On May 30 2014 12:42 TheYango wrote:
Remember that Riot has also admitted in the past that matchmaking is actually supposed to partially account for blue side imbalance. So unless this changed or is disabled in OFA, the effect is actually even more dramatic because Riot is already trying to correct for it.

It probably has to do with a number of things. Vision angle could be one of them, but it's definitely not the only one.

EDIT: One thing of note is that viewing angle as a source of imbalance does not only exist on Summoner's Rift. If you wanted to look at the effect of viewing angle in isolation, we coincidentally have a game mode that is played on an otherwise symmetrical map, but is subject to viewing angle-based imbalance (ARAM).

IIRC riot said when they matchmake 10 people they make the most even 5v5 possible and then give the stronger team purple side. It might have had significance in the past, but with such a large player base nowadays I believe that matchmade teams are so close in MMR that they are essentially even. (With exceptions at the very top or bottom of the ladder)

And blue side is advantaged in howling abyss. I'd like to see stats for TT or dom since those don't have the top/bottom camera asymmetry (although iirc left side wins more still)

Sufficiency

Canada23833 Posts

May 30 2014 04:38 GMT

#1205

Never mind, on Twisted Treeline the win rate is equal on both sides.

Camera angle it is...

Sufficiency

Canada23833 Posts

May 30 2014 04:44 GMT

#1206


OneForAll blue side win (1 = won):

 team_id | won | cnt
---------+-----+----------
     100 |   0 |     1505
     100 |   1 |     1946

For one for all, blue wins ~56% of the time.


Twisted Treeline (normal blind + ranked team)
 team_id | won | cnt
---------+-----+----------
     100 |   0 |    34865
     100 |   1 |    34722

For TT, blue wins ~50% of the time.

ShaLLoW[baY]

Canada12499 Posts

May 30 2014 04:53 GMT

#1207

one for all fizz was actually fun, thank god

Goumindong

United States3529 Posts

May 30 2014 05:09 GMT

#1208

On May 30 2014 12:22 Sufficiency wrote:

Show nested quote +

is the effect that big though? When a champion wins 55% of the time, that champion is OP. You can get the same effect just by being on the blue side.

In any case, to verify what you said, I need to collect more data and plot the blue side win % per champion. It is possible that champions with more skillshots are more heavily affected by the blue side.

Possibly more interesting. Champions which are considered OP are picked first more often. This conveys a side advantage and a champion advantage because first pick is blue side (At the very least, contested champions are more likely to land on blue side if they land anywhere)

Moreover an analysis of champion advantages which does not consider Blue/Purple* side pick discrepancy will have omitted variable bias unless we believe that the sole of the advantage comes from picking first rather than the UI/Camera angle. And any analysis of blue side advantage must take into account the strength of champions if it wants to examine the advantage irrespective of pick order.

*either check to see if it exists then ignore if it doesn't or consider blue/purple side in the analysis

edit: To see how this works, suppose a champion legitimately had no effect on the outcome of the game, his win rate in a vacuum would be 50%. Suppose as well that the blue side advantage is 100% UI/Vision and not pick order.

If he were picked 90% on blue side (56% win rate) and 10% on Purple side(44% win rate) his actual win rate would be 54.8%

Simply the perception of a champion being strong can lead to a blue side bias, which leads to an actual win rate advantage

Goumindong

United States3529 Posts

May 30 2014 05:11 GMT

#1209

On May 30 2014 13:44 Sufficiency wrote:


OneForAll blue side win (1 = won):

 team_id | won | cnt
---------+-----+----------
     100 |   0 |     1505
     100 |   1 |     1946

For one for all, blue wins ~56% of the time.


Twisted Treeline (normal blind + ranked team)
 team_id | won | cnt
---------+-----+----------
     100 |   0 |    34865
     100 |   1 |    34722

For TT, blue wins ~50% of the time.

So i guess the answer is "dodge unless you're blue side" then.

Sufficiency

Canada23833 Posts

May 30 2014 05:13 GMT

#1210

On May 30 2014 14:11 Goumindong wrote:

Show nested quote +

On May 30 2014 13:44 Sufficiency wrote:


OneForAll blue side win (1 = won):

 team_id | won | cnt
---------+-----+----------
     100 |   0 |     1505
     100 |   1 |     1946

For one for all, blue wins ~56% of the time.


Twisted Treeline (normal blind + ranked team)
 team_id | won | cnt
---------+-----+----------
     100 |   0 |    34865
     100 |   1 |    34722

For TT, blue wins ~50% of the time.

So i guess the answer is "dodge unless you're blue side" then.

I wonder how big the dodge penalty is.....

Sufficiency

Canada23833 Posts

May 30 2014 05:16 GMT

#1211

I think another exercise is to take a look at blue side advantage on this map for each champion.

I suspect long range champions, particularly the ones with skillshots, will have even greater advantage on the blue side.

Goumindong

United States3529 Posts

May 30 2014 05:27 GMT

#1212

On May 30 2014 14:13 Sufficiency wrote:

Show nested quote +

I wonder how big the dodge penalty is.....

MMR? Its Zero*. LP, its like 3 but it pretty much gets given back to you your next game. Biggest downside to dodging is the timer. Other than that, dodging a game you think you will lose is always a positive EV play in terms of your MMR and your net division ranking (so long as you play enough games to cancel out the LP loss by automatic gain)

*MMR penalties were removed because people were intentionally tanking their mmr by dodging. This intentional taking by dodging makes MMR systems not work. Besides creating lots of toxic games where these people stomp, it can create a true "ELO Hell", where games get easier the higher your MMR gets, for a certain range of MMR discrepancies. For the proof of this(iirc, its been a while since i've done the math) suppose some % of people dodge a certain MMR rating reduction, then calculate the new expected MMR distribution and take the derivative of the ratio of "dodgers to not dodgers". You should note that the amount of dodgers decreases as you increase in MMR and that the "expected mmr" of the enemy team (relative to yours) can decrease as you increase in MMR. Fortunately this situation can no longer exist.

wei2coolman

United States60033 Posts

May 30 2014 05:48 GMT

#1213

all hecarim is super fun. i'm pretty sure my q button is broken.

739

Bearded Elder29905 Posts

May 30 2014 06:00 GMT

#1214

All Ziggs works too, dodge them bombs.

Inflicted

Australia18228 Posts

May 30 2014 06:04 GMT

#1215

On May 30 2014 15:00 739 wrote:
All Ziggs works too, dodge them bombs.

More like dodge ur teams' satchels

Eiii

United States2566 Posts

May 30 2014 06:06 GMT

#1216

On May 30 2014 15:00 739 wrote:
All Ziggs works too, dodge them bombs.

The most hellish all for one game I played before was lux vs ziggs

80 minutes of pain

Now that you can control what your opponents get, I look forward to teaming up with four buddies to force alistar every single game no matter what