For starters, AscendantOne, you need to become familiar with this website or one similar to it ...
Personally, I have not run any statistical analysis on anything COD since an infamous MW3 episode which is better left in the past for fear of opening old wounds.
Nonetheless, to get a super-accurate picture of trends in kdr, spm, etc, etc ... you need to look at about 1000 different AND truly random players. However, there is a way in which you can look at fewer and determine whether or not the patterns that emerge are "reliable." To make things "easy," I suggest you pick about twenty random players - not players you know, not players you've played, not players you "think" are reflective of typical player "types." Twenty absolutely, 100% random. Using the alphabet and a random number generator, it is pretty easy to find twenty random players.
Using the "statistics" on those players' Elite profiles, you can use the site mentioned above to use actual, real, bonafide, theoretically and practically proven mathematical formulas or models or whatever you want to call them to actually determine the relationship between any of the stats you see on those Elite profiles.
Many community members have already done those things in years past. You can take our word for it or you can rack your brain out trying to do the same thing ... but you're going to get the same result ...
A + B = B + A ... that's what your GIN formula reveals. It quantifies information about the individual player, it does nothing to qualify the player against other players.
The statistical math, however, is going to reveal that you have to look at a broad spectrum of data to determine the actual "quality" of the player. Even then, player quality is subject to various conditions.
There is no 100% reliable objective statistic to determine player quality.
First off I didnt say the GIN formula was a groundbreaking discovery, I am simply proposing a more convenient way of tracking the impact of a player to the game.This is simply comparing your performance to the average of all the players you have faced in all your games.
Second, I am aware everything has standard deviation, almost nothing is CONRETE, let alone Call Of Duty Statistics. What color underwear you will choose in the morning has standard deviation, unless you dont wear underwear. The best you can do is increase sample size to minimize this deviation.
It is unnecessary to pick random players to do analyses for this statistic. The randomness is captured by the opponents a player plays in their games. For Team Death Match, lets say a player plays 1000 games of CoD, which in reality is not that much, that is roughly playing 4 hours a week for 41 weeks out of the year. In that time they will have played against 1000*6 = 6000 different and essentially random players (far greater than the 1000 you suggested). The GIN is a lifetime statistic that evolves over your playing career with you. After playing 6000 different people, it will approach a nominal value, but will obviously have some high standard deviation on a game to game basis because CoD players vary so much (like you said). What I am getting at is that after 1000 games, the GIN will be accurate, a player with a higher GIN than you who has also played 1000 games WILL produce more kills and die less than you on a regular basis, and therefore win more on a regular basis. I would guess off the top of my head that you would have to have a GIN of more than 100 from another person to witness the difference on a game to game basis playing with them.
What you are saying about variance is true, and where it is true is in this case: If a team has an average GIN score of 175 and the other has an average of 150, they might have to play 100 games against each other to see the difference.
If a team has an average GIN score of 350 and the other has 200, you could probably see the difference in only 10-15 games since the GIN statistic is exponential in nature the difference between a 350 player and a 200 player is very significant.
You have brought a valid point up though, it would require testing to quantify the value of the statistic. Testing I am willing to do when the game comes out. After my friends and I have played more than 200 games each, I will use the players to make teams based on their average GIN. One really good team with over 300-400avg GIN, one with 200avg GIN and one with 100avg. These teams will play together for a night and keep track of their wins vs losses and see if their is significant deviation between the results.
You have failed to take into account w/l. Meaning we might have a guy that has John Rambo like stats, but he knows how to win. In games that are close, he comes through and goes positive, but in games that blow outs, he is running around trying to "Misery Loves Company" everyone on the map.
I believe w/l says a lot about a person.
You are most certainly correct, a lot can be said about Win/Loss ratio. You would still have your W/L ratio available to you obviously. This statistic isnt suppose to replace any of them (Kdr, spm, wlr), but be added to them and compliment them.
There also isnt a strong correlation between WLR and the GIN. You can have someone with a high GIN and low WLR because he always plays by himself and gets horrible teammates who have a .5 KDR haha. If they have a high GIN and high WLR chances are they play with a group of friends.
I'm not following you, Ascendant. If the stat doesn't tell us anything more than we already know, then what purpose does it serve? I can think of a purpose (which I stated upon my first impression of your idea) but I'll refrain from stating it for the sake of level-headed debate.
In addition, the testing you propose would still have a bias that is blatantly apparent in your proposal - you're testing the idea based upon party play. Without a doubt, using your formula, a team of players who regularly, routinely, always play together and would have a GIN of 100 would be an extremely poor performing team. That's obvious because of something much more obvious: if their kdr, spm, etc are low to begin with, the GIN is going to come out low. Like I said A + B = B + A. Watching how the average moves over time is equally not going to reveal anything that is not already known fact - most of the time, the more you play, the better you get; the better you get, the better your stats look.
In fact, I'm really not seeing how GIN would be better than SPM. Score Per Minute already takes into account kills scored and point-providing actions that the player takes during a match. Every time I play I see guys with substantially higher SPM than I have, but with substantially lower KDR. Those guys almost always have positive WLR. But so what? I have a positive WLR, as well. In fact, my WLR is about par with theirs. As far as high SPM players who have substantially lower or substantially higher WLR than mine, it's about even either way.
In other words, going by the relationship of SPM to WLR, as for WLR I think I'm very near the top of the bell curve. As for SPM, I'm definitely in the lower fifty percentile.
What are they doing different than I am?
I'll let you answer that question before I tell you what I've found to be the most common differences, one of which overwhelmingly stands out above all others.
Excellent thread. 5 stars for originality and substance.
The only adjustment I would make is the camper that goes 12-4 and scored 350 or so in your GIN. There ought to be a minimum number of kills before they can enjoy scoring good GIN. For example, they have to have scored 15 or more kills with a KD of higher than 2 to even be considered for GIN. A 12-1 is rare. Usually, the unskilled camper, if they do well, goes 5-1 and that's no impact at all except for 75-74 games.
BTW try to ignore some of the forum regulars. I don't remember when's the last time some of these guys took the time and effort (like you did) to try to contribute rather than persistently naysay. (not you, nut.)
Does it matter how you win? Because if it does not matter how you win, just whether or not you win, then why don't you just assume losses = left early?
There is a difference between giving the ol' college try but losing and quitting/dashboarding. If there was a stat like Hob suggests I'm betting there'd be less of them. Lets try to imagine it, would a would-be stat padder find much satisfaction if the stat is there for everyone to see.
Not a bad idea at all, IMO. In fact, if they're going to add any new stat, I would pick this one.
What color underwear you will choose in the morning has standard deviation, unless you dont wear underwear.
LOL. Nice analogy (it's only inaccurate if all I wear are tighty whities.)
I wish people would take the time to understand what a stat really is and, more importantly, is not.