Stats Discussion

[pooty]

So I made this: https://docs.google.com/spreadsheets/d/ ... =671937450

Got the raw ranking data.

I put the formula in... sorted by adjusted PPH. Used 2.2 for spread.

And you just don't ever want Hector on your team.

[FuriousRabbit]

I think win % matters most.

The reason win % can be deceiving is because that's determined by the long term average quality of teams you're on. Leon, despite being a very good player that's very aggressive on nodes, only has a 49% win stat. That could be because he just unluckily gets put on lots of crappy teams. And there's players with higher win % that are definitely not better players than Leon.

The long term average of 1000+ games works the other way. It is highly unlikely that Leon is that unlucky and it is much more likely to be the influence of the multiplier. Win% is clearly the best summary number. I mean, that's the objective, to win the game. If your actions, whatever they are, contribute to winning, your win% should be higher.

[pooty]

And there's players with higher win % that are definitely not better players than Leon.

Are you saying its not how big but how you use it?

[captainsnarf]

So I made this: https://docs.google.com/spreadsheets/d/ ... =671937450

Got the raw ranking data.

I put the formula in... sorted by adjusted PPH. Used 2.2 for spread.

And you just don't ever want Hector on your team.

I think 2.2 for spread is wrong though.

I'll need to think about it more but my win rate is less than 50%. My multiplier should be less than 1.0. Using 2.2 puts it greater than 1.0.

[Enyo]

The long term average of 1000+ games works the other way. It is highly unlikely that Leon is that unlucky and it is much more likely to be the influence of the multiplier. Win% is clearly the best summary number. I mean, that's the objective, to win the game. If your actions, whatever they are, contribute to winning, your win% should be higher.

My point is that onslaught has too many variables and that win% is not the end all be all of all stats. Yes, the objective is to win the game, but what’s the best strategy (as represented by stats) to achieve the best win %? And by extension achieve the best balance.

Anon and I have very different play strategies, but have very close win%. I’m often quick to say that node heavy attackers win in ONS, but when the server puts me and Anon on the same team it’s often a slaughter because our play styles complement each other. He decimates everyone while I destroy and build nodes. And he doesn’t just blindly DM somewhere on the map, he’s strategic in protecting crucial nodes.

So, I’d say the most important aspect of onslaught play that isn’t quantifiable in the stats is situational map awareness of which nodes are the most valuable to attack at any given moment, aka strategy. Most players are oblivious of the map and don’t pay attention to which nodes to attack. Hell, I can’t even get what I consider good players to agree on a cohesive strategy on most maps despite years of evidence that certain strategies are repeatedly successful.

[captainsnarf]

but when the server puts me and Anon on the same team it’s often a slaughter because our play styles complement each other.

And that win will be represented in the win rate. It really is the best number to base any multiplier on considering team selection is random. You could easily be matched up with a vehicle waster that wastes all the flyers and power vehicles.

There are always gripes about 'DM players' on DMish maps. They might win more on those maps, but those maps don't get voted very often either, and the balancer already includes map specific PPH values so DM players on DM maps should be accounted for.

The balancer uses PPH. Any multiplier we add after should be based on something else. Nodes, damage points, kills, etc all give PPH.

[Enyo]

The balancer uses PPH. Any multiplier we add after should be based on something else. Nodes, damage points, kills, etc all give PPH.

All true, but since the balancer is based on PPH you have to give their multipliers weight based on the things that give you points. Nodes, damage, kills... you have to decide which is most important. IMO, in a game where the objective is capturing the nodes, points from nodes should have the most weight, and they currently do. BUT, if you get a lot of kills you score high there too, and not every DM player, most in fact, do not go to nodes to try and build them.

The onslaught game mode's name literally means an overwhelming aggressive assault. If everyone on a team goes after nodes aggressively, the DM players eventually get overwhelmed. Sure, the DM guys can help by stopping the other team's node attackers and are helpful in that regard, but you MUST have node attackers to win. I think the best balance of a team is 2/3 node attackers and 1/3 snipers/DM guys. Shitstains are the counter balance thwarting the node attackers by taking key vehicles.

[pooty]

I think 2.2 for spread is wrong though.

Maybe, that's why I made the spreadsheet, so we could change it and see. 2.0 felt like there wasn't enough spread to guarantee that you don't end up with a bunch of power players on one team. And the other aspect is some of the players with really good aim (eg. Sanka, Nautikal) aren't always point heavy, but they help the team MORE than other players with similar PPH.

Interestingly, with 2.2, good player multiples previously set by me were often much higher, while most of the shitstains were pretty close...

But ultimately it still comes down to:

situational map awareness of which nodes are the most valuable to attack at any given moment, aka strategy

I think Torlan-LinkerMadness is a good example -- last night we had excellent teamwork, comms and strategy (link trains), the opposite team did not.

[captainsnarf]

Multiplier formula = 1.0 + (((WinPct/100)-0.5) * scale)

In 4door's google sheet I used

Code: Select all

=ARRAYFORMULA(1 + (((D2:D35/100)-0.5) * 4.0))

If you want a bigger spread use a bigger scale.

Suggested multipliers

Code: Select all

Anonymous [38]    1.364
HalloweEny)o( [36]    1.464
Rhamp [573]    1.264
Xexx [527]    1.436
Tater [126]    1.356
leon [3]    0.972
Player 15 [77]    1.116
Meredith.Host [344]    1.072
GuythatkillsU [641]    1.068
AnKeeDo [967]    1.16
4Doors [80]    1.036
Sub-Zero [535]    0.976
candylick [548]    0.988
Spoon!! [94]    1.156
Kurt [534]    0.608
Judas [402]    1.076
EbNo [629]    1.252
Azazel [26]    0.96
YEAAAHHHHHHHHHHHH [40]    1.048
Zilch [190]    1.268
Sac [134]    1.228
FuriousRabbit [44]    1.048
nautikal [60]    0.976
MadeByTheLowestPidor [253]    1.26
Snake98 [34]    0.836
p)o(oty [22]    1.112
RP* [593]    1.092
false_flag [33]    1.116
TheAtheistGod [168]    0.988
Hector_Salamanca [27]    0.768
Super_Sanka [35]    1.152
Bisounours [117]    0.932
S)o(L{snarf [15]    0.944
Vugluskr [87]    1.052

[McLovin]

<1.0 and you're a shit stain? LOL

I resemble that remark. ^^^^^