Theodin's Book of Melee

[Theodin]

Can you identify someones average impact rating not based off the total number of kills they have. For example, what is a players impact rating at a lower average of kills and deaths(less playtime). Calculating a stat for players who are not playing every single minute of a match, could be highly beneficial for a team captains ability to identify the most impactful players. That being said I know nothing about calculating these stats, I just find it as an interesting idea to implement.

Well two things - AIR is based off of the ratios not volume so less playing time shouldn’t theoretically affect stats that way. But also tonight you’ll see Russian’s stat which directly takes rounds played into effect!

[BillTheButcher]

2014-2016 was peak NW melee

it was the golden ages of NW when shit actually mattered

and everyone on his list is from those days so................go fuck yourself

damn that's crazy

dan was never any good at melee he wouldnt understand

[Hawkince]

2014-2016 was peak NW melee

it was the golden ages of NW when shit actually mattered

and everyone on his list is from those days so................go fuck yourself

damn that's crazy

dan was never any good at melee he wouldnt understand

[Fartknocker]

2014-2016 was peak NW melee

it was the golden ages of NW when shit actually mattered

and everyone on his list is from those days so................go fuck yourself

damn that's crazy

dan was never any good at melee he wouldnt understand

Dan literally invented turning to the right. What did you invent in this game Princess?

[Theodin]

just kidding russian's stat is getting posted tmw along with (hopefully) the ratings from the match between gvt and bbg

[Dolcebowler]

2014-2016 was peak NW melee

it was the golden ages of NW when shit actually mattered

and everyone on his list is from those days so................go fuck yourself

damn that's crazy

dan was never any good at melee he wouldnt understand

Dan literally invented turning to the right. What did you invent in this game Princess?

[Theodin]

So, Russian's stat! It's got a bit more math in it - but fans of volume-based metrics and FBR will love it.
It's called RPR - Russian's Player Ratings.

This is a screenshot of the doc with the first formula:

Spoiler

[close]

The stat does (Kills-TKs-First Bloods) + First Bloods^1.5 / (Deaths-First Deaths) + First Deaths^1.5 to calculate IPR (Initial Player Rating). This is done with the idea in mind that First Bloods should be valued more than regular kills, and vice versa for First Deaths.

Now to add in rounds, the formula is

Spoiler

[close]

Which is IPR / the square root of rounds played (in this case 91) *10.

So in this case the sheer volume of rounds played increased Risk's score by 0.06!

This stat would scale like this:
0.8 or less is bad
0.8-1 is below average
1-1.25 is average
1.25-1.5 is good
1.5+ is very good

[Theodin]

Let's see these stats in action! Last night's League 2 match data has been released, so lets break down how people did!

AIR:

Spoiler

[close]

Scaled we have
0.79 or less is bad - Jolly, Cow and James had tough outings
0.8-1 is average - Dio's FBR brought him down a chunk
1-1.3 is above average - Copot and Wolfe had similar statlines, Killershark's FBR kept him in this category, Phil's great FBR night brings him up here
1.3-1.5 is spectacular - no spectacular performances last night
1.5+ is elite - Goomba and Antidote had elite nights, especially in first blood volume and ratio

I've also done AIR above Match AIR:

Spoiler

[close]

RPR:

Spoiler

[close]

Scaled we have
0.8 or less is bad - Both stats reflect tough nights for Jols, James, Dio, and Cow
0.8-1 is below average - Phil's FBR can't save him from this as it does in AIR
1-1.25 is average - No average performances tonight
1.25-1.5 is good - Good performances from Copot, Wolfe, and the Shark
1.5+ is very good - Goomba and Antidote break the charts

Some thoughts:
 - Phil's AIR shows why judging impact from KDR alone is not good - his 4-1 FDR score saved his rating
 - We changed the *adjustment in RPR because league 2 round samples are small
 - RPR rewards good performances and punishes bad ones, while AIR seems to be kinder
 - Dio and Phil are the only ones who changed scaling groups between stats. In both cases it seems to be because their FBR was wildly different than their KDR

Anyways interesting, won't be as many comments in future weeks but I'll do this for every match!

[RussianFury]

[Fartknocker]

The future of NW is now

[Hawkince]

So, Russian's stat! It's got a bit more math in it - but fans of volume-based metrics and FBR will love it.
It's called RPR - Russian's Player Ratings.

This is a screenshot of the doc with the first formula:

Spoiler

[close]

The stat does (Kills-TKs-First Bloods) + First Bloods^1.5 / (Deaths-First Deaths) + First Deaths^1.5 to calculate IPR (Initial Player Rating). This is done with the idea in mind that First Bloods should be valued more than regular kills, and vice versa for First Deaths.

Now to add in rounds, the formula is

Spoiler

[close]

Which is IPR / the square root of rounds played (in this case 91) *10.

So in this case the sheer volume of rounds played increased Risk's score by 0.06!

This stat would scale like this:
0.8 or less is bad
0.8-1 is below average
1-1.25 is average
1.25-1.5 is good
1.5+ is very good

Fancy way of saying kd with first bloods and first deaths worth more

Stop trying to revinvent the wheel son

[RussianFury]

thanks hawkince!

[Hawkince]

thanks hawkince!

no problem!

[Theodin]

thanks hawkince!

[Dan the Seagull Chef]

How do these metrics relate to the touch-grass spectrum?