persons easy one. mrakobes has beaten Tiger on NP, but cannot beat him on Anubis, which in itself makes a nonsense of the weighting system as I think we'd all agree that Tiger is a better player than mrakobes, yet surely if mrakobes can win on NP the *easier* Anubis would pose no problem?

It's a nonsense and should be treated as such. Every map is different and the sizes themselves are just an approximation.

Author:  redfox [ Mon May 31, 2004 10:38 pm ]
Post subject: 

Yeah, I'm for no weighting for the reasons sited by Quitch and Artanis. I agree with Arty's thinking about the equal randomness of small maps, and for the most part I think I've been extraodinarily lucky in most of my gams on Biz. I haven't played many games on Biz for the last month just because I want to gain mor experience on the larger maps, and I've been short on play time lately.

Author:  Maelstrom [ Mon May 31, 2004 10:49 pm ]
Post subject: 

Ok then :) I'll consider myself outvoted. I'll see what Norat can do about calculating the individual sides of each planet, then averaging them together for the final score as Quitch proposed. Considering noone else has thought it important enough to make a statement opposing this idea, I'll assume it's a consensus.

Author:  Artanis [ Mon May 31, 2004 11:18 pm ]
Post subject: 

Personally, I'd prefer if the per-side rankings were calculated seperately like the per-map rankings are.

Author:  Maelstrom [ Tue Jun 01, 2004 12:46 am ]
Post subject: 

Quitch wrote:
In my world your glicko rating changes as per normal, but takes into account the map and the sides (so if you play as PL on Emerald the score would be worked out using your PL on Emerald rating & RD against their FNU on Emerald rating & RD.) Once your new score and RD are updated, a new average glicko rating is calculated for your overall rating and RD. I don't see a better, Glicko way of doing it. Not only does this take map into account, but it also compensates for the FNU being tough on small maps.

Glicko would now break down into the following tables

Overall (your average score and RD from every map & side)
Emerald PL
Emerald FNU

That's what Quitch recommended too ;)

Author:  Quitch [ Tue Jun 01, 2004 3:41 am ]
Post subject: 

I think Artanis means the following:

Overall rating (The average of all your PL and FNU per map ratings)
Glicko FNU rating for map X
Glicko PL rating for map X

Glicko FNU rating
Glicko PL rating
Glicko rating for map X (discounting side)

I'd say that the first three are vital for as accurate as possible Glicko system as can be made for MA, while the last three make for interesting reading, but aren't required for accurate overall statistics.

i.e. The first three are required, the last three would be nice.

Author:  Three Seven [ Tue Jun 01, 2004 9:57 am ]
Post subject: 

How about taking trial players off the list, since most of them play for a bit and never come back, and get the people who were on for the free week and never came back. It just clutters up the list.

clutter.JPG [ 125.73 KiB | Viewed 26169 times ]

Author:  Artanis [ Tue Jun 01, 2004 11:08 am ]
Post subject: 

What I mean about seperately-calculated per-side rankings is:

You seem to be suggesting that it follow
PL Glicko = (Glicko formula for PL games)
FNU Glicko = (Glicko formula for FNU games)
Overall Glicko = (PL glicko + FNU glicko) / 2

I mean not to do this. Don't change how the overall glicko is calculated, keep it with how it is now with all games being run through the glicko formula, and the per-side glicko being done seperately for those who are curious about such rankings, like the per-map rankings are now.

Instead, make it
PL Glicko = (Glicko formula for PL games)
FNU Glicko = (Glicko formula for FNU games)
Overall Glicko = (Glicko formula for all games)

Author:  Quitch [ Tue Jun 01, 2004 11:19 am ]
Post subject: 

No, I suggest the following (and I have edited my previous post)

Glicko FNU Emerald
Glicko FNU Antarcticus

Glicko PL Emerald
Glicko PL Antarcticus

Overall ranking = average of all previous scores and RDs

I feel it's important to break down the sides per map, since this should help weight the FNU experience on Antarcticus appropriately, while side would have little effect on the score of a NP game.

