Some Interesting Game Statistics

[Cubey]

The biggest indicator of a player's skill is their kills/minute. Not kills/death. A very slow-reaction player with poor hand-eye coordination can hide in spawn and get a kill every 4 minutes. This does not take any skill. The reason why kills/minute is the best indicator of skill is because in order to achieve it a number of important factors need to be satisfied. First, game-sense. A player must know where to be, at the right time, and when to anticipate the enemy. If somebody is getting many K/M then it indicates they have game-sense, otherwise, they be stuck in places on the map without much enemy action. Second, reaction-time. If one frequently gets into fire-fights and wins most of them, then it generally means that player has a superior reaction time. Since the faster player shoots first, shoot first, kill first, generally speaking of course. Lastly, superior hand-eye coordination, i.e. basically aim. It is conceivable that one has superior reaction time, but very poor aim, and that can lead the player to dying more often. Thus, when somebody has a high K/M rate, then it means their game-sense, reaction-time, and hand-eye coordination are working better than expected.

Some might object to the use of K/M as a measurement of skill. But then one has to supply a better indicator. For example, suppose the alternative, "caps per minute", C/M, is proposed. The problem with this measured is that one can have poor game-sense, poor hand-eye coordination, and poor reaction time, and still, despite all of that, get many caps/minute simply because they keep on retaking open flags. A relatively new player can achieve a high C/M rate by learning the routes and making it their sole purpose to cap as much as possible. Therefore, C/M is not a good measurement of skill. And so, if one objects to K/M, then one should a better alternative statistic to use.

Now we get to the interesting part.

I took the data of the top 101 players from the server. I took the top 101 for the sole purpose because those are some of the most frequent players on the servers. Below is a histogram plot of those top 101 players.
View attachment 18102
Notice how the data is not normally distributed. The four players who are in the over 2.11 K/M range break the normality. The four players who are in that range happen to all be former competitive DOD players. When these four players are removed from the data set look how shockingly normal it looks. There is a center at 22 height, and the other bars, on each side, are almost perfectly symmetrical.

In a way it makes sense why these four players create statistical anomalies. The rest of the data set consist of casual gamers. It is irresponsible to analyze this data by including four outcasts in the data set. Therefore, I have removed these four players and reduced the data set down to 97 players.

With this adjusted I ran four statistical tests to check for normality.

Anderson-Darling Test: 92% probability of normal distribution.
Jarque?Bera Test: 98% probability of normal distribution.
d'Agostino-Pearson Test: 96% probability of normal distribution.
Shapiro-Wilks Test: 97% probability of normal distribution.
Cramer-von Mises Test: 90% probability of normal distribution.

These observations support the hypothesis that K/M is an appropriate measure of skill since the skill level is normally distributed.
---------

The mean K/M score (in the adjusted data set) is 1.2
The standard deviation is 0.35

Therefore, if your personal K/M score is one standard deviation above the average, i.e. 1.55 K/M or above, you are in the top 16% of the skill among players. If your personal K/M score is two standard deviations above the average, i.e. 1.9 K/M or above, you are in the top 2% of the skill among players.

---------
However, kills/death, K/D, as many would say, is not a good measure of skill. Here is the plot of the top 101 players and their K/D scores.
View attachment 18103
At first glance this picture looks normal. It has a spike in the center and symmetry on both sides. However, the right has, one extra column spike, which suggests this data is not normal. Let us see what various normality tests indicate.


Anderson-Darling Test: 25% probability of normal distribution.
d'Agostino-Pearson Test: 37% probability of normal distribution.
Shapiro-Wilks Test: 51% probability of normal distribution.
Cramer-von Mises Test: 21% probability of normal distribution.

The picture and these tests strongly suggest that K/D is not even log-normal.
----------

One may also wonder between the relationship between K/M and K/D.

Here is the scatter plot of K/M vs K/D for each player. In this graph the two data sets were not adjusted.
View attachment 18104

Here the x-axis are K/M scores and y-axis are K/D scores.

