Right now we just rank by kills. Why not change it slightly. The top 10 in kills get sorted out by a combination of stats.

Kill Ratio
Kills/Deaths
Kills/Round
Kills/Hour
Round win percentage

Use those 5 stats to make up the top ten. Use skill or Win percentage as a tie breaker if need be. Just an idea to make things more interesting.

sounds good (says the guy with a bad rank )

yeah...they should change the ranking a bit more but i think your plan is pretty good...nice idia

I like the idea, except I'd widen it to the top 25 or top 50. In a previous post, I had recommended something similar: Rank score = Skill + x*Kills, where x could be greater than 1 to emphasize Kills or less than 1 to emphasize Skill. Skill seems to jump around alot (at least for me it can vary by 400 pts depending on if I have a good night or not). So your idea of using a combination of the other stats might work better than using Skill. But I would still reward longevity (and create an incentive for people to play) by keeping the # of Kills in the equation. Otherwise, SwanDude or Pizza, for example, who are way ahead of everyone else and wont drop out of the top ten (or 25 or 50) in kills, could stop playing now if there stats are good enough and still win. Also, # of Kills would stabilize the Rank, so that it doesn't jump around as much and it would make the last couple weeks before the Stats Series ends much more interesting: Do I play more to increase my kills with the chance that I play badly and lower my other stats?

Well, I only said the top 10 to ensure a large enough sample size. Top 25 is fine too. But top 50 may be a big stretch. Right now the 50th has just over 500 kills. #1 has over 4700. I wouldn't want to see someone with 500 kills outrank someone with 4700.

I also tried to choose ratings that did not force you to play more in order to improve. Except for kills to actually get there. There were more, I only picked a few. But this is what the rankins would look like for the top ten.

1. Pizza
2. One
3. Jinx
4. Deathbal
5. Bruin
6. Sylar
7. Tnick
8. Loke
9. Brain
10. Swan

I see where you are going, but I would still add one more factor for # of kills so that you wouldn't have to worry about the 50th (or 25th) place player beating the 1st place player. Add the four stats as you did, but also add the number of kills of the player/the number of kills of the leader. As I write this, Pizza is the kills leader and has 4772 kills, so everyone gets their kills divided by 4772 and the result is added to the score that you calculated. Pizza gets an extra 1.0, Swan 0.95, Deathbal 0.62, Bruin 0.55, etc. The 50th player (who has 542 kills) would get 0.11 added.

The top ten under this scheme would be
1. Pizza (4.48 pts)
2. One (3.53)
3. Deathbal (3.45)
4. Jinx (3.39)
5. Bruin (3.24)
6. Sylar (2.84)
7. Tnick (2.62)
8. Swan (2.61)
9. Loke (2.3)
10. Brain Damage (2.26)

I would probably also adjust the weighting of the all the factors because the Kill Ratio and the Win Ratio are much smaller (for the top ten players between 0.39 and 0.13) than the Kills/Death and Kills/Round (between 2.2 and 0.8 ). Maybe divide the Kills/Death and Kills/Round by 2 or 3. Another way would be to normalize all of the stats (divide by the leader's score in that stat like I did for # of Kills) to make everything equal weight. The leader in each stat would get 1 pt for that stat and everybody else would get a fraction of that depending on how they compare to the leader.

Doing this would still make # of Kills important, but it would only be 1/5 of your score - some of us could get lives. Obviously I would wait for the next stat series before making a change.

Ok, let me tell you how I came to my top ten.

All I did was see where each player placed in each category, then added them up. For example.....I placed 6th, 3rd, 4th and 4th. It equals 17 and placed me 4th.........The lower the number the higher the rank. Pizza was 2,1,1,1, which placed him first. Swandude for example was 10, 10, 10, 9, which placed him 10th. If you wanted to use that method, and add kills to it, Swan would be 10, 10, 10, 9, 2........I tried to make it as simple as possible. This way it doesn't matter how low or high the actual numbers are.

We also must take account of each player's dog weight (giving additional points to those who has labradors or dobermans), which we multiply by number of hairs on grandma's head. And sure it is benefit if you like japanese food - additional points also.

...

Sorry for acidic tone.


"Skill ratio".
As there are no official comments about it I share results of my own observations. It might be not precise but very close to it. Been watching this for myself and several other persons with whom i played.
This indicator is very far from reality.
This value depends on score you earned recently (i would use term density of score) and total number of games won.
While density is high, skill slowly goes up, but several rounds without score are enough to drop it down significantly.
Total number of game victories serve as a "firm ground" for the level at which skill varies for the person.
You can compare stats from previous series for the same persons with their current to see it.

