Evaluation System

[Arcaon]

Negative Point Accumulation System (NPAS):
The best score on a map is counted as 0. The diffecence in milliseconds to rank1 is defined as the point value. Points add up over maps. The player with the least points, wins.

Example:

Code: Select all

23:000, rank1, 0 Points
23:080, rank2, 80 Points
...

This would be my choice.

23:000, rank1, 0 Points
23:080, rank2, 80 Points
24:264, rank3, ??? 1264 Points ?

24:264 is 16 times worse than 23:080.

:>

[<hk>]

but it still lacks the point where it all went wrong last year. Which binaries will be legit? Will we get another adjusted binary like last year, which wasn't all that great and some people never even noticed the announcement of the binary, or will we be able to use certain existing binaries (baseq3, dfengine, ioq3, ...) or only one specific binary?

If I knew, I would have posted it into the INFO thread.

What I can say, since I take care of the anti cheat stuff is that you can very likely use any binary you want.

[<)Ghost(>]

just so you know I didn't read everything in this post, I skimmed through it, and if you think I'm being an asshole, good.

I think you're making this more complex than it really is...

say you have 5 maps, with player 1 times like:
20.000
25.008
30.176
22.032
10.016
total is 107.232

divide that by 5 and you get 21.4464 (always rounding down; 21.446)

compared to player 2 times like:

19.992 (-.008)
25.000 (-.008)
30.184 (+.008)
22.040 (+.008)
10.000 (-.016)
total = 107.216/5 = 21.4432.... total average of 21.443

the player with the lowest average time wins. in this case player 2 wins by .003 seconds. don't count by frames, we're not talking about q3 anymore, we're talking about a ranking system.

IF for some reason there is a tie on the total average time (which is highly unlikely, [especially in the top 10], but still possible,) then it would be a tie.

forget about a ranking system with points, you're only going to make something so simple, too complex, that you'll have to explain to the moron's a million times for them to barely understand it.

fair/unfair shouldn't even be apart of this, if you think its unfair gtfo.
the player with the best time wins. who cares if ones route is faster than another, if he got a better time than some1else; well, that's it, he got a better time.

"If someone is close to another player in one map he should get the chance to make that count."
I disagree, sure hes close, and I'm sure he did his best, but he didn't get the best time. there's no reason to qualify him to the next round or give him another chance because he was close. If you don't make the cut your out, no need to be sentimental about it.

There's my official 2 cents.

/commence flaming.

[Arcaon]

say you have 5 maps, with player 1 times like:
20.000
25.008
30.176
22.032
10.016
total is 107.232

divide that by 5 and you get 21.4464 (always rounding down; 21.446)

compared to player 2 times like:

19.992 (-.008)
25.000 (-.008)
30.184 (+.008)
22.040 (+.008)
10.000 (-.016)
total = 107.216/5 = 21.4432.... total average of 21.443

the player with the lowest average time wins. in this case player 2 wins by .003 seconds. don't count by frames, we're not talking about q3 anymore, we're talking about a ranking system.

A system based on individual times would probably be one of the most fair ones, providing that the rank 1 points are fixed. E.g. 100 points for rank 1, and everything else relative to that.

If it was some system where the times were simply added together in some manner, factors like individual strengths (someone only being good at strafe maps for example), map exploits (i.e. dfwc2010-4), and map lengths would weigh too much.

I think I'll have to agree with the coolest man in the universe on this one.

[$$-albion]

A system based on individual times would probably be one of the most fair ones, providing that the rank 1 points are fixed. E.g. 100 points for rank 1, and everything else relative to that.

If it was some system where the times were simply added together in some manner, factors like individual strengths (someone only being good at strafe maps for example), map exploits (i.e. dfwc2010-4), and map lengths would weigh too much.

I think I'll have to agree with the coolest man in the universe on this one.

Can't seem to remember that I wrote that, hmmz...

