Aram is not really R

Suryadread·Apr 2, 2017·7 votes·1,569 views

I'm seeing issues in ARAM where randomness is not really true out of 52 champions I own, I landed nearly 3 times in a row on GP for 1/140608 odds of happening or 0.000711197% chance of happening and that's not the first time. So I'm starting to think that random is not a true random and that some mechanics are happening behind that prevent it from being a true random roll on champion selection.

Also I don't know what the rule is about players being matched with others that have way smaller champ pools (say aram only champ pool accounts) but the prevalence of some champ rolls in this mode makes me feel like aram should be rolled on all the champs available put together from all the players in the game for better fairness in order to prevent unbalance through specificly created aram accounts(or match similar account champ pool together for improved fairness in team composition!).

27 Comments

Mecha MaIphiteApr 2, 20179 votes

I can pick any specific combination of three champions and there would be the same odds of that happening out of 52 champs.

True random will always have small segments where it doesn't feel random (getting the same champ, getting champs of the same class, etc), because that's how random works. On the other hand, this wouldn't happen if you had an algorithm to "simulate" randomness to ensure you don't, but that wouldn't be true random.

I don't know what Riot uses but I'd assume they'd have true random. What would they gain out of deliberately making it appear non-random?

SuryadreadApr 2, 20171 votes

Entice players into playing by giving players champs that give better odds of winning in aram. In the long term that's more money for the company as a player that plays longer with the game has a higher chance of spending money as the odds of them not spending money is inferior to 1 and independent rolls on that will eventually lead to 0% of them not spending money ;)

Mecha MaIphiteApr 2, 20172 votes

Except GP is sitting at a 49% in ARAM, which only increases slightly in higher elo. A sample size of 3 games really doesn't prove anything.

SuryadreadApr 4, 20171 votes

well I have about 500 games played so the odds of that happening are still really low. And I've had doubles before on the same account(but that is much less impressive as there's only 83% chance out of 500 games to never have doubles with 52 champs).

DestaiceApr 2, 20175 votes

Random does not mean you cannot get the same thing over and over again.

A bag with four blocks. One green, one red, one blue, one yellow. It is possible to pull the yellow one out 10 times in a row.

It's just very unlikely.

SpoofghoulApr 2, 20175 votes

and then there is aram accounts

Only having a bag with 1 block

And it happens to be yellow

Pulling out the yellow block every time

aram accounts in a nutshell

SuryadreadApr 4, 20171 votes

1+7free/week unless it's included in the weeklies then it's 1+6.

RIP Tahm KenchApr 3, 20171 votes

And then there's ARURF, where you own all 100+ champions, play both weekends 6+ hours a day, and only play as 20% of the cast. I must have rolled Taric, Mordekaiser, Urgot, and Malzahar 6 times each over those two weekends, when that whole time I just wanted to play one game with the new feast stack Chogath. I think Riot definitely has some kind of internal rarity going on for random picks, where champions designated as "too good"/"cancer" have a much lower chance of appearing.

SuryadreadApr 4, 20171 votes

I don't know about that because if we go look on lolking for arams winrate+pickrate teemo is in top5 both times and 2nd place in popularity behind ashe whom i can't even tell the winrate of but is not top 15. But sure as hell does not feel like random even in one's own champ pool.

AMYS GRAVEApr 3, 20172 votes

if you own 52 champions the odds of getting one champion at least three out of four games is actually 52/52 (the first game can be whatever it wants) * 1/52 * 1/52 *52/52 (one of the four games can be any champion)= 0.0369%.

Thats about 1/2704, so with 12 mill people a day, say 1% of those play arams consistently is 120k players daily. If they all own ~52 champions such as yourself then this would happen to ~45 people every single day if the champions are being drawn randomly.

The fact that this happened isn't even statistically unexpected. You're even biasing the results by only showing us the 4 games where it happened. If you pull a streak of 4 randomly from your match history then it looks a lot more random.

SuryadreadApr 4, 20171 votes

I have about 500 games and have had doubles quite more often then you'd expect I feel like; I do not have a complete table of the champ distribution but it certainly does not feel like it's naturally balancing out like it would with true randomness.

A Basic IslandApr 4, 20172 votes

With 52 champs owned you have a little worse than a 2% chance to play the champ you just played, which really isn't that low.

SuryadreadApr 5, 20171 votes

Yeah it's not that low and yet still feels like I've had more.

AMYS GRAVEApr 4, 20171 votes

http://www.wolframalpha.com/widgets/view.jsp?id=25d70df1dbf954506a4f3015a26d03ea

roll a 52 sided die 500 times, you'll probably see some results only 3-4-5 times out of those 500, while some results will come up 15-16-17 times. Its not going to feel like its evening out for another 1000 games or so.

SuryadreadApr 5, 20172 votes

Just did one, thanks for the link. The visual average varies wildly, I guess 500 is not enough for a distribution over 52.

Lil Lewd WitchApr 2, 20171 votes

Well, looking at my own. I had 18 straight ARAM games without repeat champions. Granted, I own all but Kled so...