The Pearson correlation coefficient is equal to .13

Therefore, there is probably no relationship between a gamers K/D and skill level. Some players with low K/D scores may be higher skilled and some players with high K/D scores may be lower skilled. Everyone already agrees with this sentiment, namely, K/D does not equate to skill. However, it is nice when the data actually supports that hypothesis.

 

[wewo]

Lots of time on your hands, eh? j/k, its kinda interesting.

 

[I SURVIVED LONDON BLITZ]

I go to bed every night worrying about my D/M no I go to bed every night worrying that DOD is dying and players are worried about their D/M. No wonder DOD ids dying if that is all they have to think about. Make some good suggestions on how to keep DOD alive. Please dont come back and tell me DOD is not dying I play every day at different times and can see it dying.


LONDON BLITZ

 

[explosion]

I'm not against the general idea. players with high kpm usually are skilled players
but there also are factors that effects kpm other than skill level

1st there are matters of holding back. even if one's skill level is high if he/she care more about kpd or points
he/she wouldn't go for as many kills as they can. even with good game sense, reflex and aim one can just decide
not to use it to full length and play safe. in which case one's kpm wouldn't accurately represent one's skill level

2nd there are matters of classes, if an assault, rifle and mg are having same kpm does it mean their skill levels are alike?
I think we can agree it doesn't. there are players who have unwavering determination when it comes to their choices of weapons if one always plays assault and the other always plays mg comparing their kpms directly would be inappropriate and correction would be required

I'm not saying kpm is a bad indicatior on the other hand it's very accurate one compared to the other stats
but it also has it's limits and should be used carefully imo

p.s. and of course the relation between one's kpm and one's contribution to one's team's victory is a whole different story we should think about

 

[GoT HiTReG?]

I dunno, I only ever played allies on Ava and waiting for a bad team to moveup while waiting out the nade wall doesn't help. These stats have a few too many factors since we only really have one map played.

I see you play support mostly which opens up the apartments, which isn't as viable for riflemen as a garand in CQ against an STG44/MP40 is asking to randomly get 3 tapped before your second shot gets off.

 

[Broadsword]

I go to bed every night worrying about my D/M no I go to bed every night worrying that DOD is dying and players are worried about their D/M. No wonder DOD ids dying if that is all they have to think about. Make some good suggestions on how to keep DOD alive. Please dont come back and tell me DOD is not dying I play every day at different times and can see it dying.


LONDON BLITZ

Weather it is dying or not you need to be more positive about it, I remember when Source took over from 1.3 and it was if Satan was roaming the streets and the world was coming to an end.

All the time someone is playing the game then it's not dying, the major clans have gone and most of the leagues have now gone, you can't play the game 24/7 now unless you want to jump from server to server.

Everyone plays this game differently, it's pub. You drop in, play and move on back to RL the days of playing for hours to try and catch peoples attention so that they will offer you a position in the clan to go on and play PCWs has vanished.

If Cubes has gone to the effort to research the above there's no need to trash him for it.

 

[BebeStars]

I love the fact that LONDON BLITZ up top stated the obvious of his worries plainly that the game needs new players.. However this other stuff doesn't matter to me so much but good job on your ability to use a spreadsheet and have all the free time in the world to do it actually. May the rest of us be so fortunate in that event to have so much free time !!

"Vanity and pride are different things, though the words are often used synonymously. A person may be proud without being vain. Pride relates more to our opinion of ourselves, vanity to what we would have others think of us.” -- Jane Austen.

 

[Brewmaster]

Ive said before - Play to have fun! Its a game for petes sake. Too many players worry about stats and all that stuff. I just try to enjoy my limited time gaming. Anyway enjoy and and I hope dod keeps on going as it is a really cool game.

 

[explosion]

Anyway enjoy and I hope dod keeps on going as it is a really cool game.