For me with my above 200 it is now about 3000, if i got some not succesful rounds it can drop down to 2900-2800, if series of succesful games - rise (had it up to 3500).
One month ago the range for me was 2100-2500. Bruin, for example, now has this within 2300-2600. Db is now at level of 2100-2400.
My alias Look At My Avatar (in top 25 extended) with about 25 game wins has the skill is about 2100.

How make it growing?

You need a stable high density of score.

But what is the source?

It all depends on crowd you play with.

There are periods of time at server without strong players when it is easy to gain many kills and win games.

Once I played a game with two brave beginners that ended with 20 kills and 10 round wins, and it made my 'skill ratio' rise by about 500. (My usual level was about 2100 at that time).

Exactly after that game I realized how really stupid this indicator is.
Then to try I once played for several hours from blank account (they give 1000 skill to all at first connection) - games against beginner players - to make it above 2000 (alias Aim also in top 25 ext).

Without many game victories at this alias several rounds with low density would be enough to make 'skill ratio' insignificant. And to rise it back, again, some hours of steady high density score playing. Though, with desire to get 'highest skill ratio', would be enough to keep on catching noob time, play games 100% ending victory and bringing good score for 'skill ratio' to rise above the sky.

When two strong players play 1 on 1 mcb, it is often something like 2-3 round wins both and 6-8 kills.
You can play best game in your life but it will make your 'skill ratio' drop down.
While two noobs can have a game ended for both with 2-3 kills and 6-8 round wins and it make their 'skill ratio' rise, 'cause better score.

So in first case winning game (well they give some little points to skill ratio for recent game victory) is worse for this indicator then in second - losing.
Though it obvios in reality if you played with very good player(s) you could have not won game and it is not the same that somebody else has won but with bad players.

Also if you are a very good player and play with many sucker-nature persons, who tend to unite and kill you first shot with 5 funkies each round, you end game without a single round victory, and even though all 10 shots you had in game you made it to kill, the whole thing will make your 'skill ratio' go down.

But, average players can play games against beginners and gain many round wins and kills and hence - rise of 'skill ratio'.

So this indicator has absolutely nothing with real quality of playing skill - which can only be determined by own experience of playing together for long enough time.

To fully comprehend its ridiculous nature, check 'skill ratio' of teams server bot #1. Btw, he has 8% kill ratio, though (read further) it is not really matter, but if you ever played at teams server against this bot...

In my opinion, 'skill ratio' should be abolished.


"Kill ratio"

In general persons who have this indicator high are very good players.
But it also depends on nature of games you having.
If you play mostly mcb games you probably have it lower then you would have in ffa.

If you play ffa games with many strong players you tend to have it higher.
In that case choosing of target is mostly based on its accessibility, and also most effective weapon is used.
Usually even after first shot many are dead, so all make their best to use the only chance.
And only few who survive have to perform more difficult shots which require several tries (so making kill ratio less). But as on average it happens to each once in several rounds, you simply have not enough tries to "spoil" it.

On the other hand, when most players are not strong, better ones have to perform more difficult shots to get others from all around the map - because others have not enough skill for mutual kill even though their shots are easier then yours. And also if you know you have enough tries you use small cheap weapon, which requires more tries in general.

Your usual choice of weapon has influence on it.
Also teleports and fuel are counted as 'shots'...

So this value depends too much, and also can be kept high purposely.
One player - known as a very good - has his forum account referring to alias with a very cool kill ratio, but he does not play with that alias - probably not to spoil it?
His other aliases though have very high kill ratios, but not that high.

From real experience. I played many 1 on 1 mcb games with these players, they are all very good shooters and they would win and i would win, though difference of their kill ratios:
Outer - 36.5, Kaze - 30.3, Sabi - 30.0, Dustkiller - 23.2 (!!!) - even funny, in fact one of the best shooters I met.

So this value is not quite real.

In some mods, where fuel and shields are cheap, it looks absolutely senseless.

I would propose to use instead something like accuracy ratio.
If it would be possible for server to calculate the result of each shot taken separately, with conditions 'no others shots', 'no movement', 'no shields' and 'no chutes'. And hits of shield shall be counted also.
So we could record the real quality of shooting - even if in reality tank moved, or kill was stolen, or it was dropped down by another shot (for example happens often when you try to use napalm - it still burns and is to kill but some moron's funky comes and spoils the whole thing), and we record the effect of diggers/riot bombs (often happens then player gets kill only because another player made hole, but not even assist for that), and effect on shield, and of course all this gives different values to accuracy ratio.

