### Wondering how the random algorithm takes into account champs owned.

innervation

Senior Member

Quote:
Originally Posted by Zielmann
Just think about it logically. Unless the system simultaneously selects a random champion for each player, and repeats this for every player in the game until 10 unique champions are picked for that game, there has to be some order to the picks.

What is that order? I don't expect we'll ever know. It would make the most sense for it to either alternate teams one at a time, or possibly having it done like draft picks (1-2-2-2-2-1). But even in this situation, I would still expect the top pick when a pair is going to go first, and that their champion assignment would lock the next player out of that champion.

So, yes, you're right. Since there has to be some kind of pick order, the first player to have their champion assigned has a true chance of getting a specific champion as 1 : (player's available champion pool). Then for all the following players, the chances of getting a specific champion in their pool would adjust to take into account any champions already assigned to other players.
I don't think your 'option A' is out of the question. But either way, don't hold your breath about getting a red response OP - the more they reveal about the behind the scenes math, the more likely it is that someone can scam/job the system - and they tend to frown on that.

I would be interested in finding out as well, perhaps I can come up with a scenario that could test this with enough iterations...

Zielmann

Senior Member

Quote:
Originally Posted by innervation
I don't think your 'option A' is out of the question. But either way, don't hold your breath about getting a red response OP - the more they reveal about the behind the scenes math, the more likely it is that someone can scam/job the system - and they tend to frown on that.

I would be interested in finding out as well, perhaps I can come up with a scenario that could test this with enough iterations...
Option A is definitely not entirely out of the question, but I also don't feel like it would be very clean to do it that way, either.

Fomorian27

Senior Member

Quote:
Originally Posted by innervation
I don't think your 'option A' is out of the question. But either way, don't hold your breath about getting a red response OP - the more they reveal about the behind the scenes math, the more likely it is that someone can scam/job the system - and they tend to frown on that.

I would be interested in finding out as well, perhaps I can come up with a scenario that could test this with enough iterations...
9 smurfs with only free week champs unlocked, one person with a non free week champ ulocked, see how often he gets it.

Also, if it is the 1:2:2:2:2:1 pick order, then that means that placement in the custom game lobby matters. That is, you are a higher chance of unlocking a champ that noone on your team owns if you're last pick.

In the previous example of 10 smurfs, if you put the smurf with the unlocked champ as first pick on blue side, he would have an equal chance of having any of the 11 champions he owns. And if he was last pick on purple side, he would have a 50% chance of getting the champion he doesn't share with anyone else, and 50% chance of getting one of the 10 free week champs (whichever champ was not picked for the other nine people). This would prove that the 1:2:2:2:2:1 system is used and calculated instantaneously.

If we can wrangle up enough people we can test this.

Lycanblu

Senior Member

I have no choice but to believe it's random, but owning 64 champs and getting eve 5 times in 10 games is disheartening for a non-dodger like me. Also Tryndamere was tossed in there.

Zielmann

Senior Member

Quote:
Originally Posted by Fomorian27
9 smurfs with only free week champs unlocked, one person with a non free week champ ulocked, see how often he gets it.

Also, if it is the 1:2:2:2:2:1 pick order, then that means that placement in the custom game lobby matters. That is, you are a higher chance of unlocking a champ that noone on your team owns if you're last pick.

In the previous example of 10 smurfs, if you put the smurf with the unlocked champ as first pick on blue side, he would have an equal chance of having any of the 11 champions he owns. And if he was last pick on purple side, he would have a 50% chance of getting the champion he doesn't share with anyone else, and 50% chance of getting one of the 10 free week champs (whichever champ was not picked for the other nine people). This would prove that the 1:2:2:2:2:1 system is used and calculated instantaneously.

If we can wrangle up enough people we can test this.
This would work. But it would be a long and iterative process. You'd need enough data points to show an actual trend. And you'd need to run all those iterations with the player with one extra champion in every slot in the lobby. And I'm talking at least 50, but ideally 100 or more iterations in each location.

JustAPerson

Member

Technically, there has to be an order, because computers actually only read one line of code at a time. So even if the code says to do everyone at once, logically, there's going to be a very short delay in between picks for each person, which results in an order. Sorry, just thinking as a programmer here.
Now, there's no way for us to know what that order is, but there should be one, even if multiple cores are used, since they'll report their results in at different times. The order may even change every time, depending on what order players connect to the select screen in, but we can't know that.
As for that last theory I had, you can see that players connect to the lobby at different times, because there will often be a debate on lanes and one person will claim they didn't see the other person call that lane. I see this frequently enough that I doubt the player in question is lying. You can even see the "player x has connected" messages, in between lines of text sometimes if you connect first.

Fomorian27

Senior Member

Quote:
Originally Posted by Zielmann
This would work. But it would be a long and iterative process. You'd need enough data points to show an actual trend. And you'd need to run all those iterations with the player with one extra champion in every slot in the lobby. And I'm talking at least 50, but ideally 100 or more iterations in each location.
Not necessarily. All you would need to test is whether being in last place has a different chance than being in first place. If the discrepancy is as big as 50% chance vs 1 in 11 chance, that would show up in just a few tests. We're not trying to be super scientific here. We don't need to do hundreds upon hundreds of tests to prove beyond a shadow of a doubt that it's true. Taken to its logical conclusion, we'd need an infinite number of tests to prove beyond a shadow of a doubt that it's true, and nobody's got the time for that.

Senior Member

Quote:
Originally Posted by Fomorian27
and nobody's got the time for that.
Why not Zoid...er, Zilean?

SwaggerJ

Junior Member

If I had to guess, I would think it would be something simple like this:

1. A pick order is determined*
2. Going in order, a champion is randomly chosen for each player based on their available pool of champions. This pool includes all free week champs, but excludes champions who have already been assigned. Once every player has been assigned a champ, the game goes to the pregame lobby.
3. Wait 1-59 seconds for someone to dodge and repeat steps above.
4. After a few minutes, you can finally begin the actual game

*I have no idea how the pick order is chosen. It could be as simple as the order (or reverse order) in which you joined the game, or it could be something extremely complicated. I don't see any reason why they wouldn't use a simpler system (RNG is RNG amirite?), but only a Rioter could tell us for certain.