Say there were two maps and my scores were as follows:

FNU Emerald 1074 50
FNU Antarcticus 2035 78

PL Emerald 1253 95
PL Antarcticus 2135 62

My overall would be: 1624 71

Say I played you on Antarcticus and was the PL. If I won, my new scores would be calculated by taking my PL Antarcticus score and comparing it to your FNU Antarcticus score, then our scores (your FNU my PL) would be changed, RDs updated, and finally the new overall score calculated.

Author:  Norat [ Tue Jun 01, 2004 3:03 pm ]
Post subject: 

To Three Seven:
You can set filter to hide inactive players for 30 (for example) days.

Author:  Three Seven [ Tue Jun 01, 2004 4:01 pm ]
Post subject: 

*slaps forehead*

forgot about that

Author:  Quitch [ Wed Jun 02, 2004 5:15 am ]
Post subject: 

There needs to be an MA and MAN version of the new system, otherwise the overall score of MA players will be terrible since they won't be able to improve on their 1500 Cratorthingy score (and any new maps in the future).

Author:  redfox [ Wed Jun 02, 2004 8:56 am ]
Post subject: 

Quitch wrote:
There needs to be an MA and MAN version of the new system, otherwise the overall score of MA players will be terrible since they won't be able to improve on their 1500 Cratorthingy score (and any new maps in the future).

I didn't even know that MA games were taken into account in the current system. Then again, I was never quite clear on the integration between MA and MAN. But yes, I agree that these games shouldn't be reflected in this system.

About calculating the Overall rating, I like Quitch's approach best. Specifically, I think it will help eliminate the problem of new players racking up a huge score by only playing on EMerald using PL option. It would effectively reduce the net effect of these games to a fraction of what they count as now, as playing on only one map and as only one faction is not a truly accurate demonstration of skill.

One question about this system, because there will be relatively few games played per map and per faction, won't the average of all RDs still be very high? I guess we'll see once we crunch the numbers, but this is what I would expect.

There were some good points brought up in this thread about player skills on different maps, and now I actually have some time to address it. Basically, my thinking is that each map has a learning curve, and in general skilled players will do better than less skilled players regardless of experience on a particular map. However, when there are two players of equivilent skill playing on a map where one has much experience and the other has very little experience, the more experienced player will probably win. A lot still depends on the luck of the draw, of course, but on maps like New Paradise, there are a lot of things you just have to learn from experience (or at least I do). I'd say the biggest difference experience makes is in the initial deployment phase, where you're given a set of SAs to work with, and you have to determine what's the best combination of units to use. For example, Wass and Crateus require radically different deployments, and they depend on both your SA spread and your faction. A skilled player will recognize this, but if they've never played on either map, they might not know what the best deployment or strategy will be.

Author:  Quitch [ Wed Jun 02, 2004 9:02 am ]
Post subject: 

I was thinking of something as simple as an "Include/Exclude MAN maps" option, or a page on the MA site which excludes them, and a page on the MAN one which includes them.

Author:  Quitch [ Mon Jun 14, 2004 4:23 pm ]
Post subject: 

Will we be seeing any of the proposed ideas integrated into the system?

Author:  Norat [ Tue Jun 15, 2004 2:34 am ]
Post subject: 

Yes, in unofficial Glicko Ratings page. I just very busy at work now and with my tournament battles, but I'll try to implement averaged by planets and sides Glicko Rating in near future.

Author:  Quitch [ Sat Jun 26, 2004 6:36 pm ]
Post subject: 

Any sign of this? :)

Author:  Artanis [ Sat Jul 03, 2004 6:01 pm ]
Post subject: 

Just to let you know, Twin Islands doesn't show up on the Glicko page yet.

Page 2 of 4 All times are UTC - 5 hours
Powered by phpBB © 2000, 2002, 2005, 2007 phpBB Group