True that!

but I just wanna say there are many ways to enjoy this game. examining stats can be one of those ways imo. in this game u can learn a lot about other players. You can see one's kpd, kpm, what's one's favorite weapon, which path one likes to use and from where one usually hits his target etc etc, with those data from many players combined, u can deepen your understanding about this game(or at least about this awesome map called "Avalanche") and develop your point of view. That is one good part of the fun I think

It may sounds exaggerated but it's a little like sabermetrics(yes I know and I already said exaggerated) it shows the hidden(or at least not very visible) aspect of this game. for example let's say there's a guy with 0.7 kpd and 1.2 kpm. In my book he's a good player who does his part but his teammates might not appreciate him as much as they should or even blame him when they lose. with stats we can correct those problems at least in some limited areas. stats are not the purpose of this game but it can add something to this game imho

 

[Cubey]

One strong anti-cheat measure is to look for statistical anomalies. Even if a cheater is trying to keep a realistic K/M and K/D without drawing too much suspicion, their other statistical numbers will be way off from what they need to be. I have caught so many cheaters, without even watching demos, across many different servers, from looking at their numbers. Later I was vindicated when 2 or 3 months in they got VAC-banned.

 

[BebeStars]

One strong anti-cheat measure is to look for statistical anomalies. Even if a cheater is trying to keep a realistic K/M and K/D without drawing too much suspicion, their other statistical numbers will be way off from what they need to be. I have caught so many cheaters, without even watching demos, across many different servers, from looking at their numbers. Later I was vindicated when 2 or 3 months in they got VAC-banned.

How about the statiscal anomalies of shooting into spawn got stats on that?

How about the statiscal anomalies of capturing flags versus non capture of flags got the stats for that?

How about the statiscal anomalies of only shooting things in certain areas of map, say orange apartments, yellow apartments, double windows, crack over the 2nd flag, church, each of the teams first flags, ect..

I'd love to see this since you seem to to have this all spreadsheeted out and I'd like to see the information.

Thanks!

 

[Cubey]

How about the statiscal anomalies of shooting into spawn got stats on that?

Spawn shooting is perfectly allowed on eGO servers.
Here is the link of the official Leadership rule,
https://www.edge-gamers.com/forums/showthread.php?155457-DoD-S-FAQ-s#post1649350

How about the statiscal anomalies of only shooting things in certain areas of map, say orange apartments, yellow apartments, double windows, crack over the 2nd flag, church, each of the teams first flags, ect..

It does not matter where one is getting kills as long as they are getting a lot of kills. If somebody is getting 3 kills a minute in apartment it means the player is clearing out apartments from the enemy. If a player is sitting in apartments and nobody is going there, then their K/M will drop to 1 per minute or lower. It is all about adaption on where the enemy is. As long as the K/M is extremely high, it means the player is at the right moment at the right time.

 

[BebeStars]

It does not matter where one is getting kills as long as they are getting a lot of kills. If somebody is getting 3 kills a minute in apartment it means the player is clearing out apartments from the enemy. If a player is sitting in apartments and nobody is going there, then their K/M will drop to 1 per minute or lower. It is all about adaption on where the enemy is. As long as the K/M is extremely high, it means the player is at the right moment at the right time.

I will address the spawn shooting thing after I've read over that several times and make a conclusion about that in a moment..

But you have no statistical anomalies about people shooting in the same place at all which would allow them to get your ratio of whatever per minute. So that doesn't make anyone a better player it makes them a better CAMPER.

You my dear possible need to look at the statisics the way they are and they are quite skewed on not being able to break your numbers down for the specific points of intersection that I've asked for so that makes the system in which you analyzed it flawed.

If k/d whatever per minute is how someone needs to define there the best then possible they aren't the best..

This is all hypothetical statistics and anomalies don't you think as we are discussing them.

I find that no one player is really better than another, some are just better at camping, while others simply don't care.

I do give you credit for having alot of free time to analyze this stuff I don't know where you get all the time because my day seems to pass by quickly!

Good Job on your analyzing though my friend.. Keep up the good work..

P.S. Your more than welcome to private mail me if you decide to break down your statistics even further as I've suggest above..