[<hk>]

19.992 (-.008)
25.000 (-.008)
30.184 (+.008)
22.040 (+.008)
10.000 (-.016)
total = 107.216/5 = 21.4432.... total average of 21.443

the player with the lowest average time wins. in this case player 2 wins

Averaging only is useful if players don't play the same number of maps. In case of the DFWC it will just be the total amount of points divided by 6. Always 6. The values will just be lower. No gain in information.
So much for not making it as complex as it should be.


I can see you guys are quite attached to the old system.
How about this one?

• Winner gets 100 points. (Normalising point 1)
• Winning time+50% is 0 points (Normalising point 2)
• Everyone slower than the winning time + 50% is a loser and doesn't get any points.
• In between we just use their time relative to the best player. and normalise it in the same way.
  Every one would get a point value between 100 and 0.

This is pretty much what I was suggesting before but you can now get 100 points tops. Not more. The winner will always get 100 points and can't own others on other maps.

[RaneFire]

The normalisation points 1 and 2 you suggest still present a problem for short maps which aren't a frame-war. IMO if you're going to design a points system you have to take the map itself into question (weapons, 1st place time and time differences), otherwise you're trying to do the impossible.

One map is an apple, then other is a lettuce, then all of a sudden u get a loaf of bread. What do u do? You can't apply an arbitrary stand-alone rule to the unpredictable results defrag can give. But for an apple-of-a-map, this time is so. Same for lettuce and the bread, look at each in context.

On the plus side it's nice to see so much thought going into points and ranking systems. Good work.

[TittenIgnition]

Ok, I think this pretty much covers my 2 cents.

albion, thats less like 2 cents and more like $5.

For $5 I'd write an essay ;D :p Btw what did you think of my masterpiece? :p or did you just read the tl;dr :p

it was a good read but i dont really know what to say. ive learned to keep my mouth shut most of the time because whenever i open it something stupid comes out.

or, in the case of the internets, i try to keep my fingers off the keyboard unless im playing a game. this post is an exception

[Arcaon]

• Winner gets 100 points. (Normalising point 1)
• Winning time+50% is 0 points (Normalising point 2)
• Everyone slower than the winning time + 50% is a loser and doesn't get any points.
• In between we just use their time relative to the best player. and normalise it in the same way.
  Every one would get a point value between 100 and 0.

Sounds plausible.

Results from last DFWC, round 0-3:

Formula: 300 - 200b/a
a = Best time
b = Time

1. Arcaon - 17.984 - 100
3. Bazz - 18.200 - 97.60
5. glm - 18.312 - 96.35
46. Nt-kop4ik - 19.016 - 88.52

1. Nt-kop4ik - 32.608 - 100
3. glm - 34.280 - 89.74
4. Arcaon - 34.416 - 88.91
5. Bazz - 34.672 - 87.34

1. Nt-kop4ik - 36.736 - 100
2. glm - 37.888 - 93.73
4. Arcaon - 38.024 - 92.99
6. Bazz - 38.376 - 91.07

1. Arcaon - 35.464 - 100
3. Bazz - 36.440 - 94.50
5. Nt-kop4ik - 36.616 - 93.50
6. glm - 36.736 - 92.83

Scores:
1. kop4ik: 382.02
2. Arcaon: 381.9
3. glm: 372.65
4. Bazz: 370.51

Hmmm... the ranks look right, but the scores are still debatable.

Edit: Including round 4:

1. glm - 02.840 - 100
2. Nt-kop4ik - 02.992 - 89.30
12. Arcaon - 03.816 - 31.27
19. Bazz! - 04.312 - 0

Scores:
1. glm: 472.65
2. kop4ik: 471.32
3. Arcaon: 413.17
4. Bazz: 370.51

Hmmm... even more debatable.

[RaneFire]

Round 4 kinda proved my point lol.
These are the abnormalities, any ranking system will work if the results are expected ones.