"Kills per round"

Again, too much depends on what crowd are you playing with.
Check Vihor's from previous series. Not really high? And the reason is obvious.

"Round win percentage"

Also same thing. In games with many strong players even 2-3 pretendents for victory often have only 1-2 of them - and that is 10-20% - not high aint it?



In my opinion, rank shall be abolished (along with skill), though kills can still be default value to sort stats, and top 10 on main stats page can be called 'top 10 killers'.
As for other changes in stats, I would like to see kill ratio for each weapon in individual stats - though it is not a real indicator, but interesting.

And, would be good to have speech activity of players measured, in characters or words, and based on it - words/hour and words/kill ratio.

Well, I didn't include Skill except for a tie breaker. My suggestion was simply to improve a bit on a Kills only ranking system. Nothing is perfect, there are always flaws. I will not change how I play. I also have a feeling most people wouldn't. And of course a lot of stuff depends on who you play. Kills depend on who you play too. I got 33 kills one game, take a wild guess who I was playing? mostly noobs.....25 of those kills were probably riot bombs and diggers.

Anyway, it was just a suggestion to spice up the ranking system. I also like the ranking system. It makes this game different in a way. Usually stats are for sports games only.

You - will not. But as we know this game is a drug giving lot of pleasure and making people addicted, and for some people it also has pronounced hallucinogenic effect.
Example of this - several persons from team server top 10, who dont play at all on main or beginners. If you join when they are playing, they dont reply to greetings, dont say anything, if it is last rounds try to kill you quick with funkies or sandhogs and then disconnect when game ends, or in early rounds disconnect at once when you join.
If all you need is game with bots you can have it offline and nobody will interfere (unless you seriously schizophrenic), so I assume the purpose of their playing is seeing their names in "top 10 list", probably giving them some special kind of satisfaction.
On main also there are pretty lot of players which like to play only games they can 100% win and they do not join games where they see very good players...
And such 'dependable' stats indicators, are like a candy for that sort.

Just recalled, in fact I decided to reply to this topic after one guy in game asked me odd question: is 'skill ratio' depending on low angle shots? So I told him that i reply here and say what i know about this ratio, but it gives me another idea about what we could have in stats.
Would be nice to have a diagram for each player about angle usage, on X axis - angles 0...90, on Y - number of shots made. IMO would provide really interesting information about players.
Also a diagram with rotation directions, but in this one indicate not shots but kills. I bet pretty lot of players would have significantly more kills in 4 specific directions

I don't think the problems you bring up will have that much of an impact. It will have some, but it won't change anything by that much.

I have started games on many occasions where we have 4-6 top players and I think, WoW, this is going to be a great game. By the 5th round 1 or 2 of those great players remain. I see this happen quite often. There will be people that simply do not like playing an average game. It has little to do with stats. They do it now, they'll do it no matter what changes. So the stats will probably remain as they are now.

Keep in mind, my suggestion still requires you to have enough kills to be in the top 10 or 25.........so you cant have someone play a couple of great games and sit on the number 1 rank.

By the way, I like your idea on keeping track of players angles. But i'm not sure what you meant about the 4 directions.

I think its really difficult to "measure" the skill of a player, but one thing that hasn't been mentioned is what type of weapons they are using. How about using the kill*armslevel 10+kill*armslevel 9+etc/total kills, then you will get an average armslevel of the weapons used. If one thinks the current armslevel is too high or low it can be corrected by multiplying it with an appropiate modifier.

The average armslevel will indicate if this player is hitting his targets spot on or just funks and dh´s mainly.

Yeah you gotta include that in there!

Good point Loke.

BTW Guardian Angel...

Your posts.

Almost all of them...

Are like a book!



Wow you must be able to type really fast or you sit there all day posting a reply.



You seriosly do need to write a book!

Make it on scorched 3d or sumthing!

Here is the average armslevel pr kill for the top10.

Pizza 6.31
Swan 5.96
DB 7.35 (must have made an error somewhere;)
Bruin 6.55
tnick 6.05
Jinx 5.45
One 5.88
Loke 6.71
BD 6.40
Sylar 6.00

Vihor 7.19

Mind you the assigned armslevels(AL) need some modification to reflect the skill in using the different weapons. All rollers are AL 8 and the heavies are really heavy;)

Loke