Speculation: Is ARAM really random?

MURDERBRIDGE

Upon Murderbridge's release I found myself in love with the new ARAM that riot had put out. The banned items and summoners made player honesty and integrity a non-issue and the map forced people to stay on the single lane. The thing that really pulled me in was the random nature of ARAM, I would have to quickly learn and adapt to playing any champion against any team.  

 

The oncoming storm

Some of the issues that started my frustration were:

-Players consistently dodging (this I will never understand, its random for a reason)

-Certain champions being consistently IMBA (OP) and carrying 1v5 from the start: Galio (MR + poke), Nidalee [gets fed, spears], Kayle (just seems to beast), Graves (once he gets going GG), Ryze (1v5 no problem at 6),  Blitzcrank* (gives your team a win), Xin (guaranteed 2 kills if he dives, snowballs), Akali* (murders squishies but tanks are a problem. 90% win hits hyper mode too fast), Malz (DoT hits so hard so fast) and Varus (Q poke gets insane and he spirals out of control in a heartbeat)

-5 player poke teams

-5 tanks

-3 AD carries and 2 Supports

-Getting the same champion 2-3 games in a row or multiple times within 10 games

As you can see some frustration is really the RNG of League while the rest is player inconsistency. The problem that bugged me the most was the consistency of my randomed champions. After about 100 games where I felt I had been randomed fiddle at least 20% of the time I started to log the random champions I got. My results were very disconcerting.

 

 The Experiment

Factors:

I own 64 champions currently, I generally have 2-5 free week champions. This was done over 2 rotations so we will add 13 champions I didn't own to my total bringing my pool to 77 total champions.

 

Mathematical outcome (perfect average)

If Random was perfectly random over 410 games I would have randomed everyone between 4-6 times.

 

Hypothesis

Pretty much what I have found from the first 100 games that I didn't document. Certain champions will be selected more often and certain champs will have very few random selections.

 

Results

Of my past 410 games I found that:

1. I was significantly more likely to random: Nasus (11), Vladamir (11), Taric (10), Udyr (10), Irelia (11)

Why these champions for me? I have no idea, but the moment I hit champion selection and see Vlad or Nasus I sigh. It isn't fun to hit that select screen and get someone that often especially in Nasus' case where he doesn't fare that well in ARAM. While Vlad, Udyr and Irelia generally fit well into an ARAM team Nasus and Taric can really fall flat on their faces. My first thought with all these was build tanky and that seems to work but sometimes you really can't get going and you will get smashed by the other team. So roughly I have bad luck.

Bonus points: I main AD carries. 4 tops and a support get spam picked for me. Enjoy that laugh

 

2. I was least likely to random (owned champions): Alistar (3), Gangplank (2), Kayle (1), Nunu (3), Sivir (3), Sona (3), Twitch (3), Vayne (3)

Why is this, seriously confused with this one. Kayle, one of my favorite champions (I have a fascination with angels) gets 1 hit in 410 games. GP gets 2 hits, I mean seriously here what is driving Nasus to 11 and Kayle to 1. Its bananas, but such is RNG and horrible luck I suppose. I main Vayne, play a lot of Kayle and Nunu but the rest of the champs here are rarely touched for me. I will on occasion play GP or Twich, so there is really nothing working in the RNG to get lower used champions you own to be randomed more often (which is good and random). The fact that I play some of these champs and don't play others ever actually does mean that the randomness is random due to it not being affected by what I play or don't play.

 

3. I was likely to random free week champions at least twice while free: Diana, Gragas, Hecarim, Heimer, Jayce, Lulu, Maokai, Renekton, Sejuani, Singed, Victor, Voli, Zyra

This makes sense to me as the fact of the matter is the free week champions will have no selections and be more likely to get randomed IF there is a sort of balance failsafe in place. However I feel like that only applies to non-owned champs. Probably for the sake of forcing players to have variety outside their normal selection pool. Either way I found myself getting the free week champions a couple times which was kind of nice to know that the next random could be a free week Zyra or Nasus.

 

4. I found that the game will double up on champions about every 100 games: Fiddle, Akali, Udyr, Morgana, Nasus and Noc all had at least 2 in a row. Fiddle won with 4 in a row.

This was really odd to me. My roommate was really the one to stumble on this when he got trundle. SEVEN TIMES. We were compiling data together but he bailed on the google doc at 28 games and made a Open office one that disappeared or something so. Basically I found this a strange glitch but quite nice on Noc as I enjoy stomping around with him. Fiddle was the most annoying to get in a row, in fact I don't think I ever just got fiddle as a one off, it was always in a pair or more.

 

5. I also found that the random generator will sometimes random champions that had the same number of randoms as if adding to their total across the board

Example. anyone with 4 or 7 randoms would be more likely to be selected if the previous champion had been randomed that many times. 

This happened 3-7 times in a row a total of 6 times while running this experiment.

This was another really strange blip in the data. I noticed it happen a couple times and then started recording it. It would happen in packs of 3-5 at first but at game 289 it started a chain of Karthus, Akali, Amumu, Irelia, Yi, Teemo, Sion before going off to characters that were not at 6 selections. Kind of strange, makes me think the random selector was "capping" off champions in order of selection to maintain an average thus forcing a high and low end to the selection. Which would mean that the game's random isn't actually that random whatsoever.

My average randomed characters were selected between 6-8 times

Due to free week champions swapping in and out the low end data encompasses them as well as my owned champions that were bad luck or something. This forced the average up from 4-6 to 6-8 with the top end being 10-11. So due to the high amount of champions that had less than 4 selections, the data got skewed with a lot of champions sitting on 7 selections. It seems that for every champion above 7 selections there were 2 champions below 7 selections. I think the game is playing favorites.

 

6. I had 21 characters not be randomed whatsoever due to either not owning them or not getting them on free week. (Most notably Galio)

 

This is not strange as not owning a champion and them not being free will out them. However of the 21 champions there were 2 free week champions I didn't random at all. Looking back at the randoming free weeks at least twice this puts a nice big hole in that because there is at least 1 in each free week that did not get selected whatsoever. I want to attribute this to bad luck but honestly I can't say why this one happened with any accuracy.

 

Conclusions 

So what can I really say about the data I gathered,  ARAM is "almost random".  If I ran the data for 10,000 games I am sure there would be vast differences in champion selection with certain champions being more likely to be selected than others. However there is always a top end and a bottom end to the data which is not surprising as most champions will end up as an average in the middle. 

 

From what I know about random number generators and the seeding thereof as well as my data and speculation on Riot's random champion selection method:

- There will be about 5 champions you will find yourself being randomed more often than not (these will be unique to the player)

- There will be about 10 champions that will be randomed very rarely (these will also be unique to the player)

- You will random free week champions you do not own at an even rate (With the exception of that one guy who just seems to be elusive)

- The other champions will be randomed to you in a sometimes accelerated and sometimes decelerated rate to keep them at an average in the middle of the pack.

Images courtesy of leagueoflegends.wikia.com.