SuryadreadApr 3, 20171 votes

Weird I get duo rolls (same champ) quite often even with 52 champs that less than 0.03% and I play a lot of arams but not nearly enough to have that happen at the rate it happens given my champ pool imo.

Janna mìdApr 2, 20171 votes

I'm pretty sure u did not calculated well your chances of getting at least 3 times the same champ in 4 games, I don't remember well how to do it but it's more about 1x4/52x52+1/52x52x52 which is a 0.14% chance (actually i think it should be a bit higher since if first game you get A and second game you get B, you can actually get AA or BB then and it is the same result, however I only took in account the AA possibility) so let's say you played about 200 games in aram, then you actually had a 30% chance (almost) which is actually very likely to happen

SuryadreadApr 3, 20171 votes

Pretty sure 3 consecutive independent draws of 1 out of 52 is (1/52)^3 in my case that did not happen it was 1/52 x 1/52 x 51/52 x 1/52 as there's a sona roll which is not a gp roll (http://www.mathgoodies.com/lessons/vol6/independent_events.html)

Janna mìdApr 3, 20171 votes

actually 3 consecutive independant draws of 1 out of 52 is 1/52^2, since it's not about having for example gp 3 times in a row but having a champion 3 times in a row (so 52x1x1/52x52x52 which is equal to 1/52^2 since you have 52 chances out of 52 to play a champion in the first game, it just needs to be the same champion in the 2 next games) And since you need to have 3 times this champ out of 4 games, there are some combinations possible : AAAB (a is whatever champ, b is another one), AABA, ABAA and BAAA (and ABBB that is also a possibility) and each one of those combinations got the same probabilty (which is 1/52^2, so combined it's 5/52^2 (but I did 4/52^2 cuz i ain't so sure about the ABBB thing, if it would be a normal addition or if it won't add the same probability)), and then there is for sure the AAAA combinaison which is 1/52^3 This isn't exact but pretty close

SuryadreadApr 3, 20171 votes

My point was about drawing gp 3 times in a row if we follow your logic we can just take the previous roll and be like look I included your first roll as not champ x and still apply the 1/52^3, even the first draw is 1/52 since you're picking 1/52 champs and it's not you saying I'll take any champ, it's AAA no B or any base52 choice because that would be a 4th roll and we're only doing 3 and considering the chances of rolling a specific champ which is a specific event and not any event in the possible which is the 52.

Janna mìdApr 3, 20171 votes

So if it was evelynn you got 3 times out of those 4 games you wouldn't have made that post ? Also you got ABAA not AAA but uh i kinda failed what i did it's true, the +1/52^3 was actually included by the B. Not something major since it change the result by less than 0.5%

SuryadreadApr 5, 20171 votes

If it was sona I would've considered myself lucky but still been like wtf??!! and made a post most probably less whiny ;)

AWildAPKalistaApr 3, 20171 votes

The most important issue that Rito is blatantly ignoring is ARAM accounts. Their sole purpose is to manipulate the system to create inherently unfair advantages for themselves. This is anti-fun and IT IS CHEATING, but Rito allows it to happen.

Then aside is the issue you've presented: rolling the same champion multiple times in a row or within close games of each other. A few problems with this occur: depending on the order which the game rolls champions, each player that was rolled before you and obtained a champion you own increases your odds of getting any other champion by 1 / (nth - 1). If you have a shallow pool, say 52 champs and you roll last, your odds of getting any one champion is now 1 / (52 - 9), which inherently removes a significant portion of the randomness.

Now, Rito can fix this if they wanted to, but doing so will involve lowering the amount of randomness involved. First option: implement a priority system involving champions in your pool. Each time you roll a champion, the odds of rolling that champion from thenout is greatly reduced, while simultaneously the odds of every other champion is slightly increased. This doesn't eliminate the chance completely, but makes it less likely each time. Second option: every time you roll a champion, the odds of rolling that champion again becomes ZERO until you've rolled everyone.

SuryadreadApr 3, 20171 votes

I want independent rolls idc about removing x champ rolls from pool as you've rolled them before since that would not be true aram but I still think that the odds of the rolls I am getting are not true R and when a roll I get which is 99.963 not going to happen it's going to take a lot of arams for it to happen (and that's for 2 consecutive identical champs) ie: 1500 aram games is still 57% chance of not happening, 500 where I am at is 83%.

AWildAPKalistaApr 3, 20171 votes

Problem is, true randomness can't happen as the game is right now. First, your roll comes from your own champion pool, you can't roll champions you don't own. This blatantly contradicts randomness altogether and ARAM will NEVER be random unless this changes.

Second, players can't roll duplicates; if an ally or enemy rolls a champion, you can't roll that champion. There used to be a glitch where the enemy could roll them, but I digress. Now, although this means real randomness isn't possible, I also don't see it changing either.

SuryadreadApr 4, 20171 votes

Of course I was talking about champs that you own or are available through the weekly free ones but this could be entirely changed with all champs available to all players pools combined so it's a fair roll on both sides as the rolling pool becomes the same.

As this kills aram accounts then the lack of dupes is not as hurtful. I agree I don't see riot changing this but I felt the need to point it out.