Monday, December 10, 2007

Pablito's Useless Stats

I do a lot of looking through statistics on Bungie.net trying to figure out what, if anything, can be deduced from them. Here are my two most recent ventures. The first is each clan member's K/D ratio for the 6 most common weapons and the second is about assists and has some interesting results. Read on to find out...

KofC K/D by Weapon
Pablito Neal:
Melee = 1.15
AR = 0.69
BR = 1.20
Sniper = 0.64
Shotgun = 0.74
Frag Grenade = 1.57

The Daymonster:
Melee = 1.01
AR = 0.70
BR = 0.63
Sniper = 0.72
Shotgun = 1.86
Frag Grenade = 1.30

Squatting Bear:
Melee = 0.95
AR = 0.98
BR = 0.96
Sniper = 1.04
Shotgun = 0.93
Frag Grenade = 1.29

Josey Wales:
Melee = 0.91
AR = 0.58
BR = 1.59
Sniper = 1.64
Shotgun = 1.35
Frag Grenade = 1.31

DammerWillEatU:
Melee = 1.07
AR = 0.92
BR = 2.11
Sniper = 1.22
Shotgun = 0.76
Frag Grenade = 0.96

Tilla1234:
Melee = 1.05
AR = 0.98
BR = 2.60
Sniper = 1.26
Shotgun = 0.65
Frag Grenade = 1.30

Unky Warthog:
Melee = 0.93
AR = 0.72
BR = 1.56
Sniper = 0.75
Shotgun = 1.82
Frag Grenade = 1.40

Unky Eternal:
Melee = 0.86
AR = 0.80
BR = 2.12
Sniper = 1.00
Shotgun = 1.08
Frag Grenade = 1.05

LocoPollo:
Melee = 1.02
AR = 1.23
BR = 2.97
Sniper = 2.22
Shotgun = 1.47
Frag Grenade = 1.13

Adeclipse:
Melee = 1.04
AR = 1.26
BR = 2.94
Sniper = 2.04
Shotgun = 2.27
Frag Grenade = 1.30



This can be added to the assist discussion. It is the percent of kills which were assisted in all of my Team Slayer matches.

4v4 Wins: 51.7%
4v4 Losses: 54.4%
3v3 Wins: 38.9%
3v3 Losses: 53.1%

Does this mean assists are important, in a bad way? This is from the team total, not just my kills/assists but it clearly shows that assists are more prevalent in our losses. I could use more games, the average sample size is about 20 but what conclusions can you other clan members draw from this?

11 comments:

Jeff said...

Those are some interesting stats. It would appear some of us are BR whores and Daymonster knows how to use the shottie.

I think the information I can glean from the assists stats would lead me to think that the reason it is higher in losses is that in general, as a team you will probably lose more 1v1 battles. This would mean that the kills you do get would probably be a team effort. Just a thought. I don't really know how that would factor in to the formula though. Does this mean a different assist multiplier should be used in a win vs. a loss to counterbalance the effect? Based on those statistics if you were to in general decrease the total number of assists by 8-9 % in the formula for losses to compensate it might work out. Pablito after work I'll send you the program and also tinker with the assist factor and see if it can get equalized. Also, where are you getting these in-depth stats from, I'd like to go there and play around?

Daymonster said...

The first stat is interesting but like the title says it's completely uselss. It's almost entirely based on what weapon you use the most. I almost never pick up the sniper rifle (except shotty/snipes of course) so I will most certainly die at the hands of a sniper more than I kill with one. And my inflated Shotgun K/D is the opposite of the sniper reasoning.

Overall K/D and Map K/D are reasonable statistics because it's the same for each person. While as the weapon K/D is almost completly based on how many times you have gotten your hands on a certain weapon.

For example: Josey dominates me with a 1.64 Sniper K/D to my shitty .76. But Josey dies an average of 1.11 times each game by a sniper, while I averge exactly 1 death at the hand of a sniper bullet.

But then when we look at shotguns, I have a 1.86 K/D with a shotgun and Josey a 1.35. But my D/G on a shot gun is .96 and Josey's is .85.

Daymonster said...

I have been tinkering with a formula as well, I am wondering how important are double and triple kills. Not on the medal aspect but simply on the fact that for a few seconds there are two or three players out of the game. Enough time to grab the rockets that just respawned or to recharge your shields for the next attack?

Also what about spree's I think that could have an impact because it's keeping you in the game long enough to get a rythm, but this could me just talking out my ass trying to get the formula more complicated like SB suggested.

adeclipse said...

I like these stats, despite them being useless. It all makes sense though. I hardly ever use anything but a sniper, BR, shotty, AR, melee or grenades. I usally won't pick up other weapons either unless there is something like a needler and people that just run straight at me everytime.

As for the assists I just don't know anymore. I used to think they were important and now y'all have me starting to think ah the heck with em. Who knows anymore. Even if they aren't that big of a deal i still like the idea that someone is watching my back and helping when they can to make sure the guy i'm fighting dies and i don't.

Jeff said...

Daymonster I think the streaks/sprees are important and should be taken into account. The one longer formula I posted in the assists thread used them at their most basic. The question though, is are we basing efficiency off of how hard a killionaire is to get, or off of how helpful it is to the team. Are they going to be one and the same? If you get a killionaire in a TS game (10 kills in a row, each within 4 seconds of each other) you would basically be eliminating the other team 2.5 times (in a 4 v 4). Improbable as that sounds, it would allow your teammates to whore up and get position, and also demoralize the other team. Obv this wouldn't really be possible in a 4 v 4 unless you were playing Blood Gulch in Halo 1 and just sitting on the ridge against nubs.

Can we take demoralization into account? If so then I think there should be an opposing team demoralization factor when you play with Unky Warthog and the map has a warthog (Valhalla for example). Not only will this complicate the formula it will make little sense.

Daymonster said...

The only issue we need this formula to use factors that can be found at bungie.net. So that can mean streaks but only the longest streak is recorded.

Also should we divide the kills, deaths, assists, by the team totals? I see the point but I am worried that a rating could be the same if it was 1 out of 4 kills and 12 out of 48. Just a thought.

Pablito Neal said...

right, the k/d by weapon doesn't really show which weapons we're best at, it only shows which ones you pick up most often because generally our opponents will use them all with same frequency. for every one of us the top 2 'deaths by' was AR (usually around 20%) and melee (about 17%). i'd say this is a good sign because if people are attacking us with these it means they've just spawned and we probably have better weapons.

the assists info i think shows that the games do basically boil down to 1v1 battles and if we can't win those then it's really hard to win the game.

Daymonster said...

I haven't checked this out, but I will give you (Pablito Neal), $20 if you can find me one person that their top two deaths aren't from melee and AR.... With a minimum of 50 ranked games.

Pablito Neal said...

http://www.bungie.net/stats/Halo3/CareerStats.aspx?player=Revelation%20Six

Daymonster said...

Touché, I owe you 20 Dollars... Worth the money to show that 99% of all halo players have AR and Melee as their number one killed by.

Now be a good sport and offer me a double or nothing bet.

locopuyo said...

When you lose you get a lot of assists because you don't have control of the power weapons (sniper, rocket, sword, etc)
that get you unassisted kills.

You end up using your assault rifle all the time which doesn't deal damage